article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
there is now overwhelming observational and theoretical evidence that powerful radio galaxies possess highly collimated and relativistic twin jets of matter that emerge from a central active galactic nucleus ( agn ) . in the powerful faranoff - riley type
ii radio galaxies ( fr - ii ; fanaroff & riley 1974 ) , the jets are thought to remain relativistic along most of their length before terminating in a series of shocks resulting from interaction with the surrounding material . a substantial amount of theoretical and numerical ( hydrodynamic ) work suggests that , after passing through the terminal shock , the spent jet material inflates a broad cocoon which encases the jets ( e.g. , see scheuer 1974 ; norman et al .
1982 ; lind et al . 1989 ; begelman & cioffi 1989 ; cioffi & blondin 1992 ; hooda , mangalam & wiita 1994 ) .
in the early life of a powerful source , the cocoon is probably overpressured ( begelman & cioffi 1989 ) with respect to the surrounding ambient material ( either the interstellar medium [ ism ] of the host galaxy , or the intracluster medium [ icm ] of the host cluster ) .
the cocoon then undergoes supersonic pressure - driven expansion into the ambient medium , sweeping the ambient medium into a shocked shell .
a contact discontinuity separates the relativistic cocoon material from the shocked ambient material . from the onset of activity , kelvin - helmholtz ( kh ) instabilities act to shred the contact discontinuity . if the ambient medium has a density profile characterized by @xmath1 where @xmath2 , the contact discontinuity will accelerate and hence will be rayleigh - taylor ( rt ) unstable . however , in the more physical case of @xmath3 the contact discontinuity will possess an initial deceleration , which is sufficient to stabilize the contact discontinuity against rt modes . only at later times , once the cocoon comes into approximate pressure balance with its surroundings , will the deceleration of the contact discontinuity no longer exceed the local gravitational acceleration , thereby rendering it rt unstable .
eventually , agn activity will cease and the jets will turn off . although this stage of a radio galaxy s life has been little studied , it seems likely that the combined action of the kh and rt instabilities will transform the remnant of the cocoon into ` plumes ' that rise in the potential of the galaxy / cluster under the action of buoyancy forces ( gull & northover 1973 ; churazov et al .
2000 ; brggen & kaiser 2000 ) .
observationally , radio galaxies are often seen to be associated with galaxy or cluster cooling flows .
this raises an obvious and interesting set of questions . to what extent
is radio galaxy activity a natural _ response _ to the cooling flow phenomenon ?
do radio galaxies act back on their environment to partially offset the cooling flow , i.e. , are radio galaxies a dramatic manifestation of le chatelier s principle ?
is radio galaxy activity a crucial component for our understanding of galaxy clusters ? since it takes only a small fraction of the cooling flow mass to fuel a powerful agn , it is easy to see how radio galaxy activity can result from a cooling flow .
however , the impact of a radio galaxy on its environment is much less well understood . in this paper
, we use hydrodynamic simulations to study the evolution of a radio galaxy situated at the center of a galaxy / cluster .
in particular , we follow the evolution of the source at times after the jets have terminated .
we examine the interaction of the radio galaxy with the ism / icm and assess the environmental impact of the agn activity .
section 2 presents our set of hydrodynamic simulations and focuses on some numerical issues . in section 3 ,
we briefly discuss the ` active ' phase of the source .
section 4 presents our results for dead sources ( i.e. , the passive phase after the jets have turned off ) , and section 5 discusses some astrophysical implications as well as the limitations of our calculations .
section 6 draws our conclusions together .
we model a source with back - to - back supersonic jets situated at the center of a spherical gaseous atmosphere .
to permit parameter studies of various high resolution simulations , we make the assumption of axisymmetry and neglect magnetic fields .
we discuss these assumptions in section 5 .
the simulations were performed using the zeus-3d code ( stone & norman 1992a,1992b ; clarke , norman & fiedler 1994 ) in its 2d ( axisymmetric ) pure hydrodynamic mode .
zeus-3d is an explicit eulerian finite difference code , formally of second order , which uses an artificial viscosity to reproduce shocks .
major advantages of this code are its flexibility in the choice of the computational grid , and the fact that it has been extensively tested . the simulations were performed on a sun microsystems ultra 60 workstation . unlike many previous simulations , we model both jets and make no assumption regarding reflection symmetry in the plane normal to the jets .
this allows us to study phenomena associated with the collision of the two backflows .
our simulations were performed in spherical polar coordinates @xmath4 , and the computational domain was the region @xmath5 , except for one run ( run 4 below ) which had the domain @xmath6 .
all gas within the calculation was assumed to have an adiabatic index of @xmath7 .
the ambient galaxy / cluster gas was assumed to be initially in an isothermal configuration with a ( adiabatic ) sound speed @xmath8 and a density profile given by a @xmath0-model with @xmath9 : @xmath10^{3/4}},\ ] ] where the core radius was set to @xmath11 .
the gravitational potential , @xmath12 , was set so as to keep this ambient material in hydrostatic equilibrium @xmath13 which gives ( to within an additive constant ) @xmath14 it is assumed that the gravitational potential is dominated by a stationary distribution of background dark matter , i.e. , the self - gravity of the gas is negligible .
this reasonable assumption prevents us from having to solve poisson s equation as part of the hydrodynamic simulations .
we inject back - to - back jets into this background medium via the use of inflow boundary conditions at @xmath15 .
the jets are initially conical with a @xmath16 half opening angle , and in pressure equilibrium with the ambient medium .
they are given an initial density of @xmath17 and mach number ( with respect to the sound speed of the injected jet material ) of @xmath18 .
as discussed in the appendix , these parameters are chosen so that the kelvin - helmzholtz growth rate at the cocoon - icm contact discontinuity in this _ non - relativistic _ simulation is approximately the same as the corresponding growth rate in the real _ relativistic _ situation . at time @xmath19 , the jet activity is stopped and the inner boundary is made reflecting .
the source is then allowed to evolve passively .
we can relate quantities within the code to physical quantities by fixing the parameters of the background medium .
suppose we set @xmath20 , @xmath21 and a central number density of @xmath22 values representative of a rich galaxy cluster .
then , one code unit of time corresponds to 50myr .
the total kinetic luminosity of the source is then @xmath23 .
these parameters are relevant for powerful sources in rich galaxy clusters such as cygnus a ( arnaud et al .
1984 ; reynolds & fabian 1996 ) and 3c 295 .
unless otherwise stated , this paper shall use the above scalings when converting our simulation results into quantities that can be compared directly with real systems .
however , in the present universe at least , many radio galaxies are found in smaller galaxy clusters or poor groups . taking objects like hydra a and virgo a as prototypes , the typical relevant ism / icm parameters are @xmath24 , @xmath25 and @xmath26 ( e.g. , see the detailed analysis of the _ chandra x - ray observatory _
data for hydra a by david et al .
2000 ) . in this case
, one time unit corresponds to 10myr and the total kinetic luminosity of the source is @xmath27 .
of course , it must be appreciated that these @xmath0-models are very crude parameterizations of the real ism / icm density structure . during the active phase ,
the large scale evolution and properties of the cocoon are strongly influenced by much smaller scale shock structures that appear in the jet . as will be discussed in more detail below , a conical
shock appears in the jet which sprays the jet thrust over a large working surface ( which expands almost self - similarly ) .
failure to resolve this shock structure in a simulation leads to the jet thrust being deposited over a very small area which leads to the jet rapidly ` drilling ' a narrow cavity through the icm .
initially , we performed simulations with @xmath28 at three different resolutions : @xmath29 ( run 1 ) , @xmath30 ( run 2 ) , and @xmath31 ( run 3 ) .
the resolution was slightly enhanced at small @xmath32 and along the jet axes .
we find that run 1 does indeed fail to resolve the jet - shock structure .
on the other hand , run 2 and run 3 both well resolve the internal shocks within the jet and produce cocoons that expand in almost a self - similar manner , at least during the early active phase .
the similarity of these two simulations gives us some confidence that we have achieved a degree of numerical convergence .
we note one additional numerical technicality .
our simulations possess a relatively high dynamical range in @xmath32 . however , to maintain adequate resolution on large scales , we choose not to use a logarithmic ( `` scaled '' ) grid for the @xmath32 variable .
instead , we space our grid using a more general geometric progression ( i.e. `` ratioed '' scaling ) .
while allowing us to maintain adequate resolution on the scales that matter to our problem , it has the undesirable effect of forcing highly elongated grid elements to exist in the very central regions of the simulation .
one might then worry about the effects of an effective anisotropic artificial viscosity in this inner region .
however , this is acceptable since we are not interested primarily in the detailed hydrodynamic properties of these inner regions .
one can view the innermost region of our grid as a buffer zone , which connects smoothly to the interesting region of the grid , while protecting it from the hard inner boundary .
experiments with high resolution simulations using a logarithmic @xmath32-grid give us confidence that this is not a problem . for the rest of this paper , we discuss results from run 2 ( i.e. , @xmath33 with a grid size of @xmath34 ) , and one other simulation . this final simulation
( run 4 ) has a larger outer radius ( @xmath35 ) and a larger computational grid ( @xmath36 ) . in the spatial region covered by both
run 2 and run 4 , the resolutions are approximately ( but not exactly ) the same .
there is already a substantial body of numerical work on the hydrodynamics , and magnetohydrodynamics , of active radio galaxy jets .
thus , we shall discuss only briefly our simulation results for the active phase of our source . in a companion paper ( reynolds ,
heinz & begelman 2000 ) we focus on the x - ray appearance of such active sources .
our results during the active phase are very similar to those obtained by lind et al .
( 1989 ) , who performed axisymmetric hydrodynamic simulations of a single jet propagating through a uniform medium .
the initially conical jet is focused by the external pressure and suffers a series of oblique shocks . near the working surface with the background medium , a strong annular
shock deflects most of the flow into a wide fan .
thus , the jet thrust is deposited over a large area , thereby reducing the advance speed of the jet below the value that would be obtained if the jet just ` drilled ' into the background medium .
this is the 2-dimensional equivalent of the so - called `` dentist drill effect '' ( scheuer 1974 ) . after passing through the annular shock ,
the spent jet material enters a mildly supersonic backflow which gradually decelerates through a series of weak shocks .
powerful kelvin - helmholtz ( kh ) instabilities exist along the contact discontinuity between the backflowing jet material and the shocked / compressed ambient material .
these kh instabilities , which become more powerful as the backflow decelerates toward the region where the two backflows collide , act to mix ambient material into the cocoon .
these instabilities can be seen clearly in fig .
[ fig : entropy_map_1 ] , which shows specific entropy maps ( @xmath37 ; note that we neglect logarithms and constants in our definition of the specific entropy ) for run 2 at times @xmath38 , @xmath39 , @xmath40 and @xmath41 . during the active phase ,
the structure expands approximately self - similarly , at least as long as the cocoon is still sufficiently overpressured as to undergo supersonic lateral expansion into the ambient material .
it is worth noting explicitly that the backflow in the cocoon has a speed comparable to the sound speed of the cocoon material . in code units ,
the advance speed of the working surface of the jet is @xmath42 ( i.e. , a mach number of 3 with respect to the sound speed of the undisturbed ambient material ) .
this should be compared with a backflow speed of @xmath43 which exists in a large portion of the cocoon .
translated into terms applicable to real radio galaxies , the backflow speed would be a significant fraction of the speed of light ( since the spent jet material is a relativistic plasma with a sound speed of @xmath44 ) .
as discussed in section 2 , we turn off the jets after @xmath19 and allow the existing structure to evolve .
both run 2 and run 4 display very similar behavior during the passive phase although , of course , run 4 can be followed to much later times due to the larger spatial domain of the calculation ( see fig . 2 ) .
we will describe our findings based on run 4 .
it should be noted that 00 and 00 have also simulated the late evolution of radio galaxies .
their simulations make the assumption that the source evolution before the jets turn off is governed by simple supersonic expansion of the lobes and consequently , their initial conditions for the passive phase are based on relatively simple geometries ; namely , spherical and ellipsoidal bubbles placed in isothermal atmospheres .
however , turbulence and powerful circulatory motion seems to be present at all times inside the active cocoon , which leads to significant excitation of the kelvin - helmholtz instability before the source turns off .
whether , and by how much , the cocoon is overpressured at the moment that the jet activity ceases can be estimated by the simple analytic models of begelman & cioffi ( 1989 ) .
thus , we extend the work of 00 and 00 by simulating the active phase of the source and then shutting the jets off , rather than modelling the late stages of evolution by letting a static bubble evolve under the action of buoyancy . we also simulate a substantially larger portion of the cluster atmosphere while maintaining the required high resolution , thereby allowing us to track the source evolution at later times .
we consider these aspects to be important extensions of these previous works . as would be expected from simple qualitative arguments , turning off the jets results in the almost immediate disappearance of internal shocks in the jet channels , which subsequently collapse .
the very rapid backflow motions in the cocoon slow down accordingly , due to the lack of high - inertia material and the disappearance of the terminal shock , which previously re - directed the flow . while the radial lobe expansion during the active phase was governed mostly by the jet thrust , the absence of ram pressure from the jets in the inactive phase results in a rapid slowing of the cocoon head to subsonic speeds .
if the cocoon is still significantly overpressured when the jet activity ceases , this subsonic stage will be preceded by a sedov expansion phase . during the _ early inactive phase _ of our simulations ,
the source is still slightly overpressured with respect to the ambient medium and we do indeed find a sedov phase , during which the cocoon expands in order to achieve pressure balance .
when pressure balance is reached ( at @xmath45 ) , the driving force on the shocked / compressed icm shell ceases and this shell rapidly decelerates to subsonic speeds .
once the lobes reach rough pressure equilibrium with their surroundings , the dynamics of the lobes closely resemble a rising bubble . in the initial stages of this _ late inactive , or buoyant , stage _ , the denser igm begins to settle back into the center , squirting the light cocoon material along the major axis of the cocoon .
eventually , the icm core will reform and the cocoon will be pinched off into two separate , buoyantly rising plumes .
material that had been entrained previously is now carried along by each plume , partly due to its own inertia and partly due to the significant circulation inside the plume .
this vorticity continues to drag ambient material into the bubble at the plume s back end , where strong eddies produce a jelly - fish - like structure ( fig . [ fig : plumemap ] ) . at the head of the plume , a dense cap of swept up material forms , which is pushed ahead . ]
continuing with the bubble analogy , we can write the buoyant velocity @xmath46 of the plume as @xmath47 where @xmath48 the radius of the bubble , @xmath49 is the local gravitational acceleration , and @xmath50 and @xmath51 are the densities of the ambient and bubble materials respectively .
the top left panel of fig .
[ fig : diagnostics ] shows the volume averaged plume height ( i.e. , its distance from the center ) .
the plumes have approximately constant velocity of @xmath52 .
together with the cocoon size of @xmath53 and the gravitational acceleration at that position , this indicates that @xmath54 is slightly less than unity .
one goal of these simulations is to determine a set of physical parameters that are only weakly dependent on the details of the simulations ( most importantly , they should be weakly dependent on the numerical resolution and the dimensionality ) .
in particular , we are interested in the overall matter distribution , energetics , and thermodynamics of the system . for this discussion ,
it is interesting to separate the system into different structural components , i.e. , the cocoon / plumes , and the ambient medium ( including the shocked / compressed shell that bounds the radio source ) .
defining these regions is not trivial . since zeus-3d is an eulerian code we can not , a priori , separate cocoon plasma from ambient plasma by tracking fluid elements . due to the turbulent cocoon surface ,
a simple geometric identification of the cocoon is also not an option .
however , since the jet material is of very high entropy compared to the background gas ( especially once it has passed through the thermal shock ) , we use the specific entropy @xmath37 as a discriminant . for the simulation parameters we chose ,
the background material has specific entropy ( in code units ) @xmath55^{1/2}.\ ] ] the injected material , on the other hand , has entropy @xmath56 and cocoon material , which has gone through the terminal shock , can have entropy as high as @xmath57 .
we chose to define the cocoon as the volume inside of which the entropy satisfies @xmath58 . at all but
the very earliest times in our simulations , the shock driven into the ambient material is too weak to raise its entropy above this threshold .
conversely , an examination of entropy maps from the simulations shows that , until about @xmath59 , the contact discontinuity between the cocoon and ambient material is sharp and well tracked by the @xmath60 contour .
in fact , until @xmath61 , the location of the contact discontinuity is insensitive to the chosen entropy threshold , provided that threshold is in the range @xmath62 . for times
@xmath63 numerical mixing reduces substantial parts of the cocoon / plumes below the @xmath60 threshold . at these late times , identification of a well - defined contact discontinuity separating the plume from the ambient material becomes problematic .
, the dotted line corresponds to @xmath64 and the dash - dotted line corresponds to @xmath65 .
the second broad peak in the distribution , which corresponds to the cocoon material , gradually moves to lower entropies as it is polluted by low entropy ism / icm gas .
[ fig : s_distribution ] ] to better illustrate the effects of numerical mixing at later times , we calculate the entropy distribution of run 4 .
figure [ fig : s_distribution ] shows this function at different times . due to numerical mixing the second peak at high entropy ( i.e. , the cocoon material ) diffuses to lower entropies .
it is clear that after a certain amount of time an entropy cut is not a reasonable estimator for cocoon material anymore , namely , when numerical diffusion has lowered the entropy inside the cocoon to levels comparable with the ambient entropy .
numerical mixing is an irreversible process and hence creates entropy . on the other hand ,
` mixing ' across the contact discontinuity in real radio galaxies is not necessarily irreversible as can be seen from the following argument .
suppose that the ambient ism / icm material is not mixed on the microscopic level . instead
, assume that intact dense ism / icm filaments and clouds are simply swept up into the light cocoon and kept distinct due to being on different magnetic structures .
the fact that this is a thermodynamically reversible process can be seen by noting that , given a sufficiently long time , buoyancy effects could separate the ism / icm gas from the cocoon plasma . thus , the entropy structure of the simulated cocoon becomes increasingly less physical as time proceeds .
indeed , once numerical mixing has significantly affected a substantial part of the simulated cocoon ( which occurs at @xmath59 ) , the entropy structure of the cocoon may be a poor reflection of reality . to explore the energetics of our simulation in a quantitative manner
, we compute the following quantities : 1 .
total mass of cocoon , @xmath66 2 .
total internal energy of cocoon ( ambient ) material , @xmath67 3 .
total kinetic energy of cocoon ( ambient ) material @xmath68 4 .
total gravitational potential energy of cocoon ( ambient ) material @xmath69 where @xmath70 is the region in which @xmath71 , and @xmath72 is the region in which @xmath73 .
we then reference all energies to their values at the initial time : @xmath74 the bottom panels of fig .
[ fig : diagnostics ] shows how these energies vary with time ( upto @xmath75 ) for run 4 . during the active phase
, the energy injected by the jets is transformed into internal and kinetic energy of both the cocoon and ambient material , as well as the potential energy of the ambient material .
all of these energy forms increase linearly with time as the injected energy is shared roughly according to equipartition .
after the source turns off , the kinetic energy of the cocoon rapidly decreases as the rapid motions associated with the jet cease . during the short sedov phase that follows ( between @xmath19 and @xmath76 ) , the internal energy of the cocoon also undergoes a rapid decrease ( adding mainly to the internal energy of the ambient medium ) , as the overpressured cocoon expands adiabatically to achieve pressure equilibrium with the ambient material . after the sedov phase
, the cocoon source then enters the buoyant phase . at all times during the buoyant phase in our simulations , the rising buoyant plumes entrain and lift material ( mainly in the trailing regions ) .
this contributes to an increase in the gravitational potential energy of both the plume and ambient material .
concurrently , an approximately spherical pulse propagates radially
outwards in the ism / icm atmosphere .
this pulse is the remnant of the strong shock that bounded the supersonic cocoon at early times .
as will be discussed below , this pulse mediates a general expansion ( inflation ) of the ism / icm atmosphere which results due to a heating of the core regions by the jet activity .
the radio galaxy influences the entropy of the ism / icm atmosphere by driving shocks into it . at very long times
after the jet activity has ceased , the icm / ism will be in a relaxed hydrostatic equilibrium in which the entropy is increasing outwards ( or else there will be convective turnover until such a state is achieved )
. a parcel of gas which passed through a very strong radio - galaxy induced shock ( which may have initially been in the centralmost regions of the galaxy / cluster ) will have a high final specific entropy and achieve its new equilibrium state some large distance from the cluster core . if the same parcel of gas suffers a weaker shock , the entropy increase will be less and it will settle at a lower point in the final ism / icm atmosphere . the density , temperature , and entropy plots in fig .
[ fig : profiles ] show that , after the episode of radio galaxy activity , the ism / icm core is left on a higher adiabat than it was prior to the activity .
this is easy to understand .
since the initial radio galaxy expansion was supersonic , the material which initially had the lowest specific entropy ( i.e. at the center of the initial ism / icm distribution ) has all been shock heated .
the final ism / icm core contains a mixture of ( mildly ) shocked gas , and unshocked ism / icm which has adiabatically flowed inwards from more distant regions ( and thus possesses a higher specific entropy ) to displace highly shocked material .
the highly shocked material is expelled to the outer regions of the galaxy / cluster . for the specific case represented by our simulation , the excess entropy over the undisturbed cluster gas peaks at 20% .
this translates into an excess entropy of about 14@xmath77 .
although this is a small change in the entropy of the cluster core , one might think that repeated radio galaxy activity would lead to a significant accumulation of excess entropy in the cluster core .
this depends upon the assumed entropy profile for the ism / icm . within the @xmath0-model parameterization
used here , the entropy profile of the ism / icm is rather shallow .
thus , even if one could imagine completely shocking a large part of the central ism / icm , the entropy increase of the core seen in the final equilibrium would still be rather modest since material just outside of the strongly shocked region will simply flow inwards and take its place . if the initial entropy profile is much steeper ( as it is in some strong cooling flow models ) then the same radio galaxy could impact the entropy of the core rather more significantly .
clearly , further work is needed on the contribution of radio galaxies to the ism / icm entropy in which the entropy injection is considered in relation to particular detailed cooling flow models . using _ rosat _ data , ponman , cannon & navarro ( 1999 ) and lloyd - davies , ponman & cannon ( 2000 )
have found evidence for an `` entropy floor '' in groups and clusters .
this suggests from an observational point of view that these systems possess excess entropy ( as compared to the entropy profile expected from simple scaling relations and detailed simulations ; however , see alternative viewpoint of bryan 2000 ) .
lloyd - davies , ponman & cannon ( 2000 ) claim an excess entropy of 70140@xmath78 in a sample of 20 clusters and groups .
interestingly , these authors also conclude that this entropy injection occurred at low densities ( @xmath79 ) and hence they suggest that it occurred prior to the collapse / formation of the cluster . in principle , microscopic mixing of the icm and radio plasma _ inside _ of the cocoon / plumes might also lead to icm entropy injection at low density . however , our simulations suggest that buoyancy effects tend to transport any such mixed material completely out of the cluster center . thus ,
if the determination of the density at which the entropy injection occurred is robust , it does indeed seem likely that the bulk of the entropy injection occurred prior to cluster formation and is not due to more recent radio galaxy activity .
the energetics of the very late stage of run 4 are reported in fig .
[ fig : profiles ] .
the large scale pulse mentioned in 4.2.3 manifests itself in each of the quantities plotted in fig . [ fig : profiles ] at @xmath80 .
the top - left panel shows the difference in enclosed energies as a function of radius between late times and the initial state @xmath81 .
comparing the values of this function on each side of the pulse may suggest to the casual reader that the wave contains most of the injected energy .
this is a curious result since we have already noted appreciable entropy and internal energy increases associated with the irreversible heating of the ism / icm atmosphere .
the bottom panels of fig . [ fig : profiles ] elucidate the true nature of this pulse . in the bottom - left panel ,
we note that the major contributor to the increase of enclosed energy comes from gravitational potential energy ( rather than kinetic and internal energy as would be expected if this were a sound wave carrying away the injected energy ) . secondly , displaying the enclosed energy difference in _ mass coordinates _ ( fig .
[ fig : profiles ] bottom - right panel ) shows that the mass contained within the inner edge of the wave - packet has acquired an appreciable excess of energy during the evolution corresponding to at least half of the injected energy .
thus , it can be seen that the ism / icm atmosphere has been inflated ( or puffed up ) due to the heat deposited in the core of the atmosphere by the jet activity .
the spherical pulse ( travelling at the sound speed ) drives mass to a higher location in the potential thereby mediating this inflation . that this inflation has occurred is obvious from the reduced core density in the final state ( fig .
[ fig : profiles ] top right ) . examining the enclosed energy in mass
coordinates lets one see that this inflation actually accounts for a substantial fraction of the injected energy , implying that the injected energy has been heated the ism / icm with a high efficiency ( @xmath82 ) .
as discussed above , we find that a large fraction ( @xmath82 ) of the energy injected by the jet activity is thermalized in the core regions of the ism / icm .
as mentioned in the introduction , this raises the possibility that radio galaxy activity can have significant effects on the energetics of the ism / icm and , therefore , act as negative feedback for cooling flow activity . in particular ,
let us address the case of cluster cooling flows .
peres et al .
( 1998 ) compared the radio and x - ray luminosity of 58 clusters in a flux limited _ rosat _ sample .
they found strong radio sources in about @xmath83 of the clusters within the sample .
however , using the results of peres et al . ( 1998 ) and summing over all members of the sample ( not just those with radio sources ) , we find that the radio luminosity of the sample ( defined as the sum over all sources of @xmath84 measured at 1.4ghz ) is about 1% of the sample bolometric x - ray luminosity from within the `` cooling radius '' . to discuss the relevance of radio galaxy heating on the cluster , we must relate the radio luminosity to the kinetic luminosity of the sources .
simple dynamical models of expanding radio lobes suggest that the maximum synchrotron luminosity of homogeneous , space - filling , lobes is only a few percent of the kinetic luminosity ( e.g. , see bicknell , dopita & odea 1997 ) .
departures from equipartition always act to decrease the synchrotron efficiency , and the efficiency further decreases as the source expands ( begelman 1996 , 1999 ) .
the only way to increase this efficiency is to concentrate the radio emission into a small volume of the radio lobe ( e.g. , strong shocks within the lobe ) . in the absence of such inhomogenieties , it seems likely that the radio luminosity is at most a few percent of the kinetic luminosity .
therefore , using the peres et al .
( 1998 ) results quoted above , the sample - averaged kinetic luminosity is comparable to the x - ray luminosity from within the cooling radius .
the results of our work suggest that a large fraction of this jet power can be thermalized in the icm .
thus , from a purely energetic point of view , radio galaxies can balance the radiative cooling associated with cooling flows .
however , it is clear that the icm of real galaxy clusters does not experience a steady - state situation in which jet - induced heating balances radiative cooling . as has already been noted , most of the integrated radio luminosity in the peres et al .
( 1998 ) sample is contributed by the powerful radio galaxies that reside in only 10% of the clusters .
the arguments of the previous paragraph would suggest that , in these clusters , the radio galaxy heating would vastly exceed the radiative cooling .
conversely , radiative cooling would certainly seem to dominate in the remaining 90% of the systems ( unless they possess powerful , but invisible , nuclear outflows ; binney 1999 ) .
thus , it is clear that even if radio galaxy heating is relevant , the heating / cooling balance can be vastly different from cluster to cluster , or at different points of a given cluster s lifetime . it is beyond the scope of this paper to confront , in any direct manner , the plethora of new data that the _ chandra x - ray observatory _ and _ xmm - newton _ are providing on the properties of the icm and cooling flows . here
, we simply note some of the challenges that future models of cooling - flows and embedded radio - galaxies must meet .
_ chandra _ finds spectroscopic evidence for cooling flows ( albeit with a mass deposition rate smaller than previously estimated with _ rosat _ ; peres et al .
1998 ) even in clusters that contain energetic radio galaxies .
for example , fabian et al . ( 2000 ) find cool gas in the perseus cluster , even though the central radio galaxy is clearly interacting with the icm is a rather vigorous manner .
curiously , the cool gas seems to form a shell around the radio lobes and occupies precisely the location where we would expect shock / compressionally heated gas to reside . furthermore , _
chandra _ has observed sharp temperature discontinuities in clusters that do not appear to host powerful radio galaxies .
firstly , a few clusters ( e.g. , abell 2142 , markevich et al . 2000 ) have large scale temperature discontinuities which have become known as `` cooling fronts '' .
these may be discontinuities in the cooling flow associated with a group / cluster merger event .
secondly , mazzotta et al .
( 2001 ) has found a bubble of hot gas in the cluster mkw3s .
they associated this with some previous outburst of nuclear activity from the central galaxy .
however , if such a bubble had buoyantly risen from the centre of the cluster , we would expect kh instabilities to disrupt it to a larger extent than observed .
one would need to invoke some mechanism , such as large scale magnetic fields , to stabilize the large scale kh instabilities and prevent disruption of the bubble .
_ xmm - newton _ has also produced a further mystery . by performing detailed emission line spectroscopy of several cooling flow clusters ( e.g. , abell 1835 , peterson et al .
2001 ; abell 1795 , tamura et al . 2001 ) with the reflection grating spectrometer ( rgs ) on _ xmm - newton _ , one can probe the detailed temperature distribution of the icm .
multiple temperatures were observed in concordance with cooling flow models , with components ranging from the virial temperature down to @xmath85kev .
however , no emission lines were observed from gas below @xmath85kev whereas cooling models would predict large amounts of emission from such gas .
fabian et al . ( 2001 ) discusses various solutions for this puzzle and favours a model in which there are large abundance inhomogenieties affecting the predicting cooling rates .
another solution is to postulate a heating mechanism which targets cool ( @xmath85kev ) gas .
examples of such a mechanism are the `` percolation '' model of begelman ( 2001 ) , or the `` reconnection '' model of norman & meiksin ( 1996 ) .
heating models based upon simple hydrodynamic shocks ( such as presented in this work or that of david et al .
2001 ) do not possess such temperature dependent targeting . in order to match the observations ,
the putative heating mechanism must meet the further challenge of heating the cool gas rapidly up to the virial temperature .
finally , we note that neither _ chandra _ nor _ xmm - newton _ have yet found clear evidence for strong shocks associated with radio galaxies . using the characterization of paper i
, all cluster radio sources observed so far appear to be in the weak - shock or the no - shock regime .
although more work is required before making a rigorous statement , there appears to be a disagreement between the observed frequency of strong shocks and current hydrodynamic models of radio galaxy evolution .
this might suggest that radio jets are less powerful than we believe ( i.e. the synchrotron efficiency is higher than a few percent , implying that the radio emitting regions fill a small portion of the radio lobes ) .
this problem was first pointed out by david et al .
( 2001 ) .
the simulations that we present in this paper are just the first step in addressing the long - term evolution of radio galaxies . in order to make the problem tractable ,
we have made several simplifying assumptions . here , we briefly discuss how these assumptions might affect our results .
our biggest simplification is that we have reduced the problem to two spatial dimensions by assuming strict axisymmetry about the jet axis . given the inherent axisymmetry of the system , and the fact that observed radio cocoons are approximately axisymmetric , this is not an outrageous assumption .
however , there are several undesirable consequences of assuming axisymmetry .
firstly , we prevent the formation of non - axisymmetric kh and rt modes .
all of the available free energy is then channeled into the axisymmetric and rt modes .
the axisymmetric kh modes are most efficient at mixing ( since they present the most effective barrier to the flow ) , and hence the axisymmetry assumption maximizes the amount of mixing between the ambient and cocoon material .
secondly , the assumed axisymmetry prevents us from including the effects of any realistic random ism / icm motions ( i.e. , `` weather '' ) .
these icm motions will buffet the buoyantly rising plumes and will become crucial in determining the morphology of very old sources . we have also chosen to ignore the presence of magnetic fields ( i.e. , we perform a hydrodynamic , rather than a magnetohydrodynamic , calculation ) .
magnetic fields in the radio cocoon might well be in equipartition with the particle energies and hence could be dynamically significant .
dynamically important fields that are tangled in an isotropic manner on small scales behave like a relativistic gas and so can be handled within the context of a pure hydrodynamic simulation .
however , anisotropic tangled fields , or large scale magnetic fields , will introduce new features to the dynamics which require a full magnetohydrodynamic treatment .
similarly , the importance of neglecting the cocoon s relativistic equation of state can only be assessed once calculations are performed using relativistic hydrodynamic codes .
radio galaxies are known to be dynamical systems that evolve on timescales of @xmath86 . hence
, we expect many galaxies and clusters of galaxies to be host to dead , or relic , radio sources .
it is possible , and maybe even probable , that this past radio activity has influenced the energetics and thermodynamics of the hot , space - filling medium in these hosts .
thus , a study of the environmental impact of dead radio galaxies is important and , given the high quality x - ray data now coming from the _ chandra x - ray observatory _ and _ xmm newton _ , very timely .
we have used high resolution hydrodynamic simulations to investigate one particular scenario relevant to dying radio galaxies i.e. , a rather powerful radio - loud agn situated at the center of a galaxy or cluster of galaxies whose activity ceases abruptly .
in particular , we have simulated back - to - back jets propagating in a @xmath0-model galaxy / cluster atmosphere .
we then shut down the jet activity and let the resulting structure evolve . to make the problem more tractable , we assume axisymmetry about the jet axis ( thereby reducing the problem to two spatial dimensions ) and neglect the action of magnetic fields .
this is a significant extension of the work of 00 and 00 since we simulate the active phase of the source and then shut off the jets , rather than modelling the late stages of evolution by simply letting a static bubble evolve under the action of buoyancy . in the early lifetime of the simulated source
( while the radio jets are still active ) , the shocked jet material forms a cocoon which is bounded by a shell of shocked icm . even during the active phase , hydrodynamic instabilities start to shred this cocoon .
however , only after the jet activity ceases do kh and rt instabilities destroy the integrity of the cocoon . thereafter , the old cocoon material forms two plumes which rise in the cluster potential through the action of buoyancy forces .
these plumes entrain a significant amount of cooler material from the icm core and lift this material high up into the cluster atmosphere . at very late times , the galaxy / cluster atmosphere settles back into hydrostatic equilibrium but with a core specific entropy that has been enhanced by @xmath87 over its initial value .
this entropy enhancement is due to the shocking of the lowest entropy material by the strong shock which bounds the radio cocoon during the early active phase .
we find that a large fraction ( @xmath88 ) of the injected energy is thermalized in the ism / icm .
comparing late times to the initial state , we find that the ism / icm atmosphere has become inflated in order to maintain hydrostatic equilibrium after the thermalization of the jets energy .
such a large thermalization efficiency raises the possibility that radio galaxies are important in the overall energy budget of the ism / icm .
csr appreciates support from hubble fellowship grant hf-01113.01 - 98a .
this grant was awarded by the space telescope institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa under contract nas 5 - 26555 .
this work has also benefited from support by nasa grant nag5 - 7329 . finally , we appreciate support from nasa under ltsa grant nag5 - 6337 , and the national science foundation under grants ast-9529170 and ast-9876887 .
arnaud k. a. , fabian a. c. , eales s. a. , jones c. , forman w. , 1984 , mnras , 211 , 981 begelman m. c. , 2001 , to appear in particles and fields in radio galaxies , ed .
r. a. laing & k. m. blundell .
begelman m. c. , 1999 , in the most distant radio galaxies , ed .
h. j. a. rttering , p. n. best & m. d. lehnert ( amssterdam : royal netherlands academy of arts and sciences ) , 173 begelman m. c. , 1996 , in cygnus a study of a radio galaxy , ed . c. carilli & d. harris ( cambridge : cambridge university press ) , 209 begelman m. c. , cioffi d. f. , 1989 , apj , 345 , l21 bicknell g. v. , dopita m. a. , odea c. p. , 1997 ,
apj , 485 , 112 binney j. , 1999 , in the radio galaxy m87 , ed .
j.rser & k. meisenheimer , springer - verlag , 116 birkinshaw m. , 1984 , mnras , 208 , 887 blandford r. d. , pringle j. e. , 1976 , mnras , 176 , 443 brggen m. , kaiser c. r. , 2000 , mnras , submitted bryan g. , 2000 , apjl , in press cesarsky c. j. , kulsrud r. m. , 1981 , in origin of cosmic rays , iau symp .
g. setti , g.spada , a. w. , wolfendale , ( dordrecht : reidel ) , 251 churazov e. , brggen m. , kaiser c. r. , bhringer h. , forman w. , 2000 , submitted to apj ( astro - ph/0008215 ) .
cioffi d. f. , blondin j. m. , 1992 , apj , 392 , 458 clarke d. a. , norman m. l. , fiedler r. a. , 1994 , national center for supercomputing applications technical report 15 .
david l. r. et al . , 2001 , apj , 557 , 546 etorri s. , fabian a. c. , 2000 , mnras , in press fabian a. c. , mushotzky r. f. , nulsen p. e. j. , peterson j. r. , 2001 , mnras , 321 , l20 fabian a. c. et al .
, 2000 , mnras , 318 , l65 faranoff b. l. , riley j. m. , 1974 , mnras , 167 , l31 ferrari a. , trussoni e. , zaninetti l. , 1978 , a&a , 64 , 43 gull s. f. , northover k. j. e. , 1973 , nature , 224 , 80 hooda j. s. , mangalam a. v. , wiita p. j. , 1994 , apj , 423 , 116 lind k. r. , payne d. g. , meier d. l. , blandford r. d. , 1989 , apj , 344 , 89 lloyd - davis e. j. , ponman t. j. , cannon d. b. , 2000 , mnras , in press markevich m. et al . , 2000 , apj , 541 , 542 mazzotta p. , kaastra j. s. , paerels f. b. , ferrigno c. , colafrancesco s. , mewe r. , forman w. r. , 2001 , apjl , in press norman c. , meiksin a. , 1996 , apj , 468 , 97 norman m. l. , winkler k. -h .
a. , smarr l. , smith m. d. , 1982 , a&a , 113 , 285 peres c. b. , fabian a. c. , edge a. c. , allen s. w. , johnstone r. m. , white d. a. , 1998 , mnras , 298 , 416 peterson j. a. et al . , 2001 , a&a , 365 , 104 ponman t. j. , cannon d. b. , navarro j. f. , 1999 , nature , 397 , 135 ray t. p. , 1981 ,
mnras , 196 , 195 reynolds c. s. , fabian a. c. , 1996 , mnras , 278 , 479 reynolds c. s. , heinz s. , begelman m. c. , 2001 , apjl , in press scheuer p. a. g. , 1974 , mnras , 166 , 513 stone j. m. , norman m. l. , 1992a , apjs , 80 , 791 stone j. m. , norman m. l. , 1992b , apjs , 80 , 81 tamura t. et al .
, 2001 , a&a , 365 , l87 zaninetti l. , 1985 , a&a , 1985 , 79 , 82
in this appendix , we present a brief derivation of the relativistic dispersion relation for the kelvin - helmholtz instability in cylinderical geometry .
we then use this dispersion relation to connect our simulations to the behavior of real sources .
similar relativistic kh analyses have been performed by many authors ( blandford & pringle 1976 ; ferrari , trussoni & zaninetti 1978 ; ray 1981 ; birkinshaw 1984 ; zaninetti 1985 ) .
we repeat the derivation here for completeness .
the continuity and momentum equations of relativistic fluid dynamics for a @xmath89-law gas are @xmath90 where @xmath91 and @xmath92 are the comoving density and pressure , respectively , @xmath93 are the components of the four velocity , and @xmath89 is the polytropic index . here
, @xmath94 is the relativistic enthalpy .
we wish to use these equations to study the relativistic kelvin - helmholtz instability .
consider first - order perturbations of these equations : @xmath95 for simplicity , we shall drop the subscript ` 0 ' in the subsequent equations .
assume that the unperturbed quantities are time - independent and spatially uniform .
the linearized fluid equations can then be cast in the form @xmath96 for definiteness , we choose cylinderical polar coordinates and say that the unperturbed velocity field is along the @xmath97-direction with velocity @xmath98 .
this last equation can then be written as @xmath99 now consider the following geometry which is appropriate for the analysis of the kelvin - helmholtz instability in radio galaxy cocoons .
consider a contact discontinuity situated at the surface @xmath100 .
for @xmath101 , we have a fluid ( denoted by subscript 1 ) which has an unperturbed velocity @xmath102 in the @xmath97-direction . for @xmath103
, we have a fluid ( denoted by subscript 2 ) which has an unperturbed velocity of @xmath104 .
we search for perturbed modes of the form @xmath105 $ ] , imposing a regularity condition at @xmath106 and only permitting outgoing waves in the region @xmath103 . matching the pressure perturbations at @xmath100 gives the following dispersion relation : @xmath107 where @xmath108 with a similar expression for @xmath109 . without loss of generality , we can suppose that the fluid in the region @xmath103 is at rest ( @xmath110 , @xmath111 ) .
it can be seen from this last expression with a little algebra that , in limit of small wavelengths and only mildly relativistic wave speeds , the general solution has the form @xmath112 where @xmath113 , where @xmath114 is the sound speed in the region @xmath101 .
we can now use this general solution to comment on the connection between our simulations and the hydrodynamics of real sources .
since most of the interesting hydrodynamics is associated with the development of the kh instability , we should endeavour to approximate the expected real - life kh growth rates within our simulations .
the cocoons in real radio galaxies will , at least initially , be dominated by plasma with a relativistic equation of state ( i.e. @xmath115 ) . on the other hand ,
our simulations are inherently non - relativistic ( i.e. , @xmath116 ) .
while fully relativistic simulations are needed to address , in full , the effect of a relativistic equation of state , we can use the above dispersion relation to judge the effect on the kh growth rates of neglecting the relativistic nature of the cocoon plasma .
numerical solutions of eqn ( 19 ) show that the kh growth rates and wavelengths are little affected by changing from @xmath115 to @xmath116 .
hence , neglecting the relativistic equation of state is not a serious flaw of our simulations . noting that the mach number of the cocoon backflow is determined by the hydrodynamics of the system ( and is always approximately unity ) , the remaining parameter relevant to the kh growth rate is the ratio @xmath117 . in real systems
, the ambient thermal material will be non - relativistic ( with sound speed @xmath118 ) while the cocoon ( especially near the jet head ) will be dominated by relativistic plasma ( @xmath119 ) .
since @xmath120 , the ratio of relativistic enthalpies will be @xmath121 plausible parameters give @xmath122 , giving @xmath123 .
any pollution of the relativistic plasma in the cocoon ( due to mixing ) will reduce this ratio appreciably . for comparison , in our simulations we have @xmath124 and by measuring the sound speeds ,
we determine that we achieve @xmath125 .
we conclude that while it would be desirable to push this ratio lower by a factor of a few , our simulations are exploring the appropriate parts of parameter space . | we present a numerical investigation of dead , or relic , radio galaxies and the environmental impact that radio galaxy activity has on the host galaxy or galaxy cluster .
we perform axisymmetric hydrodynamical calculations of light , supersonic , back - to - back jets propagating in a @xmath0-model galaxy / cluster atmosphere .
we then shut down the jet activity and let the resulting structure evolve passively .
the dead source undergoes an initial phase of pressure driven expansion until it achieves pressure equilibrium with its surroundings .
thereafter , buoyancy forces drive the evolution and lead to the formation of two oppositely directed plumes that float high into the galaxy / cluster atmosphere . these plumes
entrain a significant amount of low entropy material from the galaxy / cluster core and lift it high into the atmosphere .
an important result is that a large fraction ( at least half ) of the energy injected by the jet activity is thermalized in the ism / icm core .
the whole ism / icm atmosphere inflates in order to regain hydrostatic equilibrium .
this inflation is mediated by an approximately spherical disturbance which propagates into the atmosphere at the sound speed .
the fact that such a large fraction of the injected energy is thermalized suggests that radio galaxies may have an important role in the overall energy budget of rich ism / icm atmospheres .
in particular , they may act as a strong and highly time - dependent source of negative feedback for galaxy / cluster cooling flows . #
1to 0pt#1 = = # 1 # 1 to size # 1 to size # 1 = cmr10 scaled2 = cmbx10 scaled2 = cmti10 scaled2 ps .
00churazov et al .
( 2000 ) 00brggen & kaiser ( 2000 ) = -1.5 cm cooling flows galaxies : jets hydrodynamics radio galaxies x - rays : galaxies : clusters |
an element of @xmath10 is called a _ binary string _ ( henceforth just called a _ string _ ) of length @xmath1 , with the usual concatenation notation .
for example , @xmath11 is the string of length @xmath1 consisting of @xmath12 0 bits followed by a single 1 bit .
we will denote by @xmath13 the @xmath14 unit string in @xmath10 .
let @xmath15 be a fixed integer .
the _ @xmath1-cube _
@xmath2 is the graph whose vertices are the binary strings of length @xmath1 , with an edge connecting vertices @xmath16 and @xmath17 if the underlying strings differ in exactly one position . given a graph @xmath18 , the set of vertices of @xmath18 is denoted by @xmath19 .
we use @xmath20 to denote the length of the shortest path connecting @xmath21 and @xmath22 in @xmath18 .
lastly , we will write @xmath23 to signify that the graphs @xmath18 and @xmath24 are isomorphic . for a given string @xmath3 and integer @xmath1 , the
_ generalized fibonacci cube _
@xmath0 is the subgraph of @xmath2 induced by the set of all strings of length @xmath1 that do not contain @xmath3 as a consecutive substring .
indeed , this generalizes the notion of the @xmath1-dimensional _ fibonacci cube _
@xmath25 , which is the graph obtained from the @xmath1-cube @xmath2 by removing all vertices that contain the substring @xmath26 .
fibonacci cubes were introduced by hsu @xcite as a model for interconnection networks . like the hypercube graphs ,
fibonacci cubes have several properties which make them ideal as a network topology , yet their size grows significantly slower than that of the hypercubes .
fibonacci cubes have been extensively investigated ; see , for example , the recent survey by klavar @xcite and even more recent papers of klavar and mollard @xcite and vesel @xcite . in the first of these papers , different asymptotic properties of fibonacci cubes are established , while in the latter a linear recognition algorithm is designed for recognizing fibonacci cubes , improving the previous best recognition algorithm of taranenko and vesel @xcite .
later , ili , klavar , and rho @xcite introduced the idea of generalized fibonacci cubes ( as defined above ) . under the same name ,
the graphs @xmath27 were studied by liu , hsu , and chung @xcite and zagaglia salvi @xcite .
the analysis of the properties of generalized fibonacci cubes led to the study of several problems related to the combinatorics of words .
to study their isometric embeddability into hypercubes , good and bad words were introduced by klavar and shpectorov @xcite , where it was proved that about eight percent of all words are good .
isometric embeddability and hamiltonicity of generalized fibonacci cubes motivated the ideas of the index and parity of a binary word , as defined by ili , klavar , and rho @xcite . in this paper
we consider the following fundamental question about the generalized fibonacci cubes : for which binary strings @xmath3 and @xmath8 and positive integers @xmath1 are the generalized fibonacci cubes @xmath0 and @xmath28 isomorphic ?
it is easy to see that if @xmath8 is the binary complement of @xmath3 , or if @xmath8 is the reverse of @xmath3 ( the reverse of @xmath29 is @xmath30 ) , then @xmath4 for any dimension @xmath1 .
hence we say that a pair of binary strings @xmath6 is _ trivial _ if @xmath8 can be obtained from @xmath3 by binary complementation , reversal , or composition of these mappings .
we are therefore only interested in the behavior of the other pairs , which we call the _ non - trivial pairs_. we proceed as follows . in the next section
, we prove that if @xmath4 , then @xmath5 . we also prove that there exist non - trivial pairs of strings @xmath6 such that @xmath4 , where @xmath7 . in the last section , we prove that if @xmath31 , then @xmath4 if and only if @xmath3 and @xmath8 have the same block structure .
several conjectures are posed along the way .
in this section we first prove that if @xmath4 , then @xmath5 .
then we pose the question whether there is some relation between @xmath32 ( @xmath33 ) and @xmath1 provided that @xmath4 . to this end
we prove that there exist non - trivial pairs @xmath6 such that @xmath4 and @xmath7 .
we also conjecture that for any non - trivial pair @xmath34 such that @xmath4 , we must have @xmath7 . the _ autocorrelation polynomial _ @xmath35 associated to a binary string @xmath36 is defined as @xmath37 where @xmath38 if the length @xmath39 suffix of @xmath3 is equal to the length @xmath39 prefix of @xmath3 , i.e. , if @xmath40 , and @xmath41 otherwise
; see flajolet and sedgewick @xcite .
note that the autocorrelation polynomial of @xmath42 is of degree at most @xmath43 and has degree @xmath43 if and only if the last bit of @xmath3 is equal to the first bit of @xmath3 .
observe also that @xmath44 we note in passing that more generally , @xmath45 if and only if @xmath46 or @xmath47 and that @xmath48 if and only if every non - trivial suffix of @xmath3 is different from the prefix of @xmath3 of the same length . such words were named _ prime _ in @xcite .
the following theorem establishes that only strings which have the same length can generate isomorphic generalized fibonacci cubes .
[ thm : equal - length ] if @xmath49 and @xmath4 , then @xmath5 .
set @xmath50 and assume that @xmath51 . we will show that @xmath52 , from which the theorem follows . the key idea of the proof is to apply a result of guibas and odlyzko @xcite stating that if @xmath53 and @xmath54 are binary strings such that @xmath55 , then @xmath56
. moreover , since the values @xmath57 only depends on @xmath58 , cf .
* proposition 1.4 ) , we also infer that if @xmath59 and @xmath60 , then @xmath56 .
it follows that if @xmath3 is a binary string of length @xmath61 , then @xmath62 in addition , since @xmath63 is a strict substring of @xmath64 we infer that @xmath65 combining and we conclude that @xmath66 where the last inequality is due to the assumption that @xmath67 .
hence , non - trivial pairs @xmath6 such that @xmath4 are of the same length .
we next ask what is the relation of this length with the dimension of the corresponding generalized fibonacci cubes .
the following result could be the extremal case .
[ thm:3k-1 ] if @xmath68 , then @xmath69 for any @xmath70 .
let @xmath68 be a fixed integer and set @xmath71 , @xmath72 , @xmath73 , and @xmath74 .
let @xmath75 and @xmath76 .
for any @xmath77 let @xmath78 then , by definition , @xmath79 let @xmath80 be an arbitrary vertex whose underlying string is in @xmath81 .
it then follows that @xmath82 , and so @xmath83 holds for all @xmath84 . since @xmath85 , we have @xmath86 . with a parallel argument
we infer that also @xmath87 .
this implies that @xmath88 .
consider now the mapping @xmath89 defined by @xmath90 in particular , @xmath91 fixes the first @xmath92 and the last @xmath43 coordinates .
since transposition of coordinates , complementation of a coordinate , and any composition of such mappings are all automorphisms of a hypercube , @xmath91 is an automorphism of @xmath93 .
consider now @xmath18 and @xmath94 as subgraphs of @xmath93 and the restriction @xmath95 of @xmath91 to @xmath18 .
since @xmath88 , it remains to prove that @xmath96 .
suppose on the contrary that for some @xmath97 we have @xmath98
. then @xmath99 , where @xmath100 for some @xmath77 , and @xmath101 .
it is straightforward to see that @xmath102 . since this is not a vertex of @xmath19 we have a contradiction .
we have thus proved the result for @xmath103 .
note that all of the above arguments also work for @xmath104 and when @xmath105 , the assertion is trivial .
motivated by the last theorem we pose : [ conj : dim - minus-1 ] if @xmath3 and @xmath8 are binary strings such that @xmath106 , then @xmath107 .
[ conj:2/3 ] let @xmath6 be a non - trivial pair such that @xmath4
. then @xmath7 .
we have verified these conjectures for small strings using the sage package @xcite . more precisely , we tested both conjectures for all @xmath108 and all non - trivial pairs @xmath6 .
see appendix [ appendix ] for the corresponding procedures .
in this section we characterize the binary strings @xmath109 of length @xmath12 for which @xmath110 .
it turns out that they are precisely the strings with the same block structure .
let @xmath111 denote one less than the number of blocks of @xmath112 .
for example @xmath113 .
when a bit is different from the previous bit we call its index an _ index of bit change _ and denote it by @xmath114 . therefore @xmath115 for @xmath116 .
[ thm : d-1 ] let @xmath117 and let @xmath3 , @xmath8 be binary strings of length @xmath12
. then @xmath118 if and only if @xmath110 .
. then @xmath120 .
similarly set @xmath121 .
let @xmath122 be the indices of bit change of @xmath3 and let @xmath123 be the indices of bit change of @xmath8 . since @xmath124 for @xmath125 and @xmath126 for @xmath125 , the strings @xmath127 and @xmath128 are different precisely for @xmath129 , @xmath130 .
it follows that @xmath131 and by a parallel argument @xmath132 .
assume first that @xmath118 .
we may without loss of generality assume that @xmath133 .
then @xmath134 . let @xmath135 be a permutation of @xmath136 such that @xmath137 , @xmath138 , @xmath139 .
set @xmath140 , where @xmath141 { \overline{x_{\phi^{-1}(\tau ) } } } ; & \text{otherwise.}\end{array}\right.\ ] ] as transposition of coordinates , complementation of a coordinate , and compositions of such mappings are automorphisms of a hypercube , @xmath142 is an automorphism of @xmath2 .
also @xmath142 sends the vertices of @xmath143 to @xmath144 .
thus @xmath142 is an isomorphism from @xmath0 to @xmath28 . to prove the converse assume that @xmath110 .
we may without loss of generality assume that @xmath145
. assume @xmath146 .
then @xmath147 and hence @xmath148 , while @xmath149 if @xmath150 .
therefore @xmath151 .
assume @xmath152 . then note that the vertices from @xmath143 induce a path on four vertices and hence @xmath153 .
if @xmath154 , then the vertices from @xmath144 induce two disjoint copies of @xmath155 and hence @xmath156 .
we conclude that @xmath157 . for the rest of the proof we can thus assume that @xmath158 .
the subgraph of @xmath2 induced on @xmath143 consists of two edges @xmath159 and @xmath160 where @xmath161 and @xmath162 .
denote @xmath163 , @xmath127 , @xmath128 , @xmath164 by @xmath165 , @xmath166 , @xmath167 , @xmath1 , respectively .
consider the shortest @xmath168-path constructed by changing from left to right the bits in which @xmath166 and @xmath167 differ : @xmath169 where addition is taken modulo 2 .
denote the @xmath170-th internal vertex @xmath171 of this path by @xmath172 for @xmath173 .
similarly , denote @xmath174 , @xmath175 , @xmath176 , @xmath177 by @xmath178 , @xmath179 , @xmath180 , @xmath181 , respectively .
set @xmath182 , @xmath183 , and recall that @xmath184 .
let @xmath185 be an isomorphism and let @xmath186 for @xmath187 .
assume @xmath188 .
then @xmath189 is of degree @xmath190 and hence @xmath191 .
this means that @xmath192 is adjacent to two vertices among @xmath193 . as @xmath2 is bipartite
, @xmath192 is adjacent to one of @xmath194 and one of @xmath195 . if @xmath192 is adjacent to @xmath179 and @xmath180 , then @xmath196 and hence @xmath197 . if @xmath192 is adjacent to @xmath178 and @xmath180 , then @xmath198 , a contradiction .
similarly we get contradictions if @xmath192 is adjacent to @xmath181 .
assume @xmath68 . then the vertices @xmath189 and @xmath199 of @xmath0 are of degree @xmath12 .
considering that @xmath200 is a path in @xmath0 , we see that @xmath201 .
therefore the vertices @xmath192 and @xmath202 of @xmath28 are of degree @xmath12 and @xmath203 .
we distinguish three cases .
* case 1 * : ( @xmath192 and @xmath202 are adjacent to a common vertex among @xmath193 ) + now , @xmath204 and hence @xmath205 . also , @xmath92 is odd as @xmath2 is bipartite , and thus @xmath206 .
considering that @xmath207 is a shortest path in @xmath2 , it follows that @xmath189 and @xmath208 have distance two and hence have a common neighbor which is different from @xmath209 in @xmath2 .
call it @xmath21 . then
, @xmath21 is not @xmath165 ,
@xmath166 , @xmath167 , or @xmath1 because of its distances from @xmath189 and @xmath208 .
therefore @xmath210 .
set @xmath211 .
then @xmath212 and hence @xmath213 , which means that @xmath214 .
similarly , @xmath215 .
hence in @xmath2 , @xmath192 and @xmath216 have three common neighbors : @xmath217 , @xmath218 , and one of @xmath193 .
this is a contradiction , because hypercubes are @xmath219-free .
+ + from now on we regard that @xmath192 and @xmath202 are not adjacent to a common vertex among @xmath193 .
* case 2 * : ( @xmath192 and @xmath202 are either adjacent to @xmath178 and @xmath179 or adjacent to @xmath180 and @xmath181 ) + we may without loss of generality assume the first
. then @xmath220 and hence @xmath221 . also , @xmath92 is even as @xmath2 is bipartite .
therefore @xmath222 or @xmath223 .
we distinguish two subcases . * case 2a * : ( @xmath222 ) + it is well - known that in @xmath2 a given edge lies in @xmath12 cycles of length 4 . among the @xmath224-cycles containing the edge @xmath225 ,
one contains @xmath166 and another contains @xmath167 .
this means that there are @xmath226 cycles of length 4 containing the edge @xmath225 in @xmath0 . among the @xmath224-cycles containing the edge @xmath227 ,
only one contains @xmath178 and @xmath179 together , and no other contains @xmath228 , or @xmath181 .
this means that there are @xmath190 cycles of length 4 containing the edge @xmath227 in @xmath28 , a contradiction . *
case 2b * : ( @xmath223 ) + it is known that for two given vertices at distance three , there are exactly three internally vertex - disjoint shortest paths in @xmath2 , and therefore there are such paths between @xmath189 and @xmath229 .
let @xmath230 be any one of them which is different from @xmath231 .
considering the distances of @xmath232 from @xmath233 , we obtain that @xmath234 and hence @xmath235 is a path in @xmath0 .
therefore @xmath236 is a path in @xmath28 . by the assumption that @xmath223 , there is also an @xmath237-path through @xmath178 and @xmath179 , implying that there are ( at least ) four internally disjoint shortest paths between @xmath192 and @xmath238 in @xmath2 , which is a contradiction . * case 3 * : ( @xmath192 is adjacent to one of @xmath178 and @xmath179 while @xmath202 is adjacent to one of @xmath180 and @xmath181 , or vice versa ) firstly , assume that @xmath192 is adjacent to @xmath179 while @xmath202 is adjacent to @xmath180 .
then @xmath239 hence , under this assumption @xmath240 .
alternatively , assume that @xmath192 is adjacent to @xmath178 while @xmath202 is adjacent to @xmath180
. then , @xmath241 a contradiction .
the other cases similarly lead to contradictions .
if @xmath242 , then @xmath118 in general no longer implies that @xmath110 .
for instance , it can be checked that @xmath243 despite the fact that @xmath244 . on the other hand
we pose the following conjecture .
[ conj : nu(f ) ] let @xmath6 be a non - trivial pair such that @xmath4 .
then @xmath118 .
this conjecture has also been verified for all dimensions up to 11 ( and all non - trivial pairs @xmath6 ) ; see appendix [ appendix ] for the sage code .
* acknowledgments : * this research was supported by the bilateral korean - slovenian project bi - kr/13 - 14005 and the international research & development program of the national research foundation of korea ( nrf ) funded by the ministry of science , ict and future planning ( msip ) of korea(grant number : nrf-2013k1a3a1a15003503 ) .
s.k . are supported by the ministry of science of slovenia under the grant p1 - 0297 .
j.l . and y.r .
are supported by the basic science research program through the national research foundation of korea funded by the ministry of education , science and technology grant 2011 - 00253195 .
+ + j.p.s research was sponsored by the national science foundation under grant number dms-1301692 .
the united states government is authorized to reproduce and distribute reprints not - withstanding any copyright notation herein . 99 l.j .
guibas , a.m. odlyzko , string overlaps , pattern matching , and nontransitive games , j. combin .
theory ser . a 30 ( 1981 ) 183208 .
hsu , fibonacci cubes a new interconnection topology , ieee trans .
parallel distrib .
syst . 4 ( 1993 ) 312 .
a. ili , s. klavar , y. rho , generalized fibonacci cubes , discrete math . 312
( 2012 ) 211 .
a. ili , s. klavar , y. rho , the index of a binary word , theoret .
452 ( 2012 ) 100106 .
a. ili , s. klavar , y. rho , parity index of binary words and powers of prime words , electron . j. combin . 19
( 2012 ) # p44 .
s. klavar , structure of fibonacci cubes : a survey , j. comb .
optim . 25 ( 2013 ) 505522 .
s. klavar , m. mollard , asymptotic properties of fibonacci cubes and lucas cubes , to appear in ann .
s. klavar , s. shpectorov , asymptotic number of isometric generalized fibonacci cubes , european j. combin .
33 ( 2012 ) 220226 .
j. liu , w .- j .
hsu , m. j. chung , generalized fibonacci cubes are mostly hamiltonian , j. graph theory 18 ( 1994 ) 817829 .
r. sedgewick , p. flajolet , analytic combinatorics , cambridge university press , cambridge , 2009 .
w. a. stein et al . , sage mathematics software ( version 6.0.0 ) , the sage development team , 2013 , http://www.sagemath.org .
a. taranenko , a. vesel , fast recognition of fibonacci cubes , algorithmica 49 ( 2007 ) 8193 .
n. zagaglia salvi , on the existence of cycles of every even length on generalized fibonacci cubes , matematiche ( catania ) 51 ( 1996 ) 241251 .
a. vesel , linear recognition and embedding of fibonacci cubes , to appear in algorithmica .
doi 10.1007/s00453 - 013 - 9839 - 3 .
in order to test our conjectures we define the function _ isom_classes _ that returns a dictionary whose entries are nontrivial pairs .
def allstr(k ) : ret = set ( ) for f in cartesianproduct(*[[0,1]]*k ) : f =
.join(f ) if inv(f ) not in ret and f[::-1 ] not in ret and inv(f[::-1 ] ) not in ret and inv(f)[::-1 ] not in ret : ret.add(f ) return ret | the generalized fibonacci cube @xmath0 is the subgraph of the @xmath1-cube @xmath2 induced on the set of all strings of length @xmath1 that do not contain @xmath3 as a substring .
it is proved that if @xmath4 then @xmath5 . the key tool to prove this result is a result of guibas and odlyzko about the autocorrelation polynomial associated to a binary string .
it is also proved that there exist pairs of strings @xmath6 such that @xmath4 , where @xmath7 and @xmath8 can not be obtained from @xmath3 by its reversal or binary complementation .
strings @xmath3 and @xmath8 with @xmath9 for which @xmath4 are characterized . |
i suggest to define the ( general ) misr as @xmath0 this has the following advantages : * the name misr still works . * for @xmath1
, it is the standard misr . * for @xmath2
, we have @xmath3 in close analogy to the asymptotics at @xmath2 .
essentially the @xmath4 is replaced by @xmath5 , while @xmath6 is replaced by @xmath7 . *
the definition of the gain @xmath8 is unchanged , , it is still @xmath9 this should resolve items 7 and 8 below . + here are some simulations for nakagami-@xmath4 fading .
so the efir does depend _ slightly _ on the fading parameter .
one comment about this : we have @xmath10 and since @xmath11 , @xmath12 , with very good approximation as soon as @xmath4 is not too small , we have @xmath13 quickly . + and here are the resulting gains : the gains are relatively constant .
@xmath8 increases slightly , so using @xmath8 for @xmath1 will give a conservative ( safe ) bound .
as far as @xmath14 is concerned , @xmath15 is essentially the no - fading case ( which i think has been observed elsewhere ) . 1 .
add prior work : xinchen , naoto miyoshi .
also cite li et al .
2 . simulation for lattices : + + so the efir is more sensitive to @xmath16 than the misr . in this range at least , it is roughly @xmath17 , but this obviously can not hold for smaller @xmath16 . + general observation : since for @xmath18 , @xmath19 a.s . , we have @xmath20 for all stationary point processes .
( new . ) * for advanced transmission schemes ( , bs silencing , joint transmission , interference cancellation ) , how is the efir defined ?
in particular , what is the relevant fading random variable and the relevant interference @xmath21 ? 4 .
plots of the horizontal shift @xmath22 for the lattices ( together with @xmath8 and @xmath14 ) : + + let @xmath23 , @xmath24 , and @xmath25 . +
* minimum gap . *
generally , @xmath26 and @xmath27 , , @xmath28 is finite , and close to @xmath29 in these cases .
so it appears that the gap is always first decreasing , then increasing .
+ if this is the case , then the asappp approximation ( with @xmath8 ) would be slightly optimistic for @xmath30 . +
* maximum gap . *
it is possible that the maximum is always assumed at @xmath8 or @xmath14 , i.e. , @xmath31 so a shift by the maximum of @xmath8 and @xmath14 would always result in an upper bound : + @xmath32 + here are some simulation results for the gains on a linear scale : + + + for @xmath33 , @xmath8 and @xmath14 are essentially the same , and for all @xmath16 , there is essentially no difference in @xmath14 between the square and the triangular lattice . 5 . * ( this is a less pertinent task . ) * taking this one step further , we could propose a highly accurate approximation by interpolating between @xmath8 and @xmath14 as follows : @xmath34 this way @xmath35 should be extremely accurate , with exactly the right asymptotics on both ends .
( i can try this out , say for the square lattice . )
6 . about the moment @xmath36
: is it always proportional to @xmath37 , so that the product @xmath38 ( and thus efir ) does not depend on the fading , and is it always proportional to @xmath39 ? ( we can at least simulate some scenarios . ) 7 .
gain @xmath8 with general fading : if @xmath40 , then @xmath41 \quad\rightarrow\quad g_0^{(m)}=\left(\frac{\e(\isr_{\rm ppp}^m)}{\e(\isr^m)}\right)^{1/m } \label{g0_a}\ ] ]
so it seems that @xmath8 depends on @xmath4 , but how strongly ?
simulations indicated that @xmath42 at least for modest @xmath4 .
+ in this case , the asappp approximation is @xmath43 where @xmath44 is the success probability for the ppp with fading parameter @xmath4 ( which is not known , of course ) .
+ at any rate , the second step here does _ not _ hold : @xmath45 8 .
related : is it possible to normalize " @xmath8 with respect to the fading parameter @xmath4 so that transmission techniques with different diversity can be compared ?
i am thinking of comparing the ppp with rayleigh fading with another scheme with fading parameter @xmath4 as follows : @xmath46 this way , instead of comparing against the ppp with diversity @xmath4 , the comparison can be against the ppp with diversity @xmath29 : @xmath47 what is the more natural definition , or ? 9 .
introduce the _ relative distance process ( rdp ) _
@xmath48 and show how the misr can be calculated as @xmath49 what can be said about the rdp for general stationary point processes ? 1 .
shall we also consider the relative vertical gap between the asappp approximations and the exact ( simulated ) curves ?
for @xmath50 near 0 , the relevant quantity is the relative outage @xmath51 and for @xmath52 , it is @xmath53 the two curves look as follows : + + the blue curve is the one obtained using the interpolation in .
+ the fact that the red and black solid curves in the right plot exceed @xmath29 for a short interval indicates that shifting by @xmath54 does not yield a lower bound valid for all @xmath50 .
however , a shift by @xmath55 may yield an upper bound .
2 . bounded path loss models ?
the decay at the tail will be faster than any polynomial ( * ? ? ?
* remark 6 ) .
it should be given by the fading .
3 . what is the gap in the no fading case ? 4 . a word on @xmath56 ?
the distribution of the signal - to - interference ratio ( sir ) is a key quantity in the analysis and design of interference - limited wireless systems . here
we focus on general single - tier cellular networks where users are connected to the strongest ( nearest ) base station ( bs ) .
let @xmath57 be a point process representing the locations of the bss and let @xmath58 be the serving bs of the typical user at the origin , , define @xmath59 .
assuming all bss transmit at the same power level , the downlink sir is given by @xmath60 where @xmath61 are iid random variables representing the fading and @xmath62 is the path loss law .
the complementary cumulative distribution ( ccdf ) of the sir is @xmath63 under the sir threshold model for reception , the ccdf of the sir can also be interpreted as the success probability of a transmission , , @xmath64 . in the case where @xmath65 is a homogeneous poisson point process ( ppp ) , rayleigh fading , and @xmath66 ,
the success probability was determined in @xcite .
it can be expressed in terms of the gaussian hypergeometric function @xmath67 as @xcite @xmath68 where @xmath69 . for @xmath70 ,
remarkably , this simplifies to @xmath71 in @xcite , it is shown that the same expression holds for the homogeneous independent poisson ( hip ) model , where the different tiers in a heterogeneous cellular network form independent homogeneous ppps .
for all other cases , the success probability is intractable or can at best be expressed using combinations of infinite sums and integrals .
hence there is a critical need for techniques that yield good approximations of the sir distribution for non - poisson networks .
it has recently been observed in @xcite that the sir ccdfs for different point processes and transmission techniques ( e.g. , bs cooperation or silencing ) _ appear to be merely horizontally shifted versions of each other _ ( in db ) , as long as their diversity gain is the same .
consequently , the success probability of a network model can be accurately approximated by that of a reference network model by scaling the threshold @xmath50 by this sir gain factor ( or shift in db ) @xmath72 , , @xmath73 formally , the horizontal gap at target probability @xmath74 is defined as @xmath75 where @xmath76 is the inverse of the ccdf of the sir and @xmath74 is the success probability where the gap is measured .
it is often convenient to consider the gap as a function of @xmath50 , defined as @xmath77 due to its tractability , the ppp is a sensible choice as the reference model .
if the shift @xmath22 is indeed approximately a constant , , @xmath78 , then @xmath72 can be determined by evaluating @xmath22 for an arbitrary value of @xmath50 . as shown in @xcite ,
the limit of @xmath22 as @xmath2 is relatively easy to calculate .
here we focus in addition on the positive limit @xmath52 and compare the two asymptotic gains to demonstrate the effectiveness of the idea of horizontally shifting sir distributions by a constant .
so the main focus of this paper are the asymptotic gains relative to the ppp , defined as follows . the asymptotic gains ( whenever the limits exist ) @xmath8 and @xmath14 are defined as @xmath79 where the ppp is used as the reference model .
some insights on @xmath8 are available from prior work . in @xcite
it is shown that for rayleigh fading , @xmath8 is closely connected to the mean interference - to - signal ratio ( misr ) .
the misr is the mean of the interference - to-(average)-signal ratio isr , defined as @xmath80 where @xmath81 is the mean received signal power averaged only over the fading .
not unexpectedly , the calculation of the misr for the ppp is relatively straightforward and yields @xmath82 ( * ? ? ?
( 8) ) in sec .
iv in this paper . ] .
since @xmath83 , the success probability can in general be expressed as @xmath84 where @xmath85 is the ccdf of the fading random variables .
for rayleigh fading , @xmath86 and thus @xmath87 , @xmath2 , resulting in @xmath88 and @xmath89 so , asymptotically , shifting the ccdf of the sir distribution of the ppp is exact . and the lower bound ( which is asymptotically tight ) @xmath90 for the ppp ( dash - dotted ) . the horizontal gap between the sir distributions of the ppp and the triangular lattice is 3.4 db for a wide range of @xmath50 values . the shaded band indicates the region in which the sir distributions for all stationary point process fall that are more regular than the ppp
] an example is shown in , where @xmath70 , which results in @xmath91 , while for the triangular lattice @xmath92 .
hence the horizontal shift is @xmath93 db . for rayleigh fading
, we also have the relationship @xmath94 by jensen s inequality , also shown in the figure . here
@xmath95 is a lower bound with asymptotic equality . in @xcite ,
the authors considered coherent and non - coherent joint transmission for the hip model and derived expressions for the sir distribution .
the diversity gain and the asymptotic pre - constants as @xmath2 are also derived . in @xcite ,
the benefits of bs silencing ( inter - cell interference coordination ) and re - transmissions ( intra - cell diversity ) in poisson networks with rayleigh fading are studied . for @xmath2
, it is shown that @xmath96 when the @xmath97 strongest interfering bss are silenced , while @xmath98 for intra - cell diversity with @xmath4 transmissions . for @xmath52 , @xmath99 and @xmath100 for bs
silencing and retransmissions , respectively .
the constants @xmath101 , @xmath102 , @xmath103 , and @xmath104 are also determined .
lastly , ( * ?
2 ) gives an expression for the limit @xmath105 for the ppp and the ginibre point process ( gpp ) with rayleigh fading .
for the gpp , it consists of a double integral with an infinite product . in @xcite ,
the authors consider a poisson model for the bss and define a new point process termed _ signal - to - total - interference - and - noise ratio ( stinr ) process_. they obtain the moment measures of the new process and use them to express the probability that the user is covered by @xmath106 bss . in our work , we consider a different map of the original point process based on relative distances , which results in simplified moment measures for the ppp and permits generalizations to other point process models for the base stations .
this paper makes the following contributions : * we define the _ relative distance process ( rdp ) _ , which is the relevant point process for cellular networks with nearest - bs association , and derive some of its pertinent properties , in particular the probability generating functional ( pgfl ) .
* we introduce the _ generalized misr _ , defined as @xmath107 , which is applicable to general fading models , and give an explicit expression and tight bounds for the ppp .
* we provide some evidence why the gain @xmath8 is insensitive to the path loss exponent @xmath16 and the fading statistics . *
we show that for all stationary point process models and any type of fading , the tail of the sir distribution always scales as @xmath108 , , we have @xmath109 , @xmath110 , where the constant @xmath111 captures the effects of the network geometry and fading . the asymptotic gain follows as @xmath112 and we have @xmath113 * we introduce the _ expected fading - to - interference ratio ( efir ) _ and show that the constant @xmath111 is related to the efir by @xmath114 .
consequently , @xmath14 is given by the ratio of the efir of the general point process under consideration and the efir of the ppp .
the base station locations are modeled as a simple stationary point process @xmath115 . without loss of generality , we assume that the typical user is located at the origin @xmath116 .
the path loss between the typical user and a bs at @xmath117 is given by @xmath66 , @xmath118 .
let @xmath85 denote the ccdf of the iid fading random variables , which are assumed to have mean @xmath29 .
we assume nearest - bs association , wherein a user is served by the closest bs .
let @xmath119 denote the closest bs to the typical user at the origin and define @xmath120 and @xmath121 . with the nearest - bs association rule
, the downlink sir of the typical user can be expressed as @xmath122 _ further notation : _
@xmath123 denotes the open disk of radius @xmath124 at @xmath116 , and @xmath125 is its complement .
in this section , we introduce a new point process that is a transformation of the original point process @xmath65 and helps in the analysis of the interference - to - signal ratio . from , the misr is defined as @xmath126 the first expectation is taken over @xmath65 and @xmath127 , while the second one is only over @xmath65 since @xmath128 .
since @xmath129 only depends on @xmath130 , it is apparent that the misr is determined by the relative distances of the interfering and serving bss .
accordingly , we introduce a new point process on the unit interval @xmath131 that captures only these relative distances .
[ def : rdp_def ] for a simple stationary point process
@xmath65 , let @xmath132 . the _ relative distance process ( rdp ) _ is defined as @xmath133 using the rdp , the @xmath134 can be expressed as @xmath135 and , since @xmath136 , the misr is @xmath137 for the stationary ppp , the cdfs of the elements @xmath138 of @xmath139 are @xmath140 , @xmath141 $ ] , as given in @xcite . summing the densities @xmath142 over @xmath143 yields the intensity measure @xmath144 .
it follows that the mean measure @xmath145 , @xmath146 .
the fact that the mean measure diverges near @xmath147 is consistent with the fact that @xmath139 is not locally finite on intervals @xmath148 . generally ,
since the misr only depends on the relative distances , the gain @xmath8 does not depend on the base station density .
for an rdp @xmath139 , the _ relative path loss process ( rplp ) _ is defined as @xmath149 for a ppp , the corresponding rplp has intensity @xmath150 , @xmath151 .
so @xmath33 is the neutral case , where the rplp is stationary on @xmath131 .
the success probability for rayleigh fading is given by the laplace transform of the @xmath134 : @xmath152 this rdp - based formulation has the advantage that it circumvents the usual two - step procedure , where first the conditional success probability given the distance to the serving base station @xmath153 is calculated and then an expectation with respect to @xmath153 is taken .
let @xmath154 .
the success probability equals @xmath155 expanding which we obtain the following lower bound @xmath156 the last step follows from and the identity @xmath157 since has the form of a pgfl , we first calculate the pgfl of the rdp generated by a ppp .
[ lem : pgfl_rdp ] when @xmath65 is a ppp , the probability generating functional of the rdp is given by @xmath158\triangleq \e\prod_{x\in \mathcal{r } } f(x ) = & \frac{1}{1 + 2\int_0 ^ 1 ( 1-f(x))x^{-3}\dd x } , \label{rdp_ppp_pgfl}\end{aligned}\ ] ] for functions @xmath159\mapsto[0,1]$ ] such that the integral in the denominator of is finite .
the pgfl @xmath160 $ ] can be calculated as @xmath158 & = \e\left[\prod_{x\in \mathcal{r } } f(x)\right]\\ & = \e\left[\prod_{x\in \phi\setminus\{x_0\ } } f\left ( \frac{\|x_0\|}{\|x\|}\right)\right]\\ & \stackrel{(a)}{=}\e_r \exp\left(-2 \pi \lambda\int_r^\infty a \left(1-f\left ( \frac{r}{a}\right)\right)\dd a\right)\\ & \stackrel{(b)}{=}\lambda\int\limits_0^\infty 2\pi r \exp\left(-2 \pi \lambda\int\limits_r^\infty a \left(1-f\left ( \frac{r}{a}\right)\right)\dd a\right)e^{-\lambda \pi r^2}\dd r\\ & = \lambda\int\limits_0^\infty 2\pi r \exp\left(-2 \pi r^2 \lambda\int\limits_1^\infty y \left(1-f\left ( \frac{1}{y}\right)\right)\dd y\right)e^{-\lambda \pi r^2}\dd r\\ & = \frac{1}{1 + 2\int_1^\infty y \left(1-f(1/y)\right)\dd y},\end{aligned}\ ] ] where @xmath161 is obtained from the pgfl of the ppp , which , for a general ppp @xmath162 with intensity measure @xmath163 , is given by @xmath164\triangleq \e\prod_{x\in\psi } f(x)=\exp\left(-\int_{\r^2 } [ 1-f(x)]\lambda(\dd x)\right ) .
\label{ppp_pgfl}\ ] ] writing the pgfl in polar form and conditioning on the distance to the nearest neighbor @xmath165 yields @xmath161 . in @xmath166 ,
we de - condition on @xmath153 using the the nearest - neighbor distribution of the ppp . using the substitution @xmath167
, we obtain the final result . when @xmath168 ( see ) , we retrieve the result in for poisson cellular networks with rayleigh fading .
it may be suspected that the rdp of a ppp is itself a ( non - stationary ) ppp on @xmath169 $ ] .
it is easily seen that this is not the case .
let @xmath162 be a ppp on @xmath170 with the same intensity function as @xmath139 , , @xmath144 .
if @xmath139 was a ppp , the success probability for rayleigh fading would follow from the pgfl of @xmath162 ( specializing to a ppp on @xmath169 $ ] ) as @xmath164=\exp\left(-\int_0 ^ 1 [ 1-f(r ) ] 2r^{-3}\dd r\right ) , \ ] ] and , in turn , the success probability would be given by @xmath171 instead of .
however , assuming @xmath139 to be poisson yields an approximation of the success probability , with asymptotic equality as @xmath2 .
the poisson approximation " is related to the actual value as @xmath172 this holds due to the identity @xmath157 rewriting and expanding , we have @xmath173 hence , only considering the dominant first term in the denominator as @xmath2 , we obtain @xmath174 , @xmath2 . the fact that @xmath175 for @xmath176 is an indication that the higher moment densities of the rdp are larger than those of the ppp .
this is indeed the case , as the following calculation of the moment densities shows .
[ lem : moment_densities_rdp ] when @xmath65 is a ppp , the moment densities of the rdp are given by @xmath177 first we obtain the factorial moment measures .
we use the simplified notation for @xmath178 since the rdp is not locally finite on @xmath179 $ ] . ]
@xmath180 the factorial moment measures are defined as @xmath181 where @xmath182 indicates that the sum is taken over @xmath183-tuples of _ distinct _ points .
the moment measures are related to the pgfl as @xcite @xmath184 \label{mom_measures_pgfl}\ ] ] evaluated at @xmath185 . using lemma [ lem : pgfl_rdp ]
we obtain @xmath186= \\ \frac{1}{1+\sum_{i=1}^n s_i(t_i^{-2}-1)}.\ ] ] differentiating with respect to @xmath187 and setting @xmath188 , we have @xmath189 the moment densities follow from differentiation , noting that @xmath190 denotes the start of the interval , which causes a sign change since increasing @xmath190 decreases the measure .
so the product densities are a factor @xmath191 larger than they would be if @xmath139 was a ppp .
this implies , interestingly , that the pair correlation function ( * ? ? ?
6.6 ) of the rdp of the ppp is @xmath192 , @xmath193 .
the moment densities of the rdp provide an alternative way to obtain the success probability for the ppp : @xmath194 where @xmath195 . from the definition of the moment densities
, we have @xmath196^n } \left(\prod_{i=1}^n\nu(\theta , t_i ) \right)\rho^{(n)}(t_1,t_2 , \hdots , t_n)\dd t_1\hdots \dd t_n \label{eq : pc_moment_measure}\end{aligned}\ ] ] using lemma [ lem : moment_densities_rdp ] , we have @xmath197}\nu(\theta , t ) t^{-3 } \dd t \right)^n,\\ & = \sum_{n=0}^\infty { \theta^n(-1)^n}\left ( \frac { \delta \ , _ 2f_1\left(1,1-\delta;2-\delta;-\theta \right)}{1 -\delta}\right)^n\\ & = \sum_{n=0}^\infty { \theta^n(-1)^n}\left ( \,_2f_1(1,-\delta ; 1-\delta ; -\theta)-1\right)^n\\ & = \frac{1}{\,_2f_1(1,-\delta ; 1-\delta ; -\theta)},\end{aligned}\ ] ] which equals the success probability given in .
we now characterize the pgfl of the rdp generated by a stationary point process .
let @xmath198 be a positive function of the distance @xmath165 and the point process @xmath199 .
the average @xmath200 $ ] can in principle be evaluated using the joint distribution of @xmath153 and @xmath65 , which is , however , known only for a few point processes .
thus we introduce an alternative representation of @xmath201 that is easier to work with .
the indicator variable @xmath202 , @xmath117 , equals one only when @xmath203 .
hence it follows that @xmath204 this representation of @xmath198 permits the computation of the expectation of @xmath201 using the campbell - mecke theorem ( * ? ? ?
we use the above idea in the next lemma to obtain the pgfl of a general rdp .
[ lem : rdp_general ] the pgfl of the rdp generated by a stationary point process @xmath65 is given by @xmath205=\lambda\int_{\r^2 } \calg^{!}_o\left [ f\left ( \frac{\|x\|}{\|\cdot + x\|}\right)\i(\cdot + x \in b(o,\|x\|)^{\rm c } ) \right ] \dd x , \label{rdp_gen_pgfl}\end{aligned}\ ] ] where @xmath206 is the pgfl of the point process @xmath65 with respect to the reduced palm measure .
using the representation in , we obtain @xmath158 & = \e\sum_{x\in \phi } \prod_{y\in \phi \setminus \{x\ } } f\left ( \frac{\|x\|}{\|y\|}\right)\i(\phi(b(o,\|x\|))=0 ) \\
& \stackrel{{\rm ( a)}}= \lambda\int_{\r^2 } \e_o^ !
\prod_{y\in \phi } f\left ( \frac{\|x\|}{\|y+x\|}\right)\i(\phi(b(-x,\|x\|))=0 ) \dd x\\ & = \lambda\int_{\r^2 } \e_o^ ! \prod_{y\in \phi } f\left ( \frac{\|x\|}{\|y+x\|}\right)\i(y\in b(-x,\|x\|)^{\rm c } ) \dd x\\ & = \lambda\int_{\r^2 } \mathcal{g}_o^!\left [ f\left ( \frac{\|x\|}{\|\cdot + x\|}\right)\i(\cdot + x \in b(o,\|x\|)^{\rm c } ) \right ] \dd x,\end{aligned}\ ] ] where @xmath161 follows from the campbell - mecke theorem . if @xmath65 is also rotationally invariant ( , motion - invariant ) , the reduced palm measure is also rotationally invariant and hence @xmath205 = \lambda2\pi \int\limits_{0}^\infty r \mathcal{g}_o^!\left [ f\left ( \frac{r}{\|\cdot + r\|}\right)\i(\cdot + r \in b(o , r)^{\rm c } ) \right ] \dd r , \label{eq : rot_inv}\end{aligned}\ ] ] where @xmath124 in any vector addition should be interpreted as @xmath207 .
_ taking @xmath65 to be a ppp , reduces to since @xmath158&=2\pi\lambda\int_0^\infty r\exp\bigg(-\lambda\int_{\r^2 } \bigg[1-f\left(\frac{r}{\|x+(r,0)\|}\right ) \i(x+(r,0)\in b(0,r)^{\rm c})\bigg]\dd x \bigg)\dd r\\ & = 2\pi\lambda\int\limits_0^\infty r\exp\bigg(-\lambda\pi r^2-\lambda\int\limits_{b(o , r)^c}\left[1-f\left(\frac{r}{\|x\|}\right)\right]\dd x\bigg)\dd r\\ & = 2\pi\lambda\int_0^\infty r\exp\left(-\lambda\pi r^2
- 2\pi\lambda\int_r^\infty y[1-f(r / y)]\dd y\right)\dd r.\end{aligned}\ ] ] the last expression equals the second - to - last line in the proof of lemma [ lem : pgfl_rdp ] .
similar to the case of ppp , we now obtain the moment measures of the rdp of a general stationary point process .
[ lem : rho_gen ] the factorial moment measures of the rdp @xmath139 generated by a stationary point process @xmath65 are @xmath208 \rho^{(n+1)}_\phi(y_0,y_1,\hdots , y_n ) \dd y_0\hdots \dd y_n , \label{fact_rdp}\ ] ] where @xmath209 and @xmath210 .
the product densities are @xmath211 where @xmath212^n } \e^{!}_{y_0,\left(\frac{\|y_0\|}{t_1},\varphi_1\right ) , \hdots,\left(\frac{\|y_0\|}{t_n},\varphi_n\right)}[g(\phi , y_0)]\nonumber\cdot \\ & \;\rho^{(n+1)}_\phi\left(y_0,\big(\frac{\|y_0\|}{t_1},\varphi_1\big),\ldots , \big(\frac{\|y_0\|}{t_n},\varphi_n\big)\right ) \dd \varphi_1\hdots \dd \varphi_n \dd y_0 .
\label{beta_n } \end{aligned}\ ] ] as before , we use the relationship . while the result can be obtained from the pgfl in lemma [ lem : pgfl_rdp ] , it is easier to begin with the definition of the pgfl .
we have @xmath158 & = \e\sum_{x\in \phi } \prod_{y\in \phi \setminus \{x\ } } f\left ( \frac{\|x\|}{\|y\|}\right)\i(\|y\|\geq \|x\|)\\ & = \e\sum_{x\in \phi } g(\phi\setminus \{x\},x ) \prod_{y\in \phi \setminus \{x\ } } f\left ( \frac{\|x\|}{\|y\|}\right).\end{aligned}\ ] ] we are interested in the derivative of the pgfl with the function @xmath213 .
so we have @xmath214 evaluated at @xmath185 .
expanding the inner product over the summation we obtain an infinite polynomial in the powers of @xmath215 and their products .
we observe that the only term that contributes to the derivative in a non - zero manner is the @xmath216 term .
its coefficient equals @xmath217 combining the summations , @xmath218 since @xmath219 implies @xmath220 , @xmath221 and , using ( * ? ? ?
* thm . 1 ) , @xmath222 \cdot \\
\rho^{(n+1)}_\phi(y_0,y_1,\hdots , y_n ) \dd y_0 \hdots \dd y_n,\ ] ] and the result follows . for the product densities , we convert the variables @xmath223 into polar coordinates @xmath224 , which yields @xmath225^n } \!\ !
\e^{!}_{y_0,(r_1,\varphi_1 ) , \hdots,(r_n,\varphi_n)}[g(\phi , y_0)]\nonumber \\ & \qquad\cdot \rho^{(n+1)}(y_0,(r_1,\varphi_1 ) , .. , ( r_n,\varphi_n ) ) \dd \varphi_1 \hdots
\dd \varphi_n \dd r_1\hdots \dd
r_n \dd y_0.\end{aligned}\ ] ] then differentiating using the leibniz rule with respect to @xmath226 , we obtain @xmath227^n } \e^{!}_{y_0,\left(\frac{\|y_0\|}{t_1},\varphi_1\right ) , \hdots,\left(\frac{\|y_0\|}{t_n},\varphi_n\right)}[g(\phi , y_0 ) ] \cdot \\
\rho^{(n+1)}_\phi\left(y_0,\left(\frac{\|y_0\|}{t_1},\varphi_1\right),\ldots , \left(\frac{\|y_0\|}{t_n},\varphi_n\right)\right ) \dd \varphi_1 \hdots \dd \varphi_n \dd y_0,\end{gathered}\ ] ] which equals . as in
, the moment densities of the rdp generated by a stationary point process can be used to compute its corresponding success probability .
in this section , we introduce and analyze the misr , including its generalized version , and apply it to derive a simple asymptotic expression of the sir distribution near @xmath147 using the gain @xmath8 .
we also give some insight why @xmath8 barely depends on the path loss exponent @xmath16 and the fading statistics .
the first result gives an expression for the misr for a general point process .
[ thm : misr_pp ] the misr of a motion - invariant point process @xmath65 is given by @xmath228 where @xmath229 is given in in lemma [ lem : rho_gen ] . using the rdp , the misr can be expressed as @xmath230 where @xmath161 follows from lemma [ lem : rho_gen ] . when @xmath65 is a ppp , from slivnyak s theorem and the fact that @xmath231 , we have @xmath232 and hence @xmath233 .
the _ generalized misr _ with parameter @xmath183 is defined as @xmath234 if there is a danger of confusion , we call @xmath235 the _ standard misr_. the generalized misr can be obtained by taking the corresponding derivative of the laplace transform @xmath236 at @xmath237 . in case of the ppp with rayleigh fading , the laplace transform is known and equals the success probability , thus @xmath238 for general fading , the laplace transform is not known , but we can still calculate the derivative at @xmath237 , as the following result for the ppp with general fading shows .
[ thm : gen_misr_ppp ] for a poisson cellular network with arbitrary fading , @xmath239 where @xmath240 are the ( incomplete ) bell polynomials . for @xmath241 ,
the generalized misr is lower bounded as @xmath242^{1/n } .
\label{misr_bound_ppp}\ ] ] for @xmath243 , equality holds , and for @xmath244 and @xmath245 , the lower bound is asymptotically tight .
we begin with the the laplace transform of the @xmath134 , given by @xmath246 where @xmath161 follows from lemma [ lem : pgfl_rdp ] .
let @xmath247 and @xmath248 . then @xmath249 . we are interested in the @xmath4-derivative of @xmath236 with respect to @xmath250 at @xmath237 , which can be computed using fa di bruno s formula @xcite as @xmath251 where @xmath240 are the ( incomplete ) bell polynomials .
we have @xmath252 , @xmath253 and @xmath254 which , when evaluated at @xmath237 , equals @xmath255 combining everything , we have @xmath256 from the definition of bell polynomials it follows that all the terms are positive , hence the result follows from @xmath257 .
the lower bound is obtained by only considering the terms @xmath258 and @xmath259 in the sum .
the bound becomes tight as @xmath244 _ and _ as @xmath245 since the term @xmath258 dominates the sum as @xmath245 since it is the only term with a denominator @xmath260 , while the term @xmath261 dominates as @xmath262 since it is the only one with a numerator @xmath263 .
hence we have two simpler asymptotically tight bounds for the generalized misr : @xmath264 for rayleigh fading , yields @xmath265 , @xmath244 .
shows @xmath266 for rayleigh fading as a function of the path loss exponent .
as can be observed , the term @xmath267 is dominant for @xmath268 even if the fading is severe ( rayleigh fading ) . for less severe fading ,
the term with @xmath269 is less relevant ; it only becomes dominant for unrealistically high path loss exponents ( @xmath270 ) .
the second moment of the @xmath134 follows from as @xmath271 and the third moment is @xmath272 for @xmath273 for the ppp with rayleigh fading as a function of the path loss exponent @xmath16 .
lower bound 1 " is the simple bound in , which holds irrespective of the fading and is asymptotically tight as @xmath274 , and lower bound 2 " is the bound in , which is valid for @xmath275 and is exact for @xmath243 . for @xmath268 , the two bounds are essentially identical . ]
* setting @xmath276 for all @xmath106 retrieves the result in ( * ? ? ?
3 ) on the pre - constant for @xmath4 transmissions in a poisson networks over rayleigh fading . + again the last term is @xmath277 , as asserted above .
previously we had the following bound @xmath278}{\alpha -2}\right)^n n!\right)^{1/n}.\ ] ] adding the first term ( corresponding to @xmath258 ) in , we have @xmath279}{\alpha -2}\right)^n n ! + \frac{2 \e[h^n]}{n\alpha-2}\right)^{1/n}.\ ] ] this is an asymptotically tight bound as @xmath280 . with rayleigh fading the bound equals @xmath279}{\alpha -2}\right)^n + \frac{2}{n\alpha-2}\right)^{1/n } ( n!)^{1/n}.\ ] ] this result also shows that @xmath281}{n}\right)^{1/n } \alpha^{-1/n } , \quad \alpha \to \infty .
\label{eq : asymp_misr}\end{aligned}\ ] ] writing the improved bound in terms of @xmath257 ( and setting @xmath128 ) , we have @xmath282 this bound becomes tight as @xmath244 _ and _ as @xmath283 : @xmath284 * an alternative way to derive the lower bound is as follows .
letting @xmath285 for @xmath286 , we expand @xmath287 as @xmath288 where the expression contains @xmath106 sums . ignoring all but the first and last terms of the expansion , we obtain @xmath289^n } ( x_1\cdots x_n)^{\alpha } \rho_\phi^{(n)}(x_1,\ldots , x_n ) \dd x_1\cdots\dd x_n
\\ & = \e(h^n)\frac{\delta}{n-\delta } + n!\ , \misr_1^n , \end{aligned}\ ] ] which equals the result in . for nakagami-@xmath4
fading , @xmath266 is decreasing with increasing @xmath4 since the moments @xmath269 are decreasing with @xmath4 . as the lower bound @xmath267 does not depend on the fading
, @xmath266 approaches a non - trivial limit as @xmath12
. shows @xmath266 as a function of @xmath183 .
the increase is almost linear in @xmath183 . indeed , as @xmath290 , @xmath266 is proportional to @xmath183 for the usually encountered path loss exponents , as the following corollary establishes . for the ppp with rayleigh fading and @xmath268 , @xmath291 for the ppp with rayleigh fading and @xmath292
, it follows from that @xmath293 since the dominant term in for large @xmath183 is the one with @xmath294 , which increases geometrically ( or stays constant ) with @xmath183 for @xmath292 . for the factorial term , @xmath295 ,
hence we obtain @xmath296 .
_ using stirling s formula @xmath297 , this asymptotic result can be sharpened slightly .
equipped with the results from theorem [ thm : gen_misr_ppp ] , we can now discuss the gain @xmath8 for general fading , @xmath298 , for arbitrary @xmath299 . ] . if @xmath300 , @xmath298 , then , for @xmath2 , we have @xmath301 $ ] , hence @xmath302 the asappp approximation follows as @xmath43 where @xmath44 is the success probability for the ppp with fading parameter @xmath4 , which is not known in closed - form . in @xcite , the sir ccdf for a poisson cellular network
when @xmath127 is gamma distributed is discussed .
however , we have the exact @xmath303 from and the lower bound @xmath304 . transmissions in rayleigh fading channels over a single transmission in nakagami-@xmath4 fading channels , for poisson networks with @xmath70 . ] for nakagami-@xmath4 fading , the pre - constant is @xmath305 , and we have @xmath306 \\ & \lesssim 1-\frac{m^{m-1}}{\gamma(m)}\misr_1 m!\theta^m \\ & = 1-\misr_1(m\theta)^m , \end{aligned}\ ] ] where @xmath307 indicates an upper bound with asymptotic equality . adding the second term in the lower bound and noting that @xmath308 yields the slightly sharper result @xmath309 .\ ] ] the gain for general fading is applicable to arbitrary transmission techniques that provide the same amount of diversity , not just to compare different base station deployments . as an example , we determine the gain from selection combining of the signals from @xmath4 transmissions over rayleigh fading channels with a single transmission over nakagami-@xmath4 fading channels , both for poisson distributed base stations .
the misr for the selection combining scheme follows from ( * ? ? ?
shows that there is a very small gain from selection combining .
simulation results indicate that at least for moderate @xmath4 , the scaling @xmath310 holds for arbitrary motion - invariant point processes .
this implies that @xmath311 , which indicates that @xmath8 is insensitive to the fading statistics for small to moderate @xmath4 .
next we show that the gain is also insensitive to the path loss exponent @xmath16 .
illustrates the densities of the square and triangular lattices relative to the ppp s , which is @xmath312 , @xmath313 $ ] , as derived after def .
[ def : rdp_def ] . since the relative densities are roughly constant over the @xmath169 $ ] interval , the gains do not depend strongly on @xmath16 .
indeed , if the density of the rdp of a general point process could be expressed as @xmath314 , we would have @xmath315 irrespective of @xmath16 .
another way to show the insensitivity of the gain to @xmath16 is by exploring the asymptotic behavior of the misr for general point processes given in theorem [ thm : misr_pp ] in the high-@xmath16 regime .
the result is the content of the next lemma . for a motion - invariant point process @xmath65 , @xmath316 where @xmath317 } \e^{!}_{y_0,\left(\|y_0\|,\varphi_1\right ) } [ g(\phi , y_0)]\rho^{(2)}_\phi\left(y_0,\left(\|y_0\|,\varphi_1\right)\right ) \dd \varphi_1 \dd y_0.\ ] ] the @xmath6 for a general point process
is given by theorem [ thm : misr_pp ] as @xmath318 using the laplace asymptotic technique ( * ? ? ?
6.419 ) , @xmath319 this shows that @xmath6 for arbitrary point processes decays as @xmath320 , which implies @xmath8 approaches a constant for large @xmath16 ( see ( a ) ) .
in this section , we define the _ expected fading - to - interference ratio ( efir ) _ and explore its connection to the gain @xmath322 in .
we shall see that the efir plays a similar role for @xmath52 as the misr does for @xmath2 .
[ def : efir ] for a point process
@xmath65 , let @xmath323 and let @xmath127 be a fading random variable independent of all @xmath61 . the _ expected fading - to - interference ratio _
( efir ) is defined as @xmath324\right)^{1/\delta } , \label{efir_def}\ ] ] where @xmath325 is the expectation with respect to the reduced palm measure of @xmath65 .
here we use @xmath326 for the interference term , since the interference here is the total received power from all points in @xmath65 , in contrast to the interference @xmath327 , which stems from @xmath328 . _ remark .
_ for the ppp , the efir does not depend on @xmath329 , since @xmath330 . to see this ,
let @xmath331 be a scaled version of @xmath65 .
then @xmath332 and thus @xmath333 .
multiplying by the intensities , @xmath334 since @xmath335 .
the same argument applies to all point processes for which changing the intensity by a factor @xmath336 is equivalent in distribution to scaling the process by @xmath111 , , for point processes @xmath337 where @xmath338 .
this excludes hard - core processes with fixed hard - core distance but includes lattices and hard - core processes whose hard - core distance scales with @xmath339 .
[ lem : efir_ppp ] for the ppp , with arbitrary fading , @xmath340 the term @xmath341 in can be calculated by taking the expectation of the following identity which follows from the definition of the gamma function @xmath342 .
@xmath343 hence @xmath344 from slivnyak s theorem ( * ? ? ?
8.10 ) , @xmath345 for the ppp , so we can replace @xmath346 by the unconditioned laplace transform @xmath347 , which is well known for the ppp and given by @xcite @xmath348 from , we have @xmath349\gamma(1-\delta)s^\delta } s^{-1+\delta}\dd s \\ & = \frac{1}{\lambda \pi \e(h^\delta)\gamma(1-\delta)\gamma(1+\delta)}=\frac{\sinc\delta}{\lambda\pi \e(h^\delta)}.\end{aligned}\ ] ] so @xmath350 , and the result follows .
remarkably , @xmath351 only depends on the path loss exponent .
it can be closely approximated by @xmath352 .
next we use the representation in to analyze the tail asymptotics of the ccdf @xmath353 of the sir ( or , equivalently , the success probability @xmath354 ) .
[ thm : main ] for all simple stationary bs point processes @xmath65 , where the typical user is served by the nearest bs , @xmath355 from , we have @xmath356 . using the representation given in
, it follows from the campbell - mecke theorem that the success probability equals @xmath357\dd x,\end{gathered}\ ] ] where @xmath358 is a translated version of @xmath65 .
substituting @xmath359 , @xmath360\dd x\nonumber\\ & \stackrel{(a)}{\sim}\lambda\theta^{-\delta } \int_{\r^2 } \e^{!}_o \bar f\left(\|x\|^\alpha i_\infty \right ) \dd
x,\quad \theta\to\infty \label{eq : alt}\\ & \stackrel{(b)}{=}\lambda\theta^{-\delta } \e^{!}_o(i_\infty^{-\delta } ) \int_{\r^2 } \bar f_h\left(\|x\|^\alpha \right ) \dd x,\quad \theta\to\infty,\nonumber\end{aligned}\ ] ] where @xmath161 follows since @xmath361 and hence @xmath362 .
the equality in @xmath166 follows by using the substitution @xmath363 .
changing into polar coordinates , the integral can be written as @xmath364 where @xmath161 follows since @xmath365 @xcite . since @xmath128 and @xmath366
, it follows that @xmath367 . for rayleigh fading , from the definition of the success probability and theorem [ thm : main ] , @xmath368
hence the laplace transform of @xmath134 behaves as @xmath369 for large @xmath50 . hence using the tauberian theorem in @xcite
, we can infer that @xmath370 from theorem [ thm : main ] , the gain @xmath14 immediately follows . for an arbitrary simple stationary point process @xmath65 with efir given in def .
[ def : efir ] , the asymptotic gain at @xmath52 relative to the ppp is @xmath371 from theorem [ thm : main ] , we have that the constant @xmath111 in is given by @xmath114 .
@xmath372 follows from lemma [ lem : efir_ppp ] as @xmath373 .
the laplace transform of the interference in for general point processes can be expressed as @xmath374,\end{aligned}\ ] ] where @xmath206 is the probability generating functional with respect to the reduced palm measure and @xmath375 is the laplace transform of the fading distribution .
[ rayleigh fading ] [ cor : ray ] with rayleigh fading , the expected fading - to - interference ratio simplifies to @xmath376\dd x\right)^{1/\delta},\ ] ] where @xmath377 with rayleigh fading , the power fading coefficients are exponential , , @xmath378 . from , we have @xmath379 and the result follows from the definition of the reduced probability generating functional . for rayleigh fading , the fact that @xmath380 as @xmath52 was derived in ( * ? ? ?
* thm . 2 ) .
while theorem [ thm : main ] shows that @xmath381 , it is not clear , if the scaling is mainly contributed by the received signal strength or the interference .
intuitively , since an infinite network is considered , the event of the interference being small is negligible and hence for large @xmath50 , the event @xmath382 is mainly determined by the random variable @xmath383 .
this is in fact true as is shown in the next lemma . for all stationary point processes and arbitrary fading ,
the tail of the ccdf of the desired signal strength @xmath383 is @xmath384 the cdf of the distance @xmath153 to the nearest bs is @xmath385 for all stationary point processes @xcite .
hence @xmath386.\end{aligned}\ ] ] so the tail of the received signal power @xmath383 is of the same order @xmath369 , and the interference and the fading only affect the pre - constant . in the poisson case with rayleigh fading
, @xmath387 the same holds near @xmath388 .
if for the fading cdf , @xmath389 , @xmath298 , @xmath390 for the ppp , @xmath391 _ so on both ends of the sir distribution , the interference only affects the pre - constant . _
we now explore the tail of the distribution to the maximum sir seen by the typical user for exponential @xmath127 .
assume that the typical user connects to the bs that provides the _ instantaneously _ strongest sir ( as opposed to the strongest sir _ on average _ as before ) .
also assume that @xmath392 .
let @xmath393 denote the sir between the bs at @xmath394 and the user at the origin .
then @xmath395\dd x\\ & = \lambda\theta ^{-\delta } \int_{\r^2 } \calg^{!}_o [ \delta(x,\cdot)]\dd x.\end{aligned}\ ] ] from the above we observe that ( for exponential fading ) , @xmath396 which shows that the tail with the maximum @xmath397 connectivity coincides with the nearest neighbor connectivity . for a general @xmath118 , the pre - constant
is given by @xmath398 hence , when the bss are distributed as a poisson point process , @xmath399 in particular when @xmath400 , the success probability behaves like @xmath401 .
the laplace transform of the interference in a ppp is @xmath402\gamma(1-\delta)s^\delta).\ ] ] from , we have @xmath403&= \frac{1}{\gamma(\delta)}\int_0^\infty e^{-\lambda \pi \e[h^\delta]\gamma(1-\delta)s^\delta } s^{-1+\delta}\dd s \\ & = \frac{1}{\lambda \pi \e(h^\delta)\gamma(1-\delta)\gamma(1+\delta)}=\frac{\sinc(\delta)}{\lambda\pi \e(h^\delta)}.\end{aligned}\ ] ] hence from theorem [ thm : main ] @xmath404 this shows that the tail distribution in a ppp network does not depend on the fading distribution .
let @xmath405 be iid uniform random variables in @xmath169 $ ] .
the unit intensity ( square ) lattice point process @xmath65 is defined as @xmath406 . for this lattice , with rayleigh fading ,
the laplace transform of the interference is bounded as @xcite @xmath407 where @xmath408 is the epstein zeta function , @xmath409 is the riemann zeta function , and @xmath410 is the dirichlet beta function .
hence from @xmath411 the upper bound equals @xmath412 , and it follows that for rayleigh fading , @xmath413 for the square lattice point process with rayleigh fading and @xmath70 . the asymptote ( dashed line ) is @xmath414 and is tight as @xmath110 . ] as @xmath16 increases ( @xmath244 ) , the upper and lower bounds approach each other and thus both bounds get tight .
the success probability multiplied by @xmath415 , the @xmath416 asymptote and its bounds for a square lattice process are plotted in figure [ fig : lattice ] for @xmath70 .
we observe that the lower bound , which is 1.29 , is indeed a good approximation to the numerically obtained value @xmath417 , and that for @xmath418 db , the ccdf is already quite close to the asymptote
. for the square and triangular lattices , shows the gain as a function of @xmath50 and the asymptotic gains @xmath8 and @xmath14 for rayleigh fading .
interestingly , the behavior of the gap is not monotone .
it decreases first and then ( re)increases to @xmath14 .
it appears that @xmath419 . if this holds in general , a shift by the maximum of the two asymptotic gains always results in an upper bound on the sir ccdf .
shows the dependence of @xmath8 and @xmath14 on @xmath16 . as pointed out in subs .
[ sec : insens ] , @xmath8 is very insensitive to @xmath16 .
@xmath14 appears to increase slightly and linearly with @xmath16 in this range . and @xmath14 ( linear scale ) for square and triangular lattices for rayleigh fading as a function of @xmath16 . ]
determinantal ( fermion ) point processes ( dpps ) @xcite exhibit repulsion and thus can be used to model the fact that bss have a minimum separation .
the kernel of the dpp @xmath65 is denoted by @xmath420 and due to stationarity is of the form @xmath421 .
its determinants yield the product densities of the dpp , hence the name .
the reduced palm measure @xmath422 pertaining to a dpp with kernel @xmath423 is defined as @xmath424 whenever @xmath425 .
let @xmath426 denote the kernel associated with the reduced palm distribution of the dpp process .
the reduced probability generating functional for a dpp is given by @xcite @xmath427 \triangleq \e^{!}_o\left[\prod_{x\in \phi } f(x)\right]=\detf(\mathbf{1}-(1-f)k^o ) , \label{eq : pgfl_dpp}\end{aligned}\ ] ] where @xmath428 is the fredholm determinant and @xmath429 is the identity operator .
the next lemma characterizes the efir a general dpp with rayleigh fading .
when the bss are distributed as a stationary dpp , the efir with rayleigh fading is @xmath430 follows from corollary [ cor : ray ] and .
_ ginibre point processes : _ ginibre point processes ( gpps ) are determinantal point processes with density @xmath431 and kernel @xmath432 using the properties of gpps @xcite , it can be shown that @xmath433 from which @xmath434 can be evaluated using .
for the gpp for rayleigh fading with @xmath70 .
the asymptote ( dashed line ) is at @xmath435 . ] in , the scaled success probability @xmath436 and the asymptote @xmath437 are plotted as a function of @xmath50 for the gpp .
we observe a close match even for modest values of @xmath50 .
shows the simulated values of the gains @xmath8 and @xmath14 for the gpp as a function of the path loss exponent @xmath16 .
@xmath438 for all values of @xmath16 , while @xmath439 . and @xmath14 for the gpp with rayleigh fading as a function of @xmath16 .
this paper established that the asymptotics of the sir ccdf ( or success probability ) for arbitrary stationary cellular models are of the form @xmath440 for a fading cdf @xmath441 , @xmath298 . both constants @xmath442 and @xmath443 depend on the path loss exponent and the point process model , and @xmath442 also depends on the fading statistics . depending on the point process fading _ may _ also affect @xmath443 .
@xmath442 is related to the mean interference - to - signal - ratio ( misr ) . for @xmath1 , @xmath444 , and for @xmath445
, @xmath442 depends on the generalized misr .
@xmath443 is related to the expected fading - to - interference ratio ( efir ) through @xmath446 . for the ppp , @xmath447 .
the study of the misr is enabled by the relative distance process , which is a novel type of point process that fully captures the sir statistics .
a comparison of @xmath8 and @xmath14 shows that a horizontal shift of the sir distribution of the ppp by @xmath8 provides an excellent approximation of the entire sir distribution of an arbitrary stationary point process .
for all the point process models investigated so far ( which were all repulsive and thus more regular than the ppp ) , the gains relative to the ppp are between @xmath147 and about @xmath448 db , so the shifts are relatively modest .
higher gains can be achieved using advanced transmission techniques , including adaptive frequency reuse , bs cooperation , mimo , or interference cancellation . as long as the diversity gain of the network architecture
is known and the ( generalized ) misr can be calculated ( or simulated ) , the asappp method can be applied to arbitrary cellular architectures .
such extensions will be considered in future work .
a generalization to heterogeneous networks ( hetnets ) is proposed in @xcite .
the method can be expected to be applicable whenever the misr is finite .
this excludes networks where interferers can be arbitrarily close to the receiver under consideration while the intended transmitter is further away , such as poisson bipolar networks .
x. zhang and m. haenggi , `` a stochastic geometry analysis of inter - cell interference coordination and intra - cell diversity , '' _ ieee transactions on wireless communications _ , vol . 13 , no . 12 , pp .
66556669 , dec . 2014 .
a. guo and m. haenggi , `` asymptotic deployment gain : a simple approach to characterize the sinr distribution in general cellular networks , '' _ ieee transactions on communications _ , vol . 63 , no . 3 , pp .
962976 , mar .
2015 .
, `` asappp : a simple approximative analysis framework for heterogeneous cellular networks , '' dec .
2014 , keynote presentation at the 2014 workshop on heterogeneous and small cell networks ( hetsnets14 ) .
available at http://www.nd.edu/~mhaenggi/talks/hetsnets14.pdf .
b. blaszczyszyn and h. p. keeler , `` studying the sinr process of the typical user in poisson networks by using its factorial moment measures , '' _ ieee transactions on information theory _ , vol .
61 , pp . 67746794 , dec .
2015 .
s. t. veetil , k. kuchi , a. k. krishnaswamy , and r. k. ganti , `` coverage and rate in cellular networks with multi - user spatial multiplexing , '' in _ 2013 ieee international conference on communications ( icc13 ) _ , budapest , hungary , jun .
2013 .
m. haenggi and r. k. ganti , `` interference in large wireless networks , '' _ foundations and trends in networking _ , vol . 3 , no . 2 ,
pp . 127248 , 2008 , available at http://www.nd.edu/~mhaenggi/pubs/now.pdf .
r. giacomelli , r. k. ganti , and m. haenggi , `` outage probability of general ad hoc networks in the high - reliability regime , '' _ ieee / acm transactions on networking _ , vol .
19 , no . 4 , pp .
11511163 , aug .
j. b. hough , m. krishnapur , y. peres , and b. virg , _ zeros of gaussian analytic functions and determinantal point processes _ , ser .
university lecture series 51.1em plus 0.5em minus 0.4emamerican mathematical society , 2009 .
h. wei , n. deng , w. zhou , and m. haenggi , `` a simple approximative approach to the sir analysis in general heterogeneous cellular networks , '' in _ ieee global communications conference ( globecom15 ) _ , san diego , ca , dec . 2015 . | it has recently been observed that the sir distributions of a variety of cellular network models and transmission techniques look very similar in shape . as a result , they are well approximated by a simple horizontal shift ( or gain ) of the distribution of the most tractable model , the poisson point process ( ppp ) . to study and explain this behavior , this paper focuses on general single - tier network models with nearest - base station association and studies the asymptotic gain both at 0 and at infinity .
we show that the gain at 0 is determined by the so - called mean interference - to - signal ratio ( misr ) between the ppp and the network model under consideration , while the gain at infinity is determined by the expected fading - to - interference ratio ( efir ) .
the analysis of the misr is based on a novel type of point process , the so - called relative distance process , which is a one - dimensional point process on the unit interval [ 0,1 ] that fully determines the sir .
a comparison of the gains at 0 and infinity shows that the gain at 0 indeed provides an excellent approximation for the entire sir distribution .
moreover , the gain is mostly a function of the network geometry and barely depends on the path loss exponent and the fading .
the results are illustrated using several examples of repulsive point processes .
cellular networks , stochastic geometry , signal - to - interference ratio , poisson point processes . |
the * mbr * ( minimum bias rockefeller ) monte carlo ( mc ) simulation is an event generator addressing the contributions of three diffraction - dissociation processes to the total - inelastic @xmath0 cross section , we will be assuming @xmath0 collisions throughout this paper for simplicity , except as explicitly stated . ] : single - diffraction dissociation or single dissociation ( sd ) , in which one of the incoming protons dissociates , double - diffraction dissociation or double dissociation ( dd ) , in which both protons dissociate , and central dissociation ( cd ) or double - pomeron exchange ( dpe ) , where neither proton dissociates .
these processes are tabulated below , @xmath1 where @xmath2 and @xmath3 represent diffractively dissociated protons .
schematic diagrams are shown in fig .
[ fig:1 ] , along with diagrams for the total and elastic cross sections , which are also simulated in * mbr*. * mbr * predicts the energy dependence of the total , elastic , and total - inelastic @xmath0 cross sections , and fully simulates the above three diffractive components of the total - inelastic cross section .
the diffractive - event generation is based on a phenomenological renormalized - regge - theory model@xcite , originally developed for the cdf experiment .
we have implemented * mbr * in pythia8@xcite , where it can be activated with the flag : ` diffraction : pomflux ` = 5 .
this paper is organized as follows : in sec .
[ xsecs ] we present the formulae used to calculate the total and elastic cross sections , as well as the diffractive and non - diffractive components of the total - inelastic cross section ; in sec . [ evgen ] we describe details of the generation of diffractive events ; and in sec . [ cards ] we summarize the pythia8 `` steering cards '' for employing * mbr * in the simulation .
the total cross section ( @xmath4 ) is the sum of the total - elastic ( @xmath5 ) and total - inelastic ( @xmath6 ) cross sections .
the @xmath4 and @xmath5 are calculated in sec .
[ totelxsec ] . the total - inelastic cross section ( @xmath6 )
is obtained as @xmath7 .
the @xmath6 receives contributions from the diffractive components ( sd , dd , cd or dpe ) and from the non - diffractive ( nd ) cross section , defined as : @xmath8 where @xmath9 is the cross section for either @xmath10 or @xmath11 , assumed to be equal . the diffractive cross sections are calculated in sec . [ diffxsec ] .
the @xmath12 cross sections at a @xmath0 center - of - mass - energy @xmath13 are calculated as follows : @xmath14 & \text{for } \sqrt{s } \ge 1.8 \mbox { tev } , \end{cases } \label{eqtot}\ ] ] the term for @xmath15 tev , where the @xmath16 denotes @xmath17 , is obtained from a global regge - theory fit to pre - lhc data on @xmath18 and @xmath19 cross sections @xcite , while that for @xmath20 tev is a prediction of a model based on a saturated froissart bound@xcite .
the latter , which we normalize to the cdf measurement of @xmath4 at @xmath21 tev , @xmath22 , depends on two parameters : the energy at which the saturation occurs , @xmath23 , and the energy - scale parameter , @xmath24 , for which we use @xmath25 , divided by @xmath26 to obtain the cross section in eq .
( [ eqtot ] ) in mb .
the elastic cross section , @xmath27 , is calculated using @xmath4 from eq .
( [ eqtot ] ) , multiplied by the elastic - to - total cross - section ratio , @xmath28 , obtained from the global regge fit of ref .
the total inelastic cross section is calculated as @xmath29 .
the energy dependences of @xmath4 , @xmath5 and @xmath6 are shown in fig .
[ fig:2 ] , and cross - section values at @xmath30=0.3 , 0.9 , 1.96 , 2.76 , 7 , 8 and 14 tev are presented in tab .
[ tab:1 ] .
cross sections for sd , dd and cd ( or dpe ) are calculated using a phenomenological model discussed in detail in ref .
differential cross sections are expressed in terms of the pomeron ( @xmath31 ) trajectory , @xmath32 , the pomeron - proton coupling , @xmath33 , and the ratio of the triple-@xmath31 to the @xmath31-proton couplings , @xmath34 .
for sufficiently large rapidity gaps ( @xmath35 3 ) , for which @xmath31-exchange dominates , the cross sections may be written as , @xmath36\delta y}\right ] \cdot \left\ { \kappa \beta^2(0 ) \left ( \frac{s'}{s_{0}}\right)^{\epsilon}\right\ } , \label{eqsd}\\ \frac{d^3\sigma_{dd}}{dtd\delta y dy_0 } & = & \frac{1}{n_{\rm gap}(s ) } \left [ \frac{\kappa\beta^2(0)}{16\pi}e^{2[\alpha(t)-1]\delta y}\right ] \cdot \left\ { \kappa \beta^2(0 ) \left ( \frac{s'}{s_{0}}\right)^{\epsilon}\right\ } , \label{eqdd}\\ \frac{d^4\sigma_{dpe}}{dt_1dt_2d\delta y dy_c } & = & \frac{1}{n_{\rm gap}(s ) } \left[\pi_i\left [ \frac{\beta^2(t_i)}{16\pi}e^{2[\alpha(t_i)-1]\delta y_i}\right]\right ] \cdot \kappa \left\ { \kappa \beta^2(0 ) \left ( \frac{s'}{s_{0}}\right)^{\epsilon}\right\ } , \label{eqcd}\end{aligned}\ ] ] where @xmath37 is the square of the four - momentum - transfer at the proton vertex and @xmath38 is the rapidity gap width .
the variable @xmath39 in eq .
( [ eqdd ] ) is the center of the rapidity gap . in eq .
( [ eqcd ] ) , the subscript @xmath40 enumerates pomerons in the dpe event , @xmath41 is the total ( sum of two gaps ) rapidity - gap width in the event , and @xmath42 is the center in @xmath43 of the centrally - produced hadronic system . eqs .
( [ eqsd ] ) and ( [ eqdd ] ) are equivalent to those of standard - regge theory , as @xmath44 , the fractional forward - momentum - loss of the surviving proton ( forward momentum carried by @xmath31 ) , is related to the rapidity gap by @xmath45 .
the variable @xmath44 is defined as @xmath46 and @xmath47 , where @xmath48 ( @xmath49 , @xmath50 ) are the masses of dissociated systems in sd ( dd ) events . for dd events , @xmath51 , and for dpe @xmath52 . the pomeron - proton coupling , @xmath33 , is given by : @xmath53 where @xmath54 and @xmath55 is the proton form factor from ref .
@xcite : @xmath56 ^ 2 \approx a_1e^{b_1t}+a_2e^{b_2t}. \label{eqff}\ ] ] the right - hand side of eq .
( [ eqff ] ) is a double - exponential approximation of @xmath57 , with @xmath580.9 , @xmath59=0.1 , @xmath604.6 gev@xmath61 , and @xmath620.6 gev@xmath61 .
the term in curly brackets in eqs.([eqsd])-([eqcd ] ) is the @xmath31-@xmath63 total cross section at the reduced @xmath31-@xmath63 collision energy squared , @xmath64 .
the parameter @xmath65 is set to @xmath66 @xcite , and @xmath67 , where @xmath68 defines the total pomeron - proton cross section at an energy - squared value of @xmath69 , is set to @xmath70 or @xmath71 .
pointing again to eqs.([eqsd])-([eqcd ] , the renormalization parameter , @xmath72 , is defined as @xmath73 , where @xmath74 is the integral of the term in square brackets , corresponding to a pomeron flux .
the integral is calculated over all phase space in @xmath75 and in @xmath76 ( dd ) or @xmath77 ( dpe ) for @xmath78 .
the renormalization procedure we use , which is based on interpreting the pomeron flux as a ( diffractive ) gap - formation probability , yields predictions which are in very good agreement with diffractive measurements at cdf@xcite .
the cross - section formulae are used to generate events with large ( diffractive ) rapidity gaps , @xmath38 .
we suppress cross sections at small value of @xmath38 by multiplying eqs .
( [ eqsd])-([eqcd ] ) by : @xmath79 , \label{erffun}\ ] ] where @xmath80 is the error function @xcite , centered by default at @xmath81 with a width of @xmath82 . for sd
, this choice corresponds to suppressing events above the coherence limit ( @xmath83 ) @xcite . for dd , the choice of @xmath81 is somewhat arbitrary , because of the difficulty of unambiguously distinguishing a low-@xmath38 dd event from a nd event with a rapidity gap from ( exponentially suppressed ) fluctuations .
changing @xmath84 introduces event migrations between dd and nd samples .
such migrations have no impact on the total - inelastic cross section , as can be inferred from eq .
( [ eqnd ] ) . for dpe ,
the suppression is applied to the total - gap width , @xmath41 .
[ fig:3 ] presents the energy dependence of diffractive cross sections calculated as explained above .
the non - diffractive cross section , given by eq .
( [ eqnd ] ) is also shown .
cross - section values for @xmath30=0.3 , 0.9 , 1.96 , 2.76 , 7 , 8 and 14 tev are tabulated in tab .
[ tab:1 ] .
.cross sections in _ mb _ of the processes contributing to the minimum - bias sample , calculated as explained in sec .
[ totelxsec ] and sec . [ diffxsec ] , for selected values of @xmath30 . [ cols="^,^,^,^,^,^,^,^",options="header " , ]
the four - momenta of particles produced in the event are generated as described below , in secs .
[ gensd]-[gencd ] .
the results of the simulation for sd and dd at @xmath30 = 7 tev are presented in fig .
[ fig:4 ] , which shows differential cross sections as a function of rapidity - gap width , @xmath38 , compared to predictions of pythia8 - 4c ( rescaled schler & sjstrand model@xcite , ` diffraction : pomflux = 1 ` ) .
the distribution of @xmath85 vs. the total rapidity - gap - width , @xmath86 , and vs. the width of an individual gap , @xmath87 , for cd at @xmath30 = 7 tev is shown in fig .
[ fig:5 ] .
events are generated by first choosing the rapidity - gap width , @xmath38 , according to eq .
( [ eqsd ] ) integrated over @xmath37 : @xmath88 the range of the generation is defined by @xmath89 and @xmath90 , where @xmath91`mbrm2min ` .
the term : @xmath92,\ ] ] is added to suppress events at low values of @xmath38 , as explained in sec .
[ diffxsec ] .
a value of @xmath37 is then chosen according to : @xmath93 where @xmath57 is given by eq .
( [ eqff ] ) and the integration is performed up to @xmath94 , with @xmath45 .
the diffractive mass is calculated as @xmath95 .
the four - momenta of the outgoing proton and the dissociated mass system are calculated using mandelstam variables for a two - body scattering process , as implemented in pythia8 for other ` diffraction : pomflux ` options .
events are generated by first choosing the rapidity - gap width according to eq .
( [ eqdd ] ) integrated over @xmath37 .
( [ eqdd ] ) is divergent as @xmath96 . in order to remove the divergence ,
the integration over @xmath37 is performed within the limits from @xmath97 to @xmath98 .
then , @xmath38 is chosen from the distribution : @xmath99 and the range of the generation is defined by @xmath89 and @xmath100 , where @xmath91`mbrm2min ` and @xmath69 . to further suppress events at low values of @xmath38 the term : @xmath101,\ ] ]
is used as explained in sec .
[ diffxsec ] .
the variable @xmath37 is chosen according to : @xmath102 in the range from @xmath97 to @xmath98 .
then , the center of the rapidity gap , @xmath39 , is selected uniformly within the limits : @xmath103 and the diffractive masses are calculated as : @xmath104 @xmath105 the four - momenta of the outgoing dissociated mass systems are calculated using mandelstam variables for a two - body scattering process , as implemented in pythia8 for other options of ` diffraction : pomflux ` .
events are generated by first choosing the total rapidity gap width , @xmath38 , according to eq .
( [ eqcd ] ) , integrated over @xmath106 and @xmath107 : @xmath108 where : @xmath109 and the integration is performed from @xmath89 to @xmath90 , where @xmath91`mbrm2min ` . for events at low values of @xmath38
we suppress individual gaps with the factor : @xmath110.\ ] ] then , the direction of the centrally - produced hadronic system , @xmath42 , is selected uniformly within the region : @xmath111 and rapidity gaps corresponding to each of the two pomerons are calculated as : @xmath112 @xmath113 the four - momentum transfers squared at each proton vertex , @xmath106 and @xmath107 , are generated according to : @xmath114 up to @xmath115 , where @xmath116 and @xmath117 .
then , the @xmath118 and @xmath119 of outgoing protons are calculated as @xmath120 and @xmath121 , where @xmath122 is the incoming proton momentum .
finally , the four - momentum of the hadronic system is calculated from the sum of the four - momenta of the pomerons , each calculated as a difference between the incoming and outgoing proton four - vectors .
the * mbr * generation is activated with the following flag : ` diffraction : pomflux = 5 ` .
the simulation works for the @xmath0 , @xmath123 and @xmath124 scattering , and the beam setup is checked in the pythia8 initialization step .
it is assumed that the user will veto the event generation , after ` pythia::init ( ) = false ` is returned for other beam particles .
the simulation of the cd ( dpe ) process is implemented in pythia8 for the first time .
it is activated with the flag ` softqcd : centraldiffractive = on ` .
the corresponding - process number is set to 106 , naturally extending the list of soft qcd processes . in the event record
, the outgoing protons information is written in rows 3 and 4 , and the centrally - dissociated hadronic system occupies row 5 ( represented by the ` rho_diff0 ` pseudo - particle ) .
below , we present the default values of parameters , used when the * mbr * simulation is activated : * the parameters of the pomeron trajectory , @xmath125 : + + ` diffraction : mbrepsilon = 0.104 ` + ` diffraction : mbralpha = 0.25 ` * the pomeron - proton coupling , @xmath126 ( @xmath127 ) , and the total pomeron - proton cross section , @xmath128 ( @xmath129 ) , see sec . [ diffxsec ] : + + ` diffraction : mbrbeta0 = 6.566 ` + ` diffraction : mbrsigma0 = 2.82 ` * the lowest mass - squared value of the dissociated system , @xmath130 , used to evaluate the highest allowed rapidity gap width , @xmath131 , see sec .
[ evgen ] : + + ` diffraction : mbrm2min = 1.5 ` * the minimum width of the rapidity gap used in the calculation of @xmath72 ( flux renormalization , sec .
[ diffxsec ] ) : + + ` diffraction : mbrdyminsdflux = 2.3 ` + ` diffraction : mbrdyminddflux = 2.3 ` + ` diffraction : mbrdymincdflux = 2.3 ` * the parameters @xmath132 and @xmath133 , used for the cross - section suppression at low @xmath38 ( non - diffractive region ) , see eq .
( [ erffun ] ) : + + ` diffraction : mbrdyminsd = 2.0 ` + ` diffraction : mbrdymindd = 2.0 ` + ` diffraction : mbrdymincd = 2.0 ` + + ` diffraction : mbrdyminsigsd = 0.5 ` + ` diffraction : mbrdyminsigdd = 0.5 ` + ` diffraction : mbrdyminsigcd = 0.5 ` in addition , if the option ` diffraction : pomflux = 5 ` is set , no dampening of the diffractive cross sections is performed ( no effect of ` sigmadiffractive : dampen = on ` ) ; however , the user can still set his / her own cross sections by hand ( ` sigmatotal : setown= on ` ) .
we present the * mbr * ( minimum bias rockefeller ) monte carlo simulation , which has been tested using cdf data , and discuss its implementation in the pythia8 generator .
the double - pomeron - exchange process is included in pythia8 for the first time .
the simulation is designed to apply to all minimum bias processes in the lhc energy range .
we thank torbjrn sjstrand and stephen mrenna for many useful discussions and suggestions .
we are indebted to torbjrn sjstrand for his guidance in implementing the * mbr * code in the pythia8 framework .
k. goulianos , _ hadronic diffraction : where do we stand ? _ , hep - ph/0407035 ; + k. goulianos , _ diffraction in qcd _ , arxiv : hep - ph/020314 . t. sjstrand , s. mrenna and p. skands , jhep05 ( 2006 ) 026 , comput .
( 2008 ) 852 , arxiv : hep - ph/0603175,arxiv:0710.3820 .
r. j. m. covolan , k. goulianos , j. montanha , phys .
b * 389 * , 176 ( 1996 ) .
k. goulianos , _ diffraction , saturation and pp cross sections at the lhc _ ,
a. donnachie , p. v. landshoff , nucl .
b * 303 * , 634 ( 1988 ) .
k. goulianos , phys . lett . *
b358 * , 379 ( 1995 ) .
t. affolder et al .
( cdf collab . ) , _ double diffraction dissociation at the fermilab tevatron collider _ , phys . rev
. lett . * 87 * , 141802 ( 2001 ) .
d. acosta et al .
( cdf collab . ) , _ inclusive double pomeron exchange at the fermilab tevatron @xmath124 collider _ , phys .
lett . * 93 * , 141601 ( 2004 ) .
r. j. m. covolan , k. goulianos , j. montanha , phys .
d * 59 * , 114017 ( 1999 ) .
k. goulianos , _ diffractive interactions of hadrons at high energies _ , phys . rep . 101 , 169 ( 1983 ) . | we present the * mbr * ( minimum bias rockefeller ) monte carlo simulation of ( anti ) proton - proton interactions and its implementation in the pythia8 event generator .
we discuss the total , elastic , and total - inelastic cross sections , and three contributions from diffraction dissociation processes that contribute to the latter : single diffraction , double diffraction , and central diffraction or double - pomeron exchange .
the event generation follows a renormalized - regge - theory model , successfully tested using cdf data .
based on the * mbr*-enhanced pythia8 simulation , we present cross - section predictions for the * lhc * and beyond , up to collision energies of 50 tev . |
galaxy clustering measurements of large area surveys , such as the 2df galaxy redshift survey ( 2dfgrs , @xcite ) , sloan digital sky survey ( sdss , @xcite ) , deep extragalactic evolutionary probe 2 redshift survey ( deep2 , @xcite ) , agn and galaxy evolution survey ( ages , @xcite ) , and vimos - vlt deep survey ( vvds , @xcite ) with many thousands of galaxies have quantified the clustering dependence on galaxy properties such as morphological type , luminosity , spectral type , and redshift to @xmath5 ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
the general trend is that red , more massive , brighter , and/or elliptical galaxies cluster more strongly than blue , less massive , fainter , and/or spiral galaxies .
halo occupation distribution ( hod ) modelling of these results has shown that there is an almost linear increase in the mean number of satellite ( non - central ) galaxies as a function of increasing dark matter halo ( dmh ) mass ( e.g. , @xcite ; @xcite ) . compared to projected angular clustering measurements ,
three dimensional ( 3d ) clustering measurements have greater statistical accuracy and minimize systematic errors associated with model assumptions in using limber s de - projection ( @xcite ) . however , 3d correlation functions require extremely large numbers of objects with secure redshift information .
clustering measurements of the auto - correlation function ( acf ) of active galactic nuclei ( agns ) reveal how agns are spatially distributed in the universe .
these measurements can constrain the agn host galaxies , determine the typical dmh mass in which agns reside and the distribution of agns with dhm mass , test theoretical model predictions , and address which physical processes are triggering agn activity ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . however , a major challenge in agn clustering measurements is to overcome the limitation caused by the low space density of agns .
narrow emission line diagnostics ( e.g. , @xcite ; @xcite ) have been used to identify narrow - line agns at low redshifts .
large sky area surveys with extensive spectroscopic follow - up programs have recently allowed narrow - line agn clustering measurements for samples of up to 90,000 agns ( @xcite ) .
studies such as @xcite , @xcite , @xcite , and @xcite explore the clustering strength of low redshift ( @xmath6 ) narrow - line agns samples with respect to various agn properties ( e.g. , black hole mass , radio emission ) .
optically - selected narrow - line agns have a clustering strength similar to all galaxies ( both blue and red ) .
narrow - line radio - loud agns cluster more strongly than agn without radio - emission and identical samples of quiescent galaxies .
consequently , denser environment of galaxies increase the probability of hosting a narrow - line radio - loud agn .
broad - line agns are , in general , more luminous than narrow - line agns .
moreover , their energy production dominates over the host galaxy star light .
large area optical surveys have spectroscopically identified tens of thousands of broad - line agns ( e.g. , @xcite ; @xcite ) .
the co - moving number density of broad - line agns is very low in the low redshift universe ( @xmath7 ) .
the situation improves at higher redshifts as the co - moving agn number distribution peaks at @xmath8 .
furthermore , a given observed area corresponds at higher redshift to a larger observed co - moving volume .
these two facts favor broad - line agn clustering measurements at @xmath9 ( e.g. , @xcite ; @xcite ; @xcite ) .
x - ray surveys have also been used to select agns .
these surveys cover much less sky area , e.g. , @xmath20.1 - 10 deg@xmath10 , than optical surveys .
further , they sample lower agn luminosities than optical surveys and contain a significant fraction of obscured agns . in order to probe the required large observed co - moving volume needed for clustering measurements using these relatively small sky areas , 3d clustering measurements of x - ray
selected agns have mainly been conducted at @xmath11 ( e.g. , @xcite ; @xcite ; @xcite ) .
optically - selected agns at redshifts of @xmath12 appear to reside in somewhat lower host dmh masses ( @xmath13 m@xmath14 , e.g. , @xcite ; @xcite ; @xcite ) than _ xmm - newton _ and _ chandra _ selected agn samples ( @xmath15 m@xmath14 , e.g. , @xcite ; @xcite ; @xcite ; @xcite ) .
possible explanations of the differences in the clustering signals include either the presence of a large fraction of x - ray absorbed narrow - line type ii agns in the _ xmm - newton _ and _ chandra _ samples or the different luminosities of the agn samples . at lower redshifts
3d broad - line x - ray agn clustering measurements have been associated with large uncertainties , due to small sample sizes ( e.g. , @xcite ) . to achieve smaller uncertainties in agn clustering measurements , @xcite use the cross - correlation function ( ccf ) of agns with a tracer set of galaxies that contains a large number of objects .
the agn acf is then inferred from the ccf , which has many more pairs at a given separation and hence significantly reduces the uncertainties in the spatial correlation function compared with the direct measurements of the agn acf . in @xcite ( hereafter paper i )
we use the cross - correlation technique to calculate the ccf between broad - line x - ray selected agns from the _ rosat _ all sky survey ( rass ) and sdss luminous red galaxies ( lrgs ) in the redshift range @xmath16 .
the potential of rass , which is currently the most sensitive x - ray all sky survey , can only be maximally exploited when it is combined with other large - area surveys such as sdss .
the unprecedented low uncertainties of the inferred broad - line agn acf from the rass / sdss combination allows us to split our sample into subsamples according to their x - ray luminosities . from this work , for the first time , we report an x - ray luminosity dependence of broad - line agn clustering in that higher luminosity agns cluster more strongly than their lower luminosity counterparts .
we conclude that low luminosity broad - line rass / sdss agns cluster similarly to blue galaxies at the same redshift , while high luminosity rass / sdss agns cluster similarly to red galaxies . in our second paper ( @xcite ;
hereafter paper ii ) , we apply a halo occupation distribution ( hod ) modeling technique to the agn - lrg ccf in order to move beyond determining the typical dmh mass based on the clustering signal strength and instead constrain the full distribution of agns as a function of dmh mass . to do this ,
we develop a novel method of applying the hod model directly to the ccf .
the hod modeling significantly improves the analysis of the ccf because it properly uses the fourier transformed linear power spectrum in the `` two halo term '' as well as the non - linear growth of matter in the `` one halo term '' through the formation and growth of dmhs .
this results in significant improvements over the standard method , which is used in paper i , of fitting both regimes with a phenomenological power law .
one of the important results of this analysis is that at @xmath16 the mean number of satellite agns in a dmh does not proportionally increase with halo mass , as is found for satellite galaxies .
the agn fraction among satellite galaxies actually _ decreases _ with increasing dmh mass beyond @xmath17 @xmath18 m@xmath14 .
in this paper we extend the scope of our previous papers to both somewhat lower and higher redshifts to obtain broad - line agn clustering results at @xmath7 , where very precise narrow - line agn clustering measurements exist but broad - line agn clustering is poorly constrained .
this is crucial for studying the possible evolution and luminosity dependence of broad - line agn clustering from low to high redshifts .
the dominant process that triggers agn activity could be a function of redshift and/or halo mass , which may be reflected in the clustering properties .
we also study the clustering signal of both x - ray and optically - selected broad - line agn samples , and test whether the exclusion of radio - detected broad - line agns changes our results . as the same statistical method and galaxy tracer sets are used to infer the clustering signal for x - ray and optically - selected broad - line agns
, we can explore differences in the clustering among the different selection techniques with low systematic uncertainties .
furthermore , we derive bias parameters by applying the hod modeling directly to all ccfs . in paper ii
we show that using power law fit results even in the non - linear regime , as is commonly done in the literature , is appropriate to detect difference in the clustering properties between different samples .
however , the derived bias parameters and dmh masses based on these fits should be interpreted with caution as the fit does not only consider the linear regime ( two halo term ) , but also the non - linear regime ( one halo term ) .
consequently , here we derive the bias for each agn sample using hod modeling of the ccf .
full detailed results of the hod modeling of the ccfs presented in this paper will be given in a future paper ( miyaji et al . in preparation ) .
this paper is organized as follows . in section 2
we describe the properties of the different galaxy tracers sets used at different redshifts , while section 3 gives the details of our different agn samples . in section 4
we briefly summarize the cross - correlation technique , how the agn acf is inferred from this , and present our results .
we apply the hod modeling in section 5 .
our results are discussed in section 6 and we conclude in section 7 . throughout the paper , all distances
are measured in co - moving coordinates and given in units of @xmath18mpc , where @xmath19kms@xmath20mpc@xmath20 , unless otherwise stated .
we use a cosmology of @xmath21 , @xmath22 , and @xmath23 , which is consistent with the @xmath24 data release 7 ( @xcite ; table 3 ) .
the same cosmology is used in papers i & ii . luminosities and absolute magnitudes are calculated for @xmath25 .
we use ab magnitudes throughout the paper .
all uncertainties represent to a 1@xmath3 ( 68.3% ) confidence interval unless otherwise stated .
the crucial ingredient in the cross - correlation method is the tracer set , a sample with a large number of objects that traces the underlying dark matter density distribution .
the properties of the tracer set determines the redshift range over which the method can be applied .
the agn samples of interest are necessarily limited to the same redshift range and geometry as the corresponding tracer set . as the rass / sdss selected agns ( @xcite ) are based on the sdss data release 5 ( dr5 ) , we consequently limit the tracer sets to the same survey geometry when we compute the clustering measurements of the x - ray selected agn . the sdss geometry and completeness are expressed in terms of spherical polygons ( @xcite ) .
this file is not publicly - available for dr5 , therefore , we use the latest version available prior to dr5 : the dr4 + geometry file .
however , optically - selected sdss agns ( @xcite ) make use of the full sdss survey ( dr7 ) .
consequently , we consider tracer sets from the dr7 geometry whenever we compute ccfs of optically - selected sdss agns . in the redshift range of @xmath26 we use sdss main galaxies for the tracer set , while sdss luminous red galaxies ( lrgs , @xcite ) are used for the corresponding cross - correlation sample at @xmath16 ( same as for paper i ) .
very luminous red galaxies are used as a tracer set at @xmath27 .
we will refer to the latter sample as the extended lrg sample. above @xmath28 the number of galaxies with spectroscopic redshifts in sdss decreases dramatically and does not allow the selection of further tracer sets with a sufficient density of objects . in the following subsections
we describe in detail the extraction of the various tracer sets and how we account for sdss fiber collisions . the sdss main galaxy sample ( @xcite )
is drawn from the nyu value - added galaxy catalog ( nyu vagc , @xcite ; @xcite ) , based on the sdss dr7 ( @xcite ) .
the photometric data covers an area of 10417 deg@xmath10 , while the spectroscopic data covers 7966 deg@xmath10 .
absolute magnitudes , based on petrosian fluxes , are @xmath29-corrected to @xmath30 ( @xcite ; @xcite ) , which is close to the median redshift of our sample .
we follow a scheme similar to @xcite , who use the nyu vagc to measure the clustering of various luminosity and color - selected galaxy subsamples .
similarly , we limit our sample to @xmath31 .
the bright limit avoids incompleteness due to galaxy deblending , and the faint limit accounts for the slightly - varying @xmath32-band magnitude limit over the sdss area ( nominal value @xmath33 ) .
the restriction of @xmath34 ensures a uniform flux limit throughout the whole sdss survey .
in addition , we create a volume - limited galaxy sample by selecting objects with an absolute magnitude of @xmath35 .
finally we limit the redshift range to @xmath36 . applying the above - mentioned selection criteria and considering only sdss dr7 areas with a spectroscopic completeness ratio of @xmath370.8
, we select 68273 galaxies with spectroscopic redshifts from the corresponding nyu vagc . the dr7 area covered by restricting the spectroscopic completeness ratio to @xmath370.8 is 7670 deg@xmath10 . the properties of this dr7 sdss main galaxy sample are summarized in table [ samples ] . as described above , the x - ray selected rass /
sdss agn samples are based on dr5 .
therefore , we further reconfigure the dr7 tracer sets ( in this case the sdss main galaxy sample ) to the dr4 + geometry to define a common survey geometry to use when measuring the clustering of x - ray selected agns . the restriction to
the dr4 + survey area with a dr7 spectroscopic completeness ratio of @xmath370.8 corresponds to an area of 5468 deg@xmath10 .
table [ samples ] lists the properties of the dr4 + sdss main galaxy sample .
an operational constraint of the sdss spectroscopic program is that two fibers can not be placed closer than 55 arcsec on a single plate .
overlapping spectroscopic plates compensate partially for the effect .
however , @xmath27% of the target galaxies can not be spectroscopically observed because of fiber collisions .
this observational bias is corrected by assigning to each galaxy that has not been observed the redshift of their nearest neighbor with a spectroscopic sdss redshift ( @xcite ) .
although one might be concerned that this simple method could overcorrect and result in too many close galaxy pairs at the same redshift , which would then distort clustering measurements , @xcite demonstrate that this correction procedure works very well .
they use @xmath38cdm @xmath39-body simulations and design three galaxy samples and measure the correlation function for three samples : i ) from the full simulated galaxy distribution , ii ) from simulated sdss data including fiber collision losses and not correcting for it , and iii ) from simulated sdss data that corrects for the fiber collision by assigning the redshift of their nearest spectroscopic neighbor .
they verify that the differences between i ) and iii ) are much smaller than the statistical uncertainties down to scales of @xmath40 @xmath18 mpc , while ii ) underestimates the correlation function at scales @xmath41 @xmath18 mpc .
therefore , we use the same fiber correction procedure for our main galaxy sample .
the nyu vagc provides this information through an sdss fiber collision corrected galaxy sample .
cccccccc + main galaxy & dr7 & @xmath42 & @xmath43 & 68273 & @xmath44 & 0.13 & -21.41 + sample & dr4 + & @xmath42 & @xmath43 & 48994 & @xmath44 & 0.13 & -21.41 + & & & & & & & + lrg sample & dr7 & @xmath45 & @xmath46 & 65802 & @xmath47 & 0.28 & -21.71 + & dr4 + & @xmath45 & @xmath46 & 45899 & @xmath48 & 0.28 & -21.71 + & & & & & & & + extended lrg & dr7 & @xmath49 & @xmath50 & 28781 & @xmath51 & 0.42 & -22.04 + sample & dr4 + & @xmath49 & @xmath50 & 19831 & @xmath52 & 0.42 & -22.04 + + total rass - agn & dr4 + & @xmath42 & @xmath53 log@xmath54 & 629 & @xmath55 & 0.12 & 43.59 + total rass - agn ( rq ) & dr4 + & @xmath42 & @xmath56 log@xmath57 & 504 & @xmath58 & 0.12 & 43.55 + low @xmath59 rass - agn & dr4 + & @xmath42 & @xmath60 log@xmath61 & 293 & @xmath58 & 0.11 & 43.25 + high @xmath59 rass - agn & dr4 + & @xmath42 & @xmath62 log@xmath63 & 336 & @xmath64 & 0.13 & 43.89 + low @xmath59 rass - agn ( rq ) & dr4 + & @xmath42&@xmath65 log@xmath66 & 253 & @xmath67 & 0.11 & 43.26 + high @xmath59 rass - agn ( rq ) & dr4+&@xmath42&@xmath68 log@xmath69 & 251 & @xmath70 & 0.13 & 43.85 + narrow line rass - agn & dr4 + & @xmath42 & @xmath71 log@xmath72 & 194 & @xmath73 & 0.11 & 43.32 + & & & & & & & + total rass - agn & dr4 + & @xmath45 & @xmath74 log@xmath75 & 1552 & @xmath76 & 0.25 & 44.17 +
total rass - agn ( rq ) & dr4 + & @xmath45 & @xmath74 log@xmath77 & 1337 & @xmath78 & 0.25 & 44.15 + low @xmath59 rass - agn & dr4 + & @xmath45 & @xmath79 log@xmath80 & 990 & @xmath81 & 0.24 & 43.94 + high @xmath59 rass - agn & dr4 + & @xmath45 & @xmath82 log@xmath83 & 562 & @xmath84 & 0.28 & 44.58 + low @xmath59 rass - agn ( rq ) & dr4 + & @xmath45 & @xmath85 log@xmath86 & 883 & @xmath87 & 0.24 & 43.95 + high @xmath59 rass - agn ( rq ) & dr4 + & @xmath45 & @xmath88 log@xmath89 & 454 & @xmath90 & 0.28 & 44.55 + narrow line rass - agn & dr4 + & @xmath45 & @xmath91 log@xmath92 & 187 & @xmath93 & 0.24 & 43.92 + & & & & & & & + total rass - agn & dr4 + & @xmath49 & @xmath94 log@xmath95 & 876 & @xmath96 & 0.43 & 44.64 + total rass - agn ( rq ) & dr4 + & @xmath49 & @xmath97 log@xmath98 & 722 & @xmath99 & 0.43 & 44.61 + + total sdss - agn & dr7 & @xmath42 & @xmath100 & 177 & & 0.13 & -22.52 + total sdss - agn ( rq ) & dr7 & @xmath42 & @xmath101 & 96 & & 0.13 & -22.45 + & & & & & & & + total sdss - agn & dr7 & @xmath45 & @xmath102 & 3500 & & 0.28 & -22.55 + total sdss - agn ( rq ) & dr7 & @xmath45 & @xmath103 & 2879 & & 0.29 & -22.51 + total sdss - agn ( nox ) & dr7 & @xmath45 & @xmath104 & 2367 & & 0.29 & -22.47 + total sdss - agn ( rq+nox ) & dr7 & @xmath45 & @xmath105 & 1958 & & 0.29 & -22.44 + total sdss - agn ( onlyx ) & dr7 & @xmath45 & @xmath106 & 1133 & & 0.27 & -22.72 + low @xmath107 sdss - agn & dr7 & @xmath45 & @xmath108 & 1757 & & 0.28 & -22.18 + high @xmath107 sdss - agn & dr7 & @xmath45 & @xmath109 & 1743 & & 0.29 & -22.93 + low @xmath107 sdss - agn ( rq ) & dr7 & @xmath45 & @xmath108 & 1520 & & 0.28 & -22.18 + high @xmath107 sdss - agn ( rq ) & dr7 & @xmath45 & @xmath110 & 1359 & & 0.29 & -22.88 + & & & & & & & + total sdss - agn & dr7 & @xmath49 & @xmath111 & 4404 & & 0.43 & -23.04 + total sdss - agn ( rq ) & dr7 & @xmath49 & @xmath112 & 3773 & & 0.43 & -23.01 + total sdss - agn ( nox ) & dr7 & @xmath49 & @xmath113 & 3421 & & 0.43 & -22.98 + total sdss - agn ( rq+nox ) & dr7 & @xmath49 & @xmath114 & 2960 & & 0.43 & -22.96 + low @xmath107 sdss - agn & dr7 & @xmath49 & @xmath115 & 2059 & & 0.42 & -22.55 + high @xmath107 sdss - agn & dr7 & @xmath49 & @xmath116 & 2345 & & 0.44 & -23.48 + low @xmath107 sdss - agn ( rq ) & dr7 & @xmath49 & @xmath117 & 1804 & & 0.42 & -22.56 + high @xmath107 sdss - agn ( rq ) & dr7 & @xmath49 & @xmath118 & 1969 & & 0.44 & -23.43 the random sample is another crucial ingredient required for measuring the correlation function . the purpose is to create a randomly - distributed sample of objects that exactly matches all observational biases ( window function , redshift distribution , etc . ) of the observed sample .
we follow the procedure of paper i ( for details see paper i , section 3.1 ) and generate a set of random @xmath119 and @xmath120 values within dr7 areas with a spectroscopic completeness ratio of @xmath370.8 , populate areas with higher completeness ratios more than areas with lower completeness ratios , smooth the observed redshift distribution by applying a least - square ( @xcite ) low - pass filter , and use this smoothed redshift profile to randomly assign redshifts to the objects in the sample .
the number of objects in a random catalog is chosen to have an adequate number of pairs in the ccf at the smallest scales measured here .
for clustering measurements with the main galaxy sample the random catalog contains 100 times as many objects as the observed sample .
the random catalog of the dr4 + sdss main galaxy sample also contains 100 times more objects than the data and follows the same exact procedure described for the dr7 geometry , except that we restrict that area to the dr4 + geometry .
the selection of the sdss lrg sample is described in detail in section 2.1 of paper i. here we briefly summarize the sample selection .
we extract lrgs from the web - based sdss catalog archive server jobs system using the flag galaxy_red , which is based on the selection criteria defined in @xcite .
we verify that the extracted objects meet all lrg selection criteria and create a volume - limited sample with @xmath16 and @xmath121 , where @xmath122 is based on the extinction - corrected @xmath123 magnitude to construct the @xmath124-corrected and passively evolved rest - frame @xmath125 magnitudes at @xmath126 .
we consider only lrgs that fall into the sdss area with a dr7 spectroscopic completeness ratio of @xmath370.8 and have a redshift confidence level of @xmath370.95 .
the correction for the sdss fiber collision in the sdss lrg sample is slightly different from that for the sdss main galaxy sample .
lrgs exhibit very well - defined spectra dominated by an old stellar population that evolves very slowly .
the reduced scatter in the spectral energy distribution ( sed ) of lrgs results in much lower photometric redshift uncertainties than the estimates for main galaxies , which can have ongoing star formation and therefore have a wider distribution of seds .
we make use of the precise lrg photometric redshifts to correct for the sdss fiber collision in the following manner .
we select from the sdss archive all lrgs that pass the pure photometric - based lrg selection criteria .
we identify photometric lrgs that are closer than 55 arcsec to a spectroscopic observed lrg in our redshift and absolute magnitude range .
we then assign a redshift using the following steps : we accept the spectroscopic redshift of the lrg even if its redshift confidence level is @xmath1270.95 . if there is no spectroscopic redshift available for the object , we give the photometric lrg the same redshift as the spectroscopic neighbor lrg ( within a 55 arcsec radius ) if @xmath128 if equation [ eq : redshift ] is not fulfilled , we assume that the photometric redshift of the lrg is the correct redshift .
a redshift is assigned only if the object meets the selection criteria to construct a volume - limited sample : @xmath16 and @xmath121 .
approximately 2% of the all lrgs in our sample are assigned redshifts .
the properties of the sample are shown in table [ samples ] .
the construction of the random catalogs is identical to the procedure described in section [ randomcat ] ( for details see paper i , section 3.1 ) , except that the lrg random catalogs contain 200 times as many objects as the real dr7 and dr4 + lrg samples .
this is a compromise between the required computation time to calculate the correlation functions and having sufficient counts on the smallest scales to avoid introducing noise .
more objects are required at higher redshift for a given sky area to account for the fact that with increasing redshift the same angular distance on the sky corresponds to larger physical co - moving separations . in order to extend our clustering measurements to higher redshifts
, we create an `` extended sdss lrg sample '' over the redshift range of @xmath27 .
the extraction of the sample and the correction for the sdss fiber collision problem is identical to the sdss lrg sample ( see section [ desc_lrg ] ) .
ideally this sample would also be volume - limited .
however , that would require an absolute magnitude cut of @xmath129 mag , which results in only 6292 objects .
as such a relatively low number of objects would yield a measured acf with very large uncertainties , we use a non - volume - limited sample with @xmath130 .
this is a compromise between making the sample volume - limited and retaining accuracy when computing the acf and ccfs ( see table [ samples ] ) .
we plot the absolute magnitude versus redshift for the 28781 objects ( dr7 ) in the extended lrg sample in fig .
[ extendedlrg ] .
the co - moving number density for this sample given in table [ samples ] is computed over the redshift range @xmath131 and magnitude range @xmath130 , where the sample is volume - limited .
furthermore , it assumes that there is no number density evolution at higher redshifts ( @xmath131 ) and in the range @xmath130 . in principle
, we can derive the co - moving number density by integrating the lrg luminosity function .
however , the different selection functions for low and high redshift lrgs ( cut i and ii , @xcite ) are difficult to model and could result in large uncertainties .
therefore , we decide to limit the estimate to the volume - limited redshift range . the random catalogs contain 1000 times as many objects as the data sample .
the procedure follows the description in section [ randomcat ] .
in this paper we derive the agn acf for x - ray and optically - selected sdss agn samples .
the selection of the different agn samples is described below .
the _ rosat _ all - sky survey ( rass , @xcite ) is currently the most sensitive all - sky x - ray survey .
although relatively shallow , this data set has a huge potential for science , especially when combined with other large - area spectroscopic surveys such as sdss .
@xcite positionally cross - correlated rass sources with sdss spectroscopic objects and classified rass and sdss - detected agns based on sdss dr5 .
they find 6224 agns with broad permitted emission lines in excess of of 1000 kms@xmath20 fwhm and 515 narrow permitted emission line agns matching rass sources within 1 arcmin .
more details on the sample selection are given in paper i section 2.2 and @xcite .
the rass / sdss agn sample is biased toward unabsorbed agns due to _ rosat _ s soft energy band ( 0.1 - 2.4 kev ) .
agns with broad emission lines and uv excess are , in general , known to show little to no x - ray absorption and therefore account for the vast majority of the rass / sdss agns . the latest available version of the rass / sdss agn sample is based on dr5 ( @xcite ) .
therefore , we reconfigure the sample to publically - available dr4 + geometry .
we select broad emission line agns in different redshift ranges according to the redshift ranges of the available galaxy tracer sets ( see fig .
[ lx_z ] and table [ samples ] ) . to study the x - ray luminosity dependence of the clustering
, we further subdivide the agns in the redshift ranges of @xmath26 and @xmath16 into relatively lower and higher x - ray luminosity samples .
the rass / sdss agns in the redshift range of @xmath27 are not subdivided into a lower and higher x - ray luminosity sample , as the corresponding tracer set ( extended lrgs ) has a much lower number of objects than the other tracer sets .
consequently , there are too few pairs in the ccf and prohibitively large uncertainties to usefully compare the rass / sdss agn acf of lower and higher x - ray luminosity samples .
we use the 0.12.4 kev observed luminosity reported by @xcite , which assumes a photon index of @xmath132 and is corrected for galactic absorption .
our 0.12.4 kev luminosity cuts are log @xmath133)=43.54 $ ] for @xmath26 and log @xmath133)=44.29 $ ] for @xmath16 ( see fig .
[ lx_z ] ) . using @xmath132 ,
the luminosity cuts correspond to 0.510 kev luminosities of @xmath134)= 1.5 $ ] and 8.5 , respectively . a significant overlap between broad - line rass / sdss agns _ and _ the sdss main galaxy sample tracer set exists only in the lowest redshift range ( @xmath26 ) . for the low luminosity broad - line agns in that redshift range , the agn light does not dominate the optical spectrum .
hence , 186 objects of the total rass / sdss agn sample of 629 objects are also classified as sdss main galaxies ( 52 within the low @xmath59 agn sample , 134 within the high @xmath59 agn sample ) . in the higher redshift samples ,
no match between rass / sdss agn and the lrg tracer sets is found as the selection methods for lrgs and rass / sdss agn are very different and high luminosity agns dominate the optical spectrum . to quantify the effect of the overlap between both samples
, we create a sdss main galaxy sample that does not include objects that have also been classified as rass / sdss agns . compared with the original ccf , the ccf of this new sample shows differences of less than 1% in the pair counts on scales larger than 1.4 @xmath18 mpc .
this is well below the statistical uncertainty of the ccf itself , therefore , the overlap between the agn sample and the tracer set will not significantly affect our clustering measurements in the lowest redshift range .
we calculate the co - moving number densities given in table [ samples ] as described in detail in paper i. for a given @xmath135 and @xmath120 we compute the limiting observable rass count rate and infer the absorption - corrected flux limit versus survey area for rass / sdss agns .
we then compute the co - moving volume available to each object ( @xmath136 ) for being included in the sample ( @xcite ) .
the co - moving number density follows by computing the sum of the available volume over each object @xmath137 .
radio - loud agns are known to be more clustered than radio - quiet agns and reside in very massive dmhs ( e.g. , @xcite ; @xcite ; @xcite ) .
radio - loud agns are also more luminous in the x - rays than radio - quiet agns ( @xcite ; @xcite ) .
in paper i we find an x - ray luminosity dependence in the agn clustering amplitude .
one possible explanation is that the high @xmath59 sample contains more radio - loud agns than the low @xmath59 sample , and therefore the relative overabundance of radio - loud agns in the high @xmath59 sample is causing the increase of the clustering amplitude . to test this hypothesis
, we construct radio - quiet rass / sdss agn samples .
@xcite list in their table of broad - line rass / sdss agn if an object is also detected as a radio source .
the radio information is taken from the faint images of the radio sky at twenty centimeters ( first ; @xcite ; @xcite ) using the nrao very large array .
we therefore create new agn subsamples by restricting all samples to the area covered by first and excluding all first detected sources , and refer to these subsamples as radio - quiet x - ray selected agn samples .
first and sdss cover almost the same area ( first has @xmath27% less area than sdss dr4 + and @xmath25% less area than sdss dr7 ) .
we also limit the tracer sets to the first geometry when computing ccfs of radio - quiet agn samples and the corresponding tracer set acfs . in tables [ samples ] ,
[ table : acf_ccf ] , and [ xagn_acf ] we label these subsamples with the entry ` ( rq ) ' .
note that our approach is conservative , in that we do not apply the usual radio to optical flux density criteria of @xmath138 ( @xcite ) .
instead we remove all radio - detected agns , which removes more objects than just those that are technically defined as radio - loud .
our definition of radio - quiet is that the agns are not detected by first .
however , for our goal of removing all radio - loud agns from the samples , the chosen procedure is adequate and the loss in a few additional agns will not significantly affect the clustering measurements and their uncertainties . in the redshift range of @xmath26 , 127 out of 504 objects
are classified as radio - quiet rass / sdss agns and sdss main galaxies . since we find no significant difference in the ccf of samples that include or exclude these objects , we use the full sample .
no overlap is found between radio - quiet rass / sdss agns and tracer set objects at higher redshifts . in fig .
[ lx_z_zoom_rq ] we show the distribution of the rass / sdss agns which are flagged as radio sources by @xcite , which we remove to create the ` radio - quiet ' samples .
considering only objects that fall in regions covered by sdss and first , 17% of all rass / sdss agn have radio detections in the @xmath42 range , 10% at @xmath45 , and 14% at @xmath49 .
we also report the radio - detected agn fraction in the corresponding lower and higher @xmath59 samples as the percentage values listed in fig .
[ lx_z_zoom_rq ] . in order to test narrow - line versus broad - line agn clustering , we construct x - ray selected narrow - line agn samples .
in addition to the predominant broad - line agns , @xcite classified @xmath27% of all rass / sdss agns as x - ray emitting agns having narrower permitted emission lines .
these 515 objects consist of x - ray emitting agn subclasses such as narrow - line seyfert 1 s ( nls1s ) ( 10% ) , seyfert 1.5 ( 29% ) , seyfert 1.8 ( 18% ) , seyfert 1.9 galaxies ( 22% ) , and seyfert 2 candidates ( 21% ) . in total ,
22% of these objects also have a first radio detection .
nls1s , seyfert 1.5 , 1.8 , and 1.9 agns show a mix of narrow and broad permitted typical agn line components , while seyfert 2 candidates have only narrow permitted emission lines .
the latter are called candidates because it is well known that a fair fraction of them turn out to be reclassified as nls1 , seyfert 1.8 , or seyfert 1.9 when re - observed with significantly improved spectroscopy ( e.g. , @xcite ) .
seyfert 1.5 and 1.8 galaxies have optical spectra with a broad line h@xmath139 component ( exceeding fwhm values of 2500 kms@xmath20 ) at a very low flux level . for more details about the different narrow - line agn subclasses in this sample
see @xcite .
figure [ lx_z_zoom_narrow ] shows that the narrow - line rass / sdss agns are mainly identified only at lower redshifts , to @xmath140 .
therefore , their classification as narrow - line agns is based on the permitted h@xmath139 and h@xmath141 lines .
the studied narrow - line rass / sdss agns are found to have , on average , lower observed x - ray luminosities than broad - line rass / sdss agns . however , the vast majority of broad - line rass / sdss agns are known to be unabsorbed x - ray sources , therefore , their observed x - ray luminosity is equal to their intrinsic x - ray luminosity . on the other hand ,
narrow - line rass / sdss agns may be absorbed agns and their intrinsic x - ray luminosities could be higher than the observed ones given in figure [ lx_z_zoom_narrow ] . consequently , both classes of rass / sdss agn samples may be very similar with respect to intrinsic x - ray luminosity .
due to _ rosat _ s soft energy range of 0.12.4 kev , narrow - line rass / sdss agns can be absorbed only by moderate column densities ( @xmath142 ) .
we simulate an x - ray spectrum with a photon index of @xmath143 , galactic absorption of @xmath144@xmath145 , and a typical redshift of @xmath146 , in order to estimate which intrinsic @xmath142 values are detectable with _
rosat_. compared to an unabsorbed source , the flux in the 0.12.4 kev _ rosat _ band drops down to 61% when an intrinsic column density of @xmath147@xmath145 is used ( 21% for @xmath148@xmath145 ) .
we conclude that only objects with @xmath149@xmath145 are detected in the rass .
most likely the narrow - line rass / sdss agn sample consists mainly of a mixture of unabsorbed and only moderately absorbed ( a few @xmath147@xmath145 ) agns .
although broad - line agns have a much lower projected space density than narrow - line agns and therefore yield , in general , more reliable identification with the rass counterpart , @xcite successfully demonstrate that the narrow - line rass / sdss agn are statistically very reliable identifications as well .
they estimate that less than 5% of the counterparts are spurious random chance positional coincidences . convincing evidence derives from the distribution of positional offsets relative to the x - ray positional error and equal area annuli . furthermore , the observed distribution of the ratios of rass / sdss x - ray to optical flux matches expectations for typical x - ray emitting agns . for details see @xcite .
we conclude that the narrow - line rass / sdss agns have a comparable high counterpart reliability as the broad - line rass / sdss agns
. in the redshift range of @xmath26 31 out of 194 narrow - line rass / sdss agns are also classified as sdss main galaxies , while at @xmath16 only one object belongs to the narrow - line rass / sdss agn sample and the lrg tracer set . as for the broad - line rass / sdss agns
, we create a sdss main galaxy sample that does not include galaxies that have been also classified as narrow - line rass / sdss agns and compute the corresponding ccf . on scales greater than 5 @xmath18 mpc ,
the pair counts differ by less than 0.1% from the original ccf .
therefore , the overlap in the samples does not affect the clustering results for the narrow - line rass / sdss agns .
all of our optically - selected sdss agns ( called ` quasars ' in the sdss literature ) are drawn from @xcite and use the full sdss survey ( dr7 ) .
the agn candidate selection is described in detail in @xcite and can be summarized as follows .
the highest priority is given to first - detected optical point sources .
then sources with non - stellar colors in the @xmath150 photometry data are considered .
objects with photometric redshifts of @xmath151 are targeted even if they are spatially resolved .
first - detected optical point sources and @xmath151 agn candidates are required to have a galactic extinction - corrected @xmath152 magnitude of 19.1 .
this selection method picks @xmath2 18 objects per square degree which are followed up with the sdss spectrograph .
the resulting ` primary ' sdss agn sample is then supplemented by objects targeted by other sdss spectroscopic selections ( main galaxies , lrgs , rass , stars , and serendipitous sources ) that turned out to be agns . in the redshift range of interest ( @xmath0 ) , these secondary channels account for @xmath29% of the total sdss agn sample ( vast majority from the galaxy target selection ) . if a significant number of sdss agns from these secondary channels were originally selected from the lrg sample , which is known to be strongly clustered , it could bias the agn clustering .
we verified that there is no overlap between the sdss agn and the lrg sample .
therefore , we do not expect any influences on our agn clustering measurements caused by the secondary selection channel .
@xcite ( and references within ) have constructed agn catalogs based on different sdss data releases .
they apply a luminosity selection of @xmath153 mag and require that objects have at least one emission line exceeding a fwhm of 1000 kms@xmath20 .
objects that have a spectrum with only narrow permitted agn - typical emission lines are removed . the absolute magnitude @xmath107 is computed by using the @xmath152 psf galactic extinction - corrected magnitude measurement and assuming a typical agn spectral energy slope .
historically , quasars are defined as objects at the high end of the agn luminosity function having @xmath154 mag ( e.g. , @xcite ) .
@xmath155 mag corresponds to @xmath156 mag for an typical agn at @xmath157 .
the schneider et al .
sdss agn catalog papers use the @xmath152-band instead of the @xmath158-band in part because the @xmath152-filter is less affected by galactic absorption . however , a significant disadvantage of this is that the host galaxy light may represent a larger contribution to the total flux then the agn .
agns near the @xmath155 mag cut can be equally bright as the host galaxy , e.g. , host galaxies at @xmath159 with @xmath160 mag ( within the detection limit of the sdss agn selection method ) have @xmath155 mag .
consequently , these agns may be less luminous than their quoted optical magnitude .
this effect is somewhat mitigated by the use of the psf photometric data in the agn selection but should be kept in mind when interpreting our results .
the agn candidates selection by @xcite has undergone constant modification to improve the efficiency .
@xcite use the final selection algorithm .
agns in these catalogs have two spectroscopic target selection flags : best ( final algorithm ) and target ( actual algorithm during targeting ) . best uses the latest photometric software and has the highest quality data .
the continuous modification of the agn selection method and the inclusion of agns not selected by the standard selection means that the agn catalogs are not statistically clean samples .
the catalog of @xcite contains 105,783 spectroscopically confirmed agns which have luminosities of @xmath153 , have at least one emission line exceeding a fwhm of 1000 kms@xmath20 , have highly reliable redshifts , and are fainter than @xmath161 .
we extract only objects that have a mode flag of primary. this procedure applies to 99% of all objects and limits our sample to objects that have been spectroscopically followed up based on a target selection and are not blended .
we use the best flags whenever target and best are available .
we use optically - selected sdss agns in the redshift ranges of the corresponding tracer sets ( @xmath26 , @xmath16 , @xmath27 ) . except for the @xmath26 agn sample , where the number of objects is very low
, we further subdivide the samples into lower and higher @xmath107 subsamples ( see table [ samples ] ) .
the @xmath107 cuts are chosen to yield approximately the same number of objects in the different luminosity subsamples .
we use @xmath162 for the @xmath16 sample and @xmath163 for the @xmath27 sample ( fig .
[ o_sdss_overview ] ) . in the @xmath26 redshift range
we find a high overlap between objects derived from the @xcite agn sample and the sdss main galaxy sample .
133 out of 177 sdss agns are also classified as sdss main galaxies . as with the rass / sdss agns
, we exclude these objects from the sdss main galaxy sample and compute the ccf .
the difference in the ccf occurs only on the larger scales and is less than 1% ; therefore the overlap affects the clustering measurements very little . at higher redshifts
, there is no overlap between optically - selected sdss agn samples and sdss lrgs used as our tracer sets . for the optically - selected sdss agns we do not estimate the co - moving number densities in table [ samples ] .
their selection function is very complex and requires a sophisticated modeling method , which is beyond the scope of this paper . to test the impact of radio - loud agns on the clustering signal of optically - selected agn samples ,
we further exclude first radio - detected objects in the @xcite agn samples .
an agn has a radio detection if its position coincidences with a first catalog entry within 2 arcsec .
we restrict all samples to the first area to ensure that no radio - loud agn is selected for the radio - quiet samples
. then we remove objects that have a measured first radio flux in the agn catalog .
this is the same conservative approach of excluding all first - detected agns as we do in sect .
[ radio_rass ] for the radio - quiet rass / sdss agn subsamples .
here we also use our definition of radio - quiet agn samples for the optically selected agns , namely that radio to optical flux density @xmath164 based on the first is @xmath165 .
we also account for the restriction to the first geometry in the corresponding tracer sets when we derive the acf of the radio - quiet optically - selected agn samples .
the resulting samples are labeled with the subsequent entry ` ( rq ) ' in tables [ samples ] , [ table : acf_ccf ] , and [ xagn_acf ] .
figure [ o_sdss_radio_x - ray ] ( top panel ) shows the distribution of optically - selected first - detected agns within the larger optically - selected agn sample . a fraction of 44% of the @xmath26 agns , 13% of @xmath16 agns , and 9% of the @xmath27 agns
are detected by first ( considering only regions that are covered by first and sdss ) .
the fraction is a sensitive function of redshift , as it depends on luminosity .
the higher @xmath107 subsamples contain more first radio - detections ( 17% in @xmath16 and 12% in @xmath27 ) than the lower @xmath107 subsamples ( 8% in @xmath16 and 5% in @xmath27 ) .
78 out of the 96 radio - quiet sdss agn in the redshift range @xmath26 are also classified as sdss main galaxies .
the large number of x - ray detected optically - selected sdss agns at redshifts above @xmath166 allows us to test how different selections of agns affect the clustering signal of broad - line agns . the catalog paper by @xcite lists for each individual agn the relevant information if the object has a detection in the rass faint or bright source catalog . beside the construction of a radio - quiet optically - selected sdss agn sample ( section [ radio_o_sdss ] )
, we create a sample that contains only optically - selected broad - line sdss agns that are _ not _ detected in rass .
this sample has the subsequent entry ` ( nox ) ' in the corresponding tables .
figure [ o_sdss_radio_x - ray ] ( panel b ) shows the distribution of objects with rass detections among the optically - selected sdss agn sample . a fraction of 70% of the optically - selected sdss agns
are also detected by rass in the redshift range of @xmath26 . at higher
redshifts rass detects 32% ( @xmath16 ) and 22% ( @xmath27 ) of the optically - selected broad - line sdss agns .
this is not surprising , as the low _ rosat _ sensitivity results in the rass redshift distribution rising quickly to @xmath167 and strongly decreasing at higher redshift .
the optical and x - ray luminosities of broad emission line agns are known to be strongly correlated , the ratio of which is often expressed by the optical to x - ray spectral energy index @xmath168 ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
therefore , rass detects 44% of the agns in the higher @xmath107 sample at @xmath16 ( 29% at @xmath27 ) but only 24% of the agns in the lower @xmath107 sample at @xmath16 ( 17% at @xmath27 ) .
rass contains only the highest flux x - ray emitting agns but has the complementary advantage of detecting lower agn activity compared to the optically - selected broad - line sdss agns .
x - ray luminosities of log @xmath133 ) > 42 $ ] are a clear indicator of agn activity , while in the optical a strong starlight component from the host galaxy can make it difficult to detect broad - line agns .
the location of the rass - detected agns in fig .
[ o_sdss_radio_x - ray ] ( panel b ) indicates that rass extends the detections of broad - line agns below the optical cut of @xmath169 mag used by @xcite at @xmath170 .
furthermore , we also create agn samples that contain only optically - selected sdss agn that are also detected as rass sources ( fig .
[ o_sdss_radio_x - ray ] , panel b ) .
the samples are labeled with the subsequent entry ` ( onlyx ) ' . for these object
we have both the x - ray and optical luminosities ( @xmath59 , @xmath107 ) .
in general , as expected by the @xmath168 connection , high / low @xmath107 corresponds to high / low @xmath59 .
the fraction of first - detected agns increases more with @xmath107 than with @xmath59 . finally , we design samples of optically - selected sdss agns that are neither first nor rass detections .
these samples only cover the first area .
this selection results in 56% of the @xmath16 and 67% of the @xmath27 optically - selected sdss agns .
we refer to these samples by the subsequent entry ` ( rq+nox ) ' .
we measure the two - point correlation function @xmath171 ( @xcite ) , which measures the excess probability @xmath172 above a poisson distribution of finding an object in a volume element @xmath173 at a distance @xmath32 from another randomly chosen object .
the auto - correlation function ( acf ) measures the spatial clustering of objects in the same sample , while the cross - correlation function ( ccf ) measures the clustering of objects in two different samples .
we use the same approach as described in detail in paper i ( section 3 ) . here
we reiterate the essential elements of our method .
we use the correlation estimator of @xcite in the form @xmath174 where @xmath175 is the number of data - data pairs with a separation @xmath32 , and @xmath176 is the number data - random pairs ; both pair counts have been normalized by the number density of data and random points .
we measure @xmath177 on a two - dimensional grid of separations @xmath178 , perpendicular to the line of sight , and @xmath179 , along the line of sight , to separate the effects of redshift space distortion due to peculiar velocities along the line of sight .
we obtain the projected correlation function @xmath180 by integrating @xmath181 along the @xmath179 direction . as in paper
i , we infer the agn acf from the ccf between agns and corresponding galaxy tracer set and the acf of the tracer set using @xcite : @xmath182 ^ 2}{w_p(trace|trace)}\,,\end{aligned}\ ] ] where @xmath183 , @xmath184 are the acfs of the agn and the corresponding tracer set , respectively , and @xmath185 is the ccf of the agns with the tracer set . in other words , we assume that the ccf is the geometric mean of two acfs , which has been verified to be valid by @xcite ( appendix a ) .
the ccf is computed by applying eq .
[ dd_dr ] @xmath186 for our purposes , the use of this simple estimator has several major advantages and results in only a marginal loss in the signal - to - noise ratio when compared to more advanced estimators ( e.g. , @xcite ) .
the estimator in eq .
[ dd_dr2 ] requires the generation of a random catalog only for the tracer set .
the tracer sets have well - defined selection functions and are , except for the extended lrg sample , volume - limited .
since the random catalog should exactly match all observational biases to minimize the systematic uncertainties , well understood selection effects are a key to generation proper random samples .
the agn samples suffer from very complex and hard to model selection functions .
therefore , a random catalog of x - ray selected rass / sdss agn is subject to large systematic uncertainties due to the difficulty in accurately modeling the position - dependent sensitivity limit and the variation in the flux limit of the sources ( caused by changing galactic absorption over the sky and spectrum - dependent corrections ) .
optically - selected sdss agns ( see section [ optically_selected_desc ] ) rely on constantly modified selection algorithms and the acceptance of additional incomplete agn selection methods .
consequently , the modeling of their selection function for the generation of a random catalog would be very challenging .
the errors in the adjacent bins in correlation measurements are not independent .
poisson errors will significantly underestimate the uncertainties and should not be used for error calculations .
instead , we use the jackknife resampling technique to estimate the measurement errors as well as the covariance matrix @xmath187 , which reflects the degree to which bin @xmath152 is correlated with bin @xmath188 .
the covariance matrix is used to obtain reliable power law fits to @xmath180 by minimizing the correlated @xmath189 values . in our jackknife resampling
, we divide the survey area into @xmath190 subsection for the dr4 + geometry and 131 subsections for dr7 , each of which is @xmath25060 deg@xmath10 .
these @xmath191 jackknife - resampled correlation functions define the covariance matrix ( eq . [ jackknife ] ) : @xmath192 \ , \end{aligned}\ ] ] we calculate @xmath180 @xmath191 times , where each jackknife sample excludes one section and @xmath193 and @xmath194 are from the @xmath124-th jackknife samples of the agn acf and @xmath195 , @xmath196 are the averages over all of the jackknife samples .
the uncertainties represent a 1@xmath3 ( 68.3% ) confidence interval .
the generation of covariance matrix for the inferred agn acf considers the @xmath191 jackknife - resampled correlation functions of the ccf ( agn and corresponding tracer set ) and the tracer set acf . for each jacknife sample
we calculate the inferred agn acf by using equation [ acf_rassagn ] .
the resulting @xmath191 @xmath180 jackknife - resampled projected correlation functions of the inferred acfs are then used to compute the covariance matrix of the inferred agn acf . to infer the agn acf
, we measure the ccf of the agn sample with the tracer set and the acf of the tracer set . in both cases we measure @xmath178 in a range of 0.0540 @xmath18 mpc in 15 bins in a logarithmic scale .
the upper 11 bins are identical with the bins used in paper i and cover the @xmath178 range of 0.340 @xmath18 mpc .
consequently , we extend the measurements by four additional bins to smaller scales .
we compute @xmath179 in steps of 5 @xmath18 mpc in a range of @xmath197 @xmath18 mpc .
the resulting @xmath198 are shown in fig .
[ contour_acf ] for the acf of the tracer sets and for the ccf of the total agn samples ( x - ray & optically - selected ) with the corresponding tracer sets .
although the projected correlation function is computed by integrating over @xmath179 to infinity ( see paper i , equations 5 , 11 ) , in practice an upper bound of the integration ( @xmath199 ) is used to include most of the correlated pairs , give stable solutions , and suppress the noise introduced by distant , uncorrelated pairs .
we compute @xmath180 for a set of @xmath199 ranging from 10160 @xmath18 mpc in steps of 10 @xmath18 mpc .
we then fit @xmath180 over a @xmath178 range of 0.340 @xmath18 mpc with a fixed @xmath200 and determine the correlation length @xmath201 for the individual @xmath199 measurements . as in paper
i , we find that the lrg acfs ( lrg sample and extended lrg sample ) saturate at @xmath202 @xmath18 mpc .
all ccfs and the main galaxy sample acfs saturate at @xmath203 @xmath18 mpc .
in addition , above these values the corresponding correlation lengths do not change by more than 1@xmath3 , considering the increased uncertainties with increasing @xmath199 values .
the use of @xmath202 @xmath18 mpc for the lrg acfs and @xmath203 @xmath18 mpc for the main galaxy acfs and all ccfs matches the @xmath199 values used for these samples by other studies , e.g. , @xcite ( lrg ) and @xcite ( sdss galaxies ) .
the @xmath180 ccfs for the different total agn samples with the corresponding tracer sets are shown in fig .
[ wp_ccf_agn ] , while the resulting power law fits for the acfs and ccfs based on @xmath204 where @xmath205 is the gamma function , are listed in table [ table : acf_ccf ] .
lccc + main galaxy ( dr4 + ) & @xmath206 & 6.30@xmath207 & 1.85@xmath208 + main galaxy ( dr7 ) & @xmath206 & 6.27@xmath207 & 1.84@xmath208 + lrg ( dr4 + ) & @xmath209 & 9.63@xmath210 & 1.98@xmath208 + lrg ( dr7 ) & @xmath209 & 9.54@xmath211 & 1.95@xmath208 + extended lrg ( dr4 + ) & @xmath212 & 10.90@xmath213 & 1.91@xmath214 + extended lrg ( dr7 ) & @xmath212 & 10.87@xmath215 & 1.89@xmath216 + + total rass - agn & @xmath206 & 5.79@xmath217 & 1.84@xmath218 + total rass - agn(rq ) & @xmath206 & 5.78@xmath219 & 1.86@xmath218 + low @xmath59 rass - agn & @xmath206 & 5.19@xmath220 & 1.96@xmath221 + high @xmath59 rass - agn & @xmath206 & 6.07@xmath222 & 1.79@xmath223 + low @xmath59 rass - agn ( rq)&@xmath206 & 5.49@xmath224 & 1.94@xmath225 + high @xmath59 rass - agn ( rq)&@xmath206 & 6.06@xmath226 & 1.79@xmath223 + narrow line rass - agn & @xmath206 & 4.99@xmath227 & 1.82@xmath228 + & & & + total rass - agn & @xmath209 & 6.88@xmath229 & 1.85@xmath230 + total rass - agn(rq ) & @xmath209 & 7.01@xmath231 & 1.86@xmath218 + low @xmath59 rass - agn & @xmath209 & 6.22@xmath232 & 1.85@xmath233 + high @xmath59 rass - agn & @xmath209 & 7.50@xmath234 & 1.90@xmath235 + low @xmath59 rass - agn ( rq ) & @xmath209 & 6.09@xmath236 & 1.85@xmath233 + high @xmath59 rass - agn ( rq ) & @xmath209 & 7.83@xmath237 & 1.94@xmath238 + narrow line rass - agn & @xmath209 & 5.78@xmath239 & 1.77@xmath210 + & & & + total rass - agn & @xmath212 & 6.82@xmath226 & 1.97@xmath221 + total rass - agn(rq ) & @xmath212 & 6.67@xmath240 & 2.08@xmath241 + + total sdss - agn & @xmath206 & 4.93@xmath242 & 1.98@xmath243 + total sdss - agn(rq ) & @xmath206 & 4.70@xmath244 & 2.11@xmath245 + & & & + total sdss - agn & @xmath209 & 6.91@xmath246 & 1.91@xmath230 + total sdss - agn(rq ) & @xmath209 & 6.98@xmath247 & 1.88@xmath230 + total sdss - agn ( nox ) & @xmath209 & 6.93@xmath248 & 1.93@xmath230 + total sdss - agn ( rq+nox ) & @xmath209 & 6.96@xmath249 & 1.89@xmath250 + total sdss - agn ( onlyx ) & @xmath209 & 7.10@xmath251 & 1.85@xmath218 + low @xmath107 sdss - agn & @xmath209 & 6.91@xmath217 & 1.91@xmath218 + high @xmath107 sdss - agn & @xmath209 & 6.74@xmath252 & 1.90@xmath218 + low @xmath107 sdss - agn ( rq ) & @xmath209 & 6.88@xmath253 & 1.90@xmath218 + high @xmath107 sdss - agn
( rq ) & @xmath209 & 6.91@xmath229 & 1.87@xmath218 + & & & + total sdss - agn & @xmath212 & 7.21@xmath248 & 1.87@xmath230 + total sdss - agn ( rq ) & @xmath212 & 7.24@xmath254 & 1.91@xmath230 + total sdss - agn ( nox ) & @xmath212 & 7.02@xmath217 & 1.81@xmath230 + total sdss - agn ( rq+nox ) & @xmath212 & 7.08@xmath255 & 1.84@xmath230 + low @xmath107 sdss - agn & @xmath212 & 7.12@xmath256 & 1.91@xmath223 + high @xmath107 sdss - agn & @xmath212 & 6.96@xmath257 & 1.81@xmath235 + low @xmath107 sdss - agn ( rq ) & @xmath212 & 7.12@xmath258 & 1.95@xmath223 + high @xmath107 sdss - agn ( rq ) & @xmath212 & 7.16@xmath257 & 1.84@xmath223 our values for the acfs of the tracer sets agree well with measurements from other studies . using a slightly different magnitude cut of @xmath259 in the redshift range @xmath26 ,
@xcite find for the sdss dr2 main galaxy sample a correlation length of @xmath260 @xmath18 mpc and a slope of @xmath261 , fitting over the range @xmath178=0.1320 @xmath18 mpc .
@xcite study the clustering of @xmath230,000 lrgs with @xmath121 and measure @xmath262 @xmath18 mpc and a slope of @xmath263 , fitting over the range @xmath178=0.330 @xmath18 mpc .
their clustering measurement in a sample of high luminosity lrgs ( @xmath264 , @xmath265 ) yields @xmath266 @xmath18 mpc and @xmath267 . in table
[ xagn_acf ] we list the redshift range , the median effective redshift of @xmath268 for the corresponding agn samples , the derived best @xmath201 and @xmath269 values based on power law fits , and @xmath201 for a power law fit with a fixed slope of @xmath270 .
the data are fitted over the range @xmath271 @xmath18 mpc to be consistent with paper i. since we measure the ccf to infer the acf , the resulting effective redshift distribution for the clustering signal is determined by both the redshift distribution of the tracer set and the agn sample : @xmath272 .
the clustering strength is commonly expressed in terms of the rms fluctuations within a sphere with a co - moving radius of 8 @xmath18 mpc ( @xmath273 , see equation 13 in paper i ) .
we derive @xmath273 from the best fit parameters of the power law fits .
the uncertainties on @xmath273 are derived from the @xmath201 versus @xmath269 confidence contours of the one - parameter fit based on a correlated @xmath274 .
using @xmath273 , we further derive the bias parameter @xmath275 based on our power law fits and give these values in column ` @xmath276 pl - fits ' of table [ xagn_acf ] .
the parameter @xmath277 indicates the clustering strength by comparing the observed agn clustering to that of the underlying mass distribution from linear growth theory ( @xcite ) , with @xmath278 , where @xmath279 is the linear growth factor ( section 7.5 in @xcite ) .
we use @xmath280 ( see section [ introduction ] ) .
the bias quantifies the amplification factor of the contrast of the object distribution with respect to that of the dark matter density distribution .
the uncertainties of @xmath277 are derived from the standard deviation of @xmath273 .
column ` @xmath276 hod ' of table [ xagn_acf ] lists the bias parameter derived using halo occupation distribution ( hod ) modeling , which is described below in section [ hod_model ] .
using equation 8 of @xcite and the improved fit for this equation given by @xcite , we compute the expected large - scale eulerian bias factor for different dmh masses at different redshifts . comparing the observed @xmath277 value from hod modeling with the dmh bias factor from @xmath38cdm cosmological simulations provides an estimate of the typical dmh mass ( @xmath281 ) in which the different agn samples reside , listed in the last column of table [ xagn_acf ] .
lcccccccc + total rass - agn & @xmath206 & 0.13 & 4.96@xmath282 & 1.80@xmath283 & 4.84@xmath234 & 1.19@xmath284 & 1.23@xmath285 & 13.22@xmath286 + total rass - agn(rq ) & @xmath206 & 0.14 & 4.42@xmath287 & 1.85@xmath245 & 4.33@xmath288 & 1.08@xmath289 & 1.25@xmath290 & 13.24@xmath291 + low @xmath59 rass - agn & @xmath206 & 0.13 & 3.84@xmath292 & 2.10@xmath293 & 3.85@xmath294 & 0.98@xmath295 & 1.11@xmath296 & 13.03@xmath297 + high @xmath59 rass - agn & @xmath206 & 0.14 & 5.35@xmath298 & 1.66@xmath299 & 4.94@xmath300 & 1.25@xmath301 & 1.33@xmath302 & 13.35@xmath303 + low @xmath59 rass - agn ( rq ) & @xmath206 & 0.12 & 4.18@xmath304 & 2.11@xmath305 & 4.28@xmath306 & 1.07@xmath307 & 1.17@xmath290 & 13.15@xmath308 + high @xmath59 rass - agn ( rq ) & @xmath206 & 0.14 & 4.67@xmath309 & 1.61@xmath310 & 3.86@xmath311 & 1.10@xmath312 & 1.36@xmath313 & 13.38@xmath314 + narrow line rass - agn & @xmath206 & 0.13 & 3.14@xmath315 & 1.64@xmath316 & 3.10@xmath317 & 0.80@xmath318 & 1.24@xmath319 & 13.24@xmath320 + & & & & & & & + total rass - agn & @xmath209 & 0.27 & 4.05@xmath321 & 1.63@xmath207 & 4.07@xmath322 & 1.06@xmath323 & 1.30@xmath324 & 13.17@xmath243 + total rass - agn(rq ) & @xmath209 & 0.26 & 4.50@xmath325 & 1.73@xmath207 & 4.50@xmath326 & 1.16@xmath323 & 1.27@xmath238 & 13.14@xmath299 + low @xmath59 rass - agn & @xmath209 & 0.24 & 3.16@xmath327 & 1.78@xmath328 & 3.18@xmath329 & 0.85@xmath330 & 1.17@xmath291 & 13.01@xmath248 + high @xmath59 rass - agn & @xmath209 & 0.31 & 5.06@xmath331 & 1.82@xmath246 & 5.19@xmath332 & 1.34@xmath333 & 1.48@xmath301 & 13.34@xmath334 + low @xmath59 rass - agn ( rq ) & @xmath209 & 0.24 & 3.62@xmath335 & 1.85@xmath336 & 3.59@xmath337 & 0.95@xmath338 & 1.18@xmath339 & 13.03@xmath340 + high @xmath59 rass - agn ( rq ) & @xmath209 & 0.30 & 6.00@xmath341 & 1.97@xmath342 & 5.90@xmath343 & 1.63@xmath344 & 1.50@xmath345 & 13.37@xmath346 + & & & & & & & + total rass - agn & @xmath212 & 0.42 & 3.24@xmath347 & 1.59@xmath348 & 3.83@xmath294 & 0.96@xmath349 & 1.02@xmath350 & 12.51@xmath351 + total rass - agn(rq ) & @xmath212 & 0.42 & 4.13@xmath352 & 1.98@xmath353 & 4.01@xmath354 & 1.20@xmath355 & 0.99@xmath356 & 12.43@xmath357 + + total sdss - agn & @xmath206 & 0.14 & 3.90@xmath358 & 1.99@xmath359 & 3.94@xmath360 & 0.98@xmath289 & 0.95@xmath361 & 12.67@xmath362 + total sdss - agn(rq ) & @xmath206 & 0.15 & 4.16@xmath363 & 2.12@xmath364 & 4.26@xmath365 & 1.08@xmath366 & 0.94@xmath367 & 12.63@xmath368 + & & & & & & & + total sdss - agn & @xmath209 & 0.31 & 4.80@xmath369 & 1.79@xmath225 & 4.81@xmath370 & 1.26@xmath235 & 1.29@xmath218 & 13.11@xmath235 + total sdss - agn(rq ) & @xmath209 & 0.32 & 4.78@xmath371 & 1.72@xmath283 & 4.77@xmath372 & 1.25@xmath235 & 1.31@xmath373 & 13.13@xmath221 + total sdss - agn ( nox ) & @xmath209 & 0.32 & 4.95@xmath374 & 1.85@xmath283 & 4.94@xmath375 & 1.31@xmath225 & 1.26@xmath373 & 13.06@xmath376 + total sdss - agn ( rq+nox ) & @xmath209 & 0.32 & 4.80@xmath377 & 1.78@xmath243 & 4.72@xmath374 & 1.27@xmath378 & 1.29@xmath379 & 13.10@xmath380 + total sdss - agn ( onlyx ) & @xmath209 & 0.29 & 4.70@xmath381 & 1.71@xmath207 & 4.77@xmath382 & 1.22@xmath301 & 1.35@xmath383 & 13.21@xmath301 + low @xmath107 sdss - agn & @xmath209 & 0.31 & 4.73@xmath384 & 1.72@xmath339 & 4.73@xmath258 & 1.23@xmath385 & 1.31@xmath376 & 13.14@xmath211 + high @xmath107 sdss - agn & @xmath209 & 0.32 & 4.36@xmath386 & 1.75@xmath207 & 4.48@xmath375 & 1.17@xmath387 & 1.24@xmath238 & 13.03@xmath330 + low @xmath107 sdss - agn ( rq ) & @xmath209 & 0.31 & 4.39@xmath388 & 1.65@xmath389 & 4.50@xmath390 & 1.16@xmath391 & 1.30@xmath392 & 13.13@xmath393 + high @xmath107 sdss - agn ( rq ) & @xmath209 & 0.32 & 4.53@xmath394 & 1.70@xmath339 & 4.64@xmath395 & 1.19@xmath391 & 1.30@xmath396 & 13.12@xmath291 + & & & & & & & + total sdss - agn & @xmath212 & 0.42 & 4.41@xmath397 & 1.69@xmath283 & 4.58@xmath398 & 1.23@xmath323 & 1.33@xmath399 & 13.05@xmath392 + total sdss - agn ( rq ) & @xmath212 & 0.42 & 4.66@xmath384 & 1.78@xmath241 & 4.76@xmath258 & 1.30@xmath392 & 1.32@xmath238 & 13.03@xmath241 + total sdss - agn ( nox ) & @xmath212 & 0.42 & 2.93@xmath400 & 1.39@xmath283 & 3.16@xmath401 & 0.92@xmath392 & 1.42@xmath284 & 13.16@xmath323 + total sdss - agn ( rq+nox ) & @xmath212 & 0.42 & 3.79@xmath402 & 1.52@xmath283 & 3.80@xmath403 & 1.09@xmath323 & 1.43@xmath221 & 13.17@xmath241 + low @xmath107 sdss - agn & @xmath212 & 0.40 & 4.59@xmath404 & 1.72@xmath405 & 4.70@xmath406 & 1.26@xmath301 & 1.29@xmath376 & 13.02@xmath210 + high @xmath107 sdss - agn & @xmath212 & 0.43 & 2.92@xmath407 & 1.48@xmath408 & 3.42@xmath409 & 0.90@xmath410 & 1.27@xmath411 & 12.95@xmath412 + low @xmath107 sdss - agn
( rq ) & @xmath212 & 0.40 & 4.68@xmath413 & 1.82@xmath310 & 4.74@xmath414 & 1.31@xmath338 & 1.26@xmath286 & 12.97@xmath319 + high @xmath107 sdss - agn ( rq ) & @xmath212 & 0.44 & 3.59@xmath415 & 1.56@xmath416 & 3.87@xmath417 & 1.05@xmath410 & 1.35@xmath228 & 13.05@xmath210 in this section , we verify the stability of our result against possible observational biases and systematic effects .
as shown in paper i , the somewhat non - contiguous coverage of the sdss dr4 + survey does not influence the clustering results significantly , given the uncertainties .
for dr7 the situations improves as the sdss geometry is much more contiguous .
we verify that the number of random points is large enough to lead to a high number of pair counts at the smallest scales measured and not introduce noise .
@xcite note that the largest structure detected in sdss ( the sloan great wall ) influences the galaxy clustering significantly for samples with @xmath418 and @xmath419 .
therefore , we explore this effect on our sdss main galaxy sample .
we measure the acf for a main galaxy sample that excludes the sloan great wall ( @xmath420 and @xmath421 ) .
we find a correlation length of @xmath422 @xmath18 mpc and @xmath423 , which agrees well with the clustering measurements in which the sloan great wall is included ( see table [ table : acf_ccf ] ) .
we conclude that our main galaxy acfs and the ccfs using the main galaxy sample as a tracer set are not affected by this supercluster at @xmath424 . for several of the agn subsamples split by luminosity
, we have tested that slightly changing the luminosity cuts by up to @xmath425 mags ( both brighter and fainter ) does not significantly change the measured ccfs .
the combination of the different tests listed above provides convincing evidence that our results are not significantly influenced by systematic effects and demonstrates their robustness .
in paper ii , we develop a novel method to infer the halo occupation distribution ( hod ) of rass / sdss agns directly from the well - constrained cross - correlation function of rass / sdss agns with lrgs .
the results from paper ii show that the linear bias parameters and typical dmh masses derived from the best power law fits down to @xmath426 @xmath18 mpc are subject to systematic errors .
this is mainly because the power law fits include scales in the non - linear regime ( @xmath427 @xmath18 mpc ) , where the contribution from pairs of objects that belong to the same dmh ( the one halo term ) is substantial . in this non - linear regime
the bias - dmh mass relation based on linear theory , in principle , should not be applied .
another , less significant source of systematic error is that even in the linear regime ( @xmath428 @xmath18 mpc ) , the underlying matter correlation function deviates from a power law . in order to avoid these issues ,
@xcite derive the bias parameters of their agn samples in the xmm - cosmos survey by modeling their agn acfs at @xmath428 @xmath18 mpc with @xmath429 , where @xmath430 is the agn bias parameter and @xmath431 is the two halo term of the dark matter correlation function modeled as the fourier transform of the linear power spectrum . in this paper , instead , we use the hod modeling developed in paper ii to derive the bias parameter down to @xmath432 @xmath18 mpc , instead of limiting ourselves to @xmath428 @xmath18 mpc .
this allows for better constraints on @xmath433 , especially in cases where the two halo dominated ( linear ) regime extends below @xmath434 @xmath18 mpc , and a better treatment of the cases where the one halo term contribution is still important at @xmath428 @xmath18 mpc .
thus , in our approach , the main constraints are derived from the two halo term , while including the one halo term contribution in the model serves as a first - order perturbation from linear theory .
paper ii discusses the hod modeling of the ccf between rass / sdss agns and lrgs at @xmath16 for three different models : in model a all agns are satellites within the same dmh as the lrgs , while models b and c include different realizations of the cases where central and satellite agns are included and explicitly parameterized .
see paper ii for details of these models .
we repeat this exercise here for the agn samples used in this paper to derive their bias parameters .
the detailed results of the extensive hod modeling will be presented in a separate paper ( miyaji et al . in preparation ) , where a number of new improvements in the modeling over that presented in paper ii will be included . in this paper , we follow exactly the method presented in paper ii . here
we reiterate the main procedure : 1 .
determine the central and satellite hods of the tracer set galaxies ( @xmath435 and @xmath436 , respectively ) from their acf .
the space density constraint is additionally used for volume - limited samples ( the sdss main galaxy and the lrg sample ) .
2 . using the derived tracer set hods ( @xmath437 & @xmath436 ) and using a parameterized model of the agn central and satellite hods ( @xmath438 & @xmath439 ) , we fit the measured ccf between the agn sample and the tracer set to constrain the agn hods
. since the galaxy acfs have much higher statistical accuracy , the uncertainties in the tracer set hods are negligible compared to the agn hod constraints .
3 . derive the bias parameter of the agn sample using : @xmath440 where @xmath441 , @xmath442 is the bias of dmhs with a mass @xmath443 , and @xmath444 is the dmh mass function .
we use equation 8 of @xcite with parameters from @xcite for @xmath442 .
the 1@xmath3 uncertainties of @xmath433 corresponds to the @xmath445 region in the parameter space of the @xmath446 model . as explained above in section [ maingal ] , the tracer set for the low redshift agn samples
is selected from the sdss main galaxy sample with an absolute magnitude range of @xmath35 .
the number density of this sample is @xmath447 @xmath448 mpc@xmath449 .
we use the five - parameter model by @xcite to represent the central and satellite hods of our low redshift tracer set : @xmath450\nonumber\\ \langle n_{\rm g , s}\rangle ( m_{\rm h } ) & = & \langle n_{\rm g , c}\rangle ( m_{\rm h})\ ; \left(\frac{m_{\rm h}-m_0}{m_1^\prime}\right)^{\alpha_{\rm
s}}. \label{eq : zheng07hod}\end{aligned}\ ] ] this form involves a step function with a lower mass cutoff @xmath451 , which is smoothed by incorporating the error function ( erf ) with the width of the cutoff profile @xmath452 . for the detailed description of the different parameters ,
see sect .
3.2 of @xcite .
a limitation of our current fitting software is that it allows a maximum of two simultaneous variable parameters .
thus , we search for acceptable fits in a two parameter space , while fixing other parameters to reasonable values .
we note that , for our current purposes in this paper of obtaining correct @xmath433 values , which is mainly constrained by the two halo term , the detail of the hod is not critical ( see discussions below in sect . [
sec : agn_bias_mdmh ] ) .
a code that searches for the best - fit values and confidence ranges of model parameters in a many - parameter space using the markov - chain monte - carlo ( mcmc ) is under development and will be used in a future paper .
@xcite perform fits to the full five parameter space for several luminosity - threshold subsets of the sdss main galaxy sample .
we take advantage of the results of their @xmath453 sample , which has a similar selection criteria to our sdss main galaxy sample , and use their results to fix three parameters : @xmath454)=11.92 $ ] , @xmath455)=13.94 $ ] , and @xmath456 .
we then search for the best fit model for our measurement of the sdss main galaxy acf in the remaining two parameters ( @xmath457 and @xmath458 ) by minimizing the correlated @xmath189 , taking into account the density constraint ( see eq .
18 of paper ii ) .
we find an excellent fit to the data with the best fit values and 1@xmath3 uncertainties of @xmath459)=12.81\pm 0.01 $ ] and @xmath460 , with an associated bias parameter of @xmath461 .
the uncertainties in these values are lower than those in @xcite as we perform the fit with only two free parameters instead of five . for the fit
, we use @xmath462 measurements in the range @xmath463<40 $ ] .
first , we follow our hod modeling described in paper ii and exclude data points that fall in the transition region between the one halo dominated and two halo dominated regimes . in the case of the sdss main galaxy sample , using scales of 0.4 @xmath464<$ ] 1 results in essentially the identical best - fit hod .
the sdss main galaxy acf and the best fit model are shown in fig .
[ wp_ccf_agn ] ( upper left panel ) .
the tracer set in the intermediate redshift range is the lrg sample with @xmath121 ( sect .
[ desc_lrg ] ) .
the derivation of central and satellite hods of this sample is described in detail in paper ii . in summary , we start from the results of @xcite and make a two parameter adjustment to find the hod that fits best to our lrg acf , including the number density constraint . for satellite lrgs ,
we interpolate between the @xmath465 upper and lower bounds of the satellite hod by @xcite , while for the central lrg hod we shift their central hod horizontally along the @xmath466 axis .
first , we fit @xmath462 measurements in the @xmath467<40 $ ] range .
significant residuals remain in the transition region between the one halo and two halo term dominated regimes , due to the fact that our hod modeling neglect the effects of halo - halo collisions in the two halo term . unlike with the main galaxy sample , this effect is not negligible with the lrg sample , due to a larger transition region .
we therefore neglect data points in a range of @xmath468<2.8 $ ] and find a good fit to the data ( fig .
[ wp_ccf_agn ] , middle left panel ) .
the associated lrg bias parameter is @xmath469 , where this error includes only the statistical 1@xmath3 uncertainty of the fit . unlike for the cases of the sdss main galaxy and lrg samples
, there is no template hod model in the literature for a sample with almost identical selection criteria to those for our extended lrg sample , i.e. , a non - volume limited sample of lrgs with @xmath130 at @xmath27 . the closest sample that we can use as our template is the @xmath264 , @xmath470 lrg sample , for which @xcite made a detailed hod investigation , in addition to the @xmath121 lrg sample .
thus , we follow the approach in paper ii for the lrg sample and search for the best - fit hod model of the acf of our extended lrg sample by adjusting @xcite hod results for the @xmath264 lrg sample . in short , we take the central and satellite hods from fig . 1 ( b ) of @xcite and search for the best - fit hod by tweaking the template .
we shift their central hod horizontally by @xmath471 in @xmath466 and interpolate between their upper and lower bounds ( @xmath472 ) of their satellite hod , with the dividing ratio of @xmath473 , where @xmath474 ( @xmath475 ) represents their lower ( upper ) bound on the satellite hod ( see paper ii for details ) .
we search for the best - fit hod model in the @xmath476 space , where the hod - model predicted @xmath462 function is calculated at @xmath477 . as with the lrgs , we exclude the @xmath478<2.8 $ ] range from the fit .
we obtain the best - fit with @xmath479 and @xmath480 .
we do not use the density constraint in the fit , because the extended lrg sample is not volume - limited and therefore the number density is not accurately determined .
however , the best - fit model gives a density of @xmath481 @xmath448 mpc@xmath449 , which is close to the number density of the extended lrgs calculated in the volume - limited portion of the sample ( see table [ samples ] ) .
for the agn hod model , we use model b of paper ii , which parameterizes the number of central ( @xmath482 ) and satellite ( @xmath483 ) agns in a dmh : @xmath484 where @xmath485 is the step function ( equal to 1 at @xmath486 ; 0 at @xmath487 ) and @xmath488 represents the agn fraction ( duty cycle ) among central galaxies at @xmath489 .
we use @xmath490 in eq .
[ eq : step_pl ] , which @xcite find to be a typical value at which a dmh hosts on average one satellite galaxy in addition to a central galaxy .
we model the ccfs between our tracer sets and the agns using the same method as in paper ii : @xmath491\ , dm_{\rm h } , \label{eq : pag1h}\\ \nonumber\\ p_{\rm ag , 2h}(k ) & \approx & b_{\rm a}b_{\rm g } p_{\rm lin}(k ) , \label{eq : p_ccf_2h}\end{aligned}\ ] ] and @xmath492j_0(kr_{\rm p})dk \label{eq : wp_pk}\ ] ] where @xmath493(@xmath494 ) is the number density of agns ( tracers ) , @xmath495 is the fourier transform of the navarro , frenk , & white profile ( nfw ; @xcite ) , @xmath496 is the linear power spectrum of the density field with the transfer function by @xcite , and @xmath497 is the zeroth - order bessel function of the first kind . due to the low space density of agns ,
most ccfs do not have a signal - to - noise ratio on small scales that is sufficient for applying the @xmath189 statistics .
while a majority of ccfs contain @xmath498 pairs per bin at @xmath499 @xmath18 mpc , we have to exclude @xmath500 @xmath18 mpc bins for several ccfs in order to have at least 16 agn - galaxy pairs per bin .
to derive @xmath433 in a consistent way for all ccfs , we use only @xmath501 @xmath18 mpc bins to derive column ` @xmath276 hod ' for all of the agn samples given in table [ xagn_acf ] .
since the purpose of using the hod modeling in this paper is to derive reliable values of @xmath433 , we do not discuss detailed results using other models , which will presented in miyaji et al .
( in preparation ) .
the derived bias parameter , which is mainly constrained by the two halo term , is not very sensitive to our particular choice of hod model .
to verfiy this , we repeat the hod fits to the ccfs with different values of the parameter @xmath502 ( eq . [ eq : step_pl ] ) . for various sdss luminosity - threshold galaxy samples
, @xcite found the range of this parameter to be @xmath503 .
thus we fix @xmath502 to 1.0 , 1.36 ( our default ) , and 1.6 and obtain @xmath433 values in each case .
the best fit values of @xmath433 typically vary only by @xmath504 among these three cases , demonstrating the robustness of deriving the bias parameter using this method . in fig .
[ xray_optical_b_hod ] we present the main results of our study .
we show the hod bias parameter for our different x - ray and optically - selected agn samples as a function of redshift .
all agn samples are consistent with a host dmh mass of log @xmath1 ) \sim 12.4 - 13.4 $ ] ( see table [ xagn_acf ] ) . samples in which the radio - loud agns have been excluded have very similar clustering amplitudes as the total samples , in all three redshift ranges .
the clustering signal of narrow - line rass / sdss agns at @xmath26 is also very similar to broad - line rass / sdss agns at the same redshift .
furthermore , weak x - ray luminosity dependences on the broad - line agn clustering amplitude are found at both @xmath26 and @xmath16 .
the various agn samples studied here allow us to compare the power law fit derived bias parameters with those from hod modeling . in fig .
[ compare_b ] we plot the bias values from the power law fits and hod modeling listed in table [ xagn_acf ] .
the @xmath277(hod ) values are often larger than the @xmath277(power law fit ) values .
furthermore , the uncertainties on the hod bias are often lower than on the power law bias .
this is because the power law fits are based on the inferred acf , while the hod modeling is directly applied to the ccf which has lower statistical uncertainties . the relatively narrower distribution and on - average higher @xmath277(hod ) values compared to the @xmath277(power law fit ) values
are caused by strong variations between the samples in the one halo term , while the variations in the two halo term are smaller .
as described in the beginning of section [ hod_model ] , power law fit bias measurements commonly use smaller scales that are in the one halo term ( our fitted range is @xmath271 @xmath18 mpc ) in order to increase the statistical significance .
if power law fits are restricted only to larger scales , the method suffers from the problem that the lowest scale where the linear biasing scheme can still be applied varies from sample to sample and remains ambiguous .
hod modeling allows , in principle , the use of the full range of scales since the method first determines the one and two halo terms and then constrains the linear using data down to the smallest @xmath178 values that are dominated by the two halo term for each individual sample .
therefore , an apparently low power law bias value of one sample compared to another can be due to a lower clustering strength in the one halo term , while the `` true '' linear bias parameters may be similar .
if we use scales of @xmath505 @xmath506 for the power law fits , the errors on the bias are much larger , but the values are statistically consistent which those derived from the hod model fits .
in summary , on small ( @xmath507 mpc ) and large ( @xmath508 mpc ) scales a power law model is not a good fit to the data .
hod modeling is currently the optimal method to establish the large - scale bias parameter , provided the adopted hod parameterization can adequately describe the true galaxy and agn dmh population . with larger data sets in the future , the uncertainties in the measurements will be decreased further . to derive a reliable picture of agn clustering
, bias parameters should be inferred from hod modeling , or at least from the comparison of the correlation function with that of the dark matter only in the linear regime @xcite , because systematic errors based on power law bias parameters will be larger than the statistical uncertainties of the clustering measurement . in the following discussions , we use the hod model bias parameters whenever we compare our different agn samples . in the case where we compare our results to other studies , which list only power law fit bias values , we also use power law fit bias values to be consistent .
our clustering measurements of luminous broad - line agns yield three independent data points in the poorly - studied low - redshift range for both x - ray and optically - selected broad - line agns .
in addition , we measure the clustering signal of x - ray detected narrow - line agns .
we derive the bias parameter of the different samples based on power law fits and hod modeling .
there are no published clustering measurements of x - ray selected broad - line agns with comparable low uncertainties at low redshifts .
for example , @xcite measure the clustering strength of broad - line agns in the _ rosat _ nep survey , for which we derive a bias parameter of @xmath509 ( @xmath510 , paper i ) . at
higher redshift @xcite compute the clustering strength of x - ray ( 0.52 kev ) unabsorbed & absorbed and narrow & broad - line agn in the _ xmm - newton _ cosmos field in different bins where the median redshift of the subsamples varies from @xmath511 to @xmath512 .
they find that broad - line agns reside in dmh of log @xmath1 ) \sim 13.2 $ ] , independent of redshift .
the average luminosity of their broad - line agn sample is log ( @xmath513 } ) \sim 44.1 $ ] ( intrinsic absorption - corrected luminosity ; v. allevato , private communication 2011 ) which corresponds to log ( @xmath514 } ) \sim 44.2 $ ] assuming @xmath515 .
consequently , the broad - line agns studied by @xcite have x - ray luminosities comparable to our total rass / sdss agn sample at @xmath16 . the derived dmh mass of log @xmath1 ) \sim 13.2 $ ] for our low - redshift rass / sdss agn sample is in excellent agreement with the value found at @xmath4 for the broad - line agn sample of @xcite .
our study allows us to compare x - ray and optically - selected agn clustering measurements over three independent low redshift ranges .
as the same procedure and identical tracer sets are used for calculating the clustering strengths of the x - ray and optically - selected agn samples , we expect there to be minimal systematic errors when comparing the clustering properties among our different agn samples .
since most of the results in the literature are based on power law fits to the correlation function , we compare our power law bias parameters from various clustering measurements of x - ray and optically - selected sdss agn samples and galaxy clustering measurements in fig .
[ xray_optical_b ] .
the bias values of our x - ray and optically - selected agn populations clearly fall within the region occupied by @xmath2@xmath516 galaxies .
we do not find a very significant difference between the clustering amplitude of our broad - line optical and x - ray agn samples in the different redshift ranges considered here .
the differences in the hod model bias parameters between the total x - ray and optically - selected agn samples in our three ( increasing ) redshift ranges are 1.5@xmath3 , 0.1@xmath3 , and 2.0@xmath3 . as our samples at @xmath16 have many more x - ray and optically - selected agns than the samples at lower and higher redshifts , we consider the results in this redshift range to be the most reliable . we do not find a significant difference in the clustering of x - ray and optically - selected agn in this redshift range .
moreover , given that only one out of three sample has a difference of @xmath22.0@xmath3 , we conclude that it is likely not significant .
although the rass / sdss agn samples extend to lower agn luminosities than the optically - selected sdss agns , both samples span roughly the same luminosity range for broad - line agns , which may account for the similar clustering properties .
rass and sdss are also well matched in terms of the depth and selection of broad - line agns ( see @xcite ) .
optical surveys are known to find at brighter magnitudes predominantly x - ray unabsorbed broad - emission line agns .
_ rosat _ s soft energy range allows primarily for the detection of x - ray unabsorbed agn and , hence , is also biased also toward broad - line agns .
therefore , we do not expect strong systematic differences between these sampes due to the rass / sdss agn selection . for the optically - selected sdss broad - line agns in the redshift ranges of @xmath16 and @xmath27 ,
the sample is large enough to create subsamples of optically - selected agn that are not detected in rass .
furthermore , at @xmath16 we select optically - selected agns that are also detected by rass .
we use these samples as a consistency check to verify that the clustering properties between x - ray and optically - selected agn samples are not significantly different .
as shown in table 3 all of these agn samples agree well within their 1@xmath3 uncertainties in their hod model bias parameters .
this provides compelling evidence that there is indeed no strong difference between broad - line x - ray and optically - selected agns at similar luminosities at low redshifts .
figure [ xray_optical_b ] compares our agn clustering results to other x - ray and optically - selected agn clustering studies .
the properties of various clustering studies are given in table 4 of paper i. @xcite study the clustering properties of agns in the ages survey .
as they publish only the agn ccfs with galaxies , we use their best power law fits ( r. hickox , private communication 2011 ) to infer their agn acf and the bias parameter , following our approach described in sect . [ inferringacf ] . from the redshift distributions presented in @xcite , we compute the effective median redshift for the ccfs ( @xmath517 ) and derive a bias value of @xmath518 .
our finding of detecting no significant difference in the agn clustering properties between x - ray and optically - selected agns at low redshifts may appear to be in contrast to agn clustering measurements at higher redshifts ( @xmath519 ) , where optically - selected agn samples have a lower clustering strength than x - ray selected agn samples ( fig .
[ xray_optical_b ] ) .
furthermore , the clustering of x - ray selected agns is roughly consistent with red galaxies at higher redshifts .
however , some of the x - ray clustering studies significantly underestimate their uncertainties by using poisson errors instead of jackknife errors . moreover , x - ray and optically - selected agn samples at these redshifts select agns with different intrinsic properties .
while the optical agns are mainly drawn from large sky area surveys and sample luminous predominantly broad - line agns , the x - ray selected agn samples derive from very deep observations covering only a few square degrees on the sky .
consequently , the x - ray selected agns have , on average , much lower luminosities . additionally , the x - ray samples include absorbed agns , which results in a large fraction of narrow - line agns that are missed in the optical agn samples at these redshifts .
the differences in the clustering signals between x - ray and optically - selected agn samples at @xmath519 can potentially be accounted for either as the result of a large luminosity difference between the samples or the large fraction of x - ray absorbed , optically narrow - line agns in the x - ray agn samples . to test the latter assumption , we measure the clustering properties of narrow - line rass / sdss agns classified by @xcite in the redshift ranges of @xmath26 and @xmath16 . the low number density of narrow - line rass / sdss agns at @xmath16 forces us to use scales of @xmath520 @xmath18 mpc to apply the @xmath189-statistics during the hod modeling .
the clustering strength of the narrow - line and total broad - line rass / sdss agn samples at @xmath26 agree well with each other ( table [ xagn_acf ] ) , although the uncertainty for the narrow - line agn sample is large . the power law fit bias parameter for the narrow - line rass / sdss agn sample is significantly lower than the bias derived from the hod model ( see fig .
[ xray_optical_b_hod ] ) and would suggest a significantly lower clustering amplitude than the broad - line rass / sdss agn sample .
this is caused by differences between the samples on small scales where the one halo term dominates .
the power law fits use these small scales , while the hod modeling mainly relies on the two halo term to determine the large - scale clustering . at @xmath16 the narrow - line agn hod bias parameter
is too poorly constrained ( @xmath521 ) to allow for a detailed interpretation . while we can not determine whether narrow - line rass / sdss agns cluster similarly or less than broad - line agns in that redshift range , we can rule out the conclusion that they are significantly more clustered .
two important points are worth noting : first , as _ rosat _ can detect only moderately x - ray absorbed agns ( log @xmath522)\la 22 $ ] ) , while _ xmm - newton _ and _ chandra _ are sensitive to much more absorbed ( and lower luminosity ) agns , the narrow - line agns detected by rass may not be as common .
second , the relative clustering strength of narrow - line agns could change with redshift if there is a difference in how agn activity is triggered at different cosmological epochs . however , other studies confirm that low - redshift narrow - line ( radio - quiet ) agns are not strongly clustered and are hosted in galaxies that do not differ significantly from typical non - agn galaxies.(e.g , @xcite , @xcite ) .
@xcite find that x - ray selected narrow - line agns in the _ xmm - newton _ cosmos field cluster slightly lower ( 2.3@xmath3 ) than x - ray selected broad - line agns and reside in dmhs of log @xmath1 ) \sim 13.0 $ ] in the redshift range @xmath523 .
however , their narrow - line agns have an average intrinsic ( absorption corrected ) log ( @xmath513 } ) = 43.1 $ ] ( v. allevato , private communication 2011 ) which is an order of magnitude lower than the average luminosity of their broad - line agns .
when @xcite consider only the x - ray properties of the sources to create x - ray absorbed and x - ray unabsorbed subsamples , in which both have an almost identical mean luminosity of log ( @xmath513 } ) \sim 43.65 $ ] , they find that x - ray absorbed agns cluster slightly less ( 2.6@xmath3 ) than x - ray unabsorbed agns .
to summarize , the difference between the agn clustering properties between x - ray and optically - selected agn samples at @xmath519 is likely not due to a strongly clustered population of narrow - line agns in the x - ray samples . however , these objects do have significantly lower luminosities than optically - selected broad - line agns . for each of the various broad - line agn samples studied here , we create subsamples where we have excluded radio - detected agns to study the impact of radio - loud agns on the derived clustering strength .
we do not find any significant differences between the agn samples that include or exclude radio - detected agns .
the hod bias parameters for these samples agree well within their 1@xmath3 uncertainties in all three redshift ranges studied ( see fig .
[ xray_optical_b_hod ] ) .
however , the samples are similar as only approximately 1020% of all broad - line agns have also radio ( first ) detections , so this result may not be particularly constraining . as mentioned in section [ optically_selected_desc ] , the sdss agn target selection gives the highest priority to point sources that are detected in first ( above a certain flux limit ) without considering their colors .
we select all agns ( @xmath524 ) with @xmath16 which have the sdss first target selection flag equal to 1 .
hence , these objects were only selected on the basis of having a significant first radio flux and can be understood as a well - defined radio - selected agn sample ( @xmath525@xmath526@xmath527=0.28 , @xmath525@xmath107@xmath527=-23.28 ) .
we compute the ccf of this agn sample with the lrg tracer set . due to the low number of agns in this sample ,
the hod bias parameter has large uncertainties , with a value of @xmath528 .
a sample of all 423 sdss agns that have a radio - detection ( including the 187 radio - selected sdss agns ; @xmath525@xmath526@xmath527=0.27 , @xmath525@xmath107@xmath527=-22.84 ) yields @xmath529 . furthermore , we compute the ccf of only the rass / sdss agns with @xmath16 that are marked as radio sources in @xcite . the sample , which contains 144 objects and has @xmath525@xmath526@xmath527=0.25 and log @xmath525@xmath514})$]@xmath527@xmath530 , yields a hod bias parameter of @xmath531 . no constraining results can be drawn by using only radio - detected sdss agn samples , because these samples contain too few objects .
however , all values are consistent with the clustering strengths of the other agn samples in the same redshift range . many previous studies ( e.g. , @xcite ; @xcite ; @xcite ) found that radio - loud agns cluster more strongly than agns without the presence of radio emission .
at first glance our results may appear to be in contrast to these findings . however , those studies focus on the clustering properties of optical narrow - line ( instead of broad - line ) agns based on diagnostic emission - line ratios and radio luminosity . moreover , when comparing the clustering strength of different samples it is essential to take into account the involved luminosities .
figure [ lowz_b ] shows the bias of various agn and galaxy samples at lower redshift ( @xmath532 ) , focusing on comparing radio - selected agn samples . for the radio ( non - broad - line )
agn sample of @xcite , we derive a bias value of @xmath533 ( @xmath534 ) by following the description given in sect .
[ discussion_optical_xray ] .
@xcite find that radio ( first ) galaxies with agn activity cluster more strongly than both radio - detected 2dfgrs galaxies without agn activity and 2dfgrs galaxies without radio emission .
it is not clear , however , whether the different galaxy samples have similar luminosities and colors .
if the galaxy samples that are used for comparison are mainly blue , star - forming galaxies , one would expect a significant difference in the clustering strength based on the different host galaxy properties .
consequently , the clustering difference could reflect not by agn activity but host galaxy type .
the high bias parameter for radio - loud agns ( @xmath535 ) and moderate bias ( @xmath536 ) for x - ray selected agns in @xcite can also be explained by the different host galaxy populations ( see their fig .
@xcite compare the clustering strength of optically - selected narrow - line agns ( @xmath537 and @xmath538 ) with and without radio emission at _ fixed _ stellar mass , i.e. , at fixed luminosity and color .
they find that at fixed stellar mass radio - loud narrow - line agns cluster more strongly and hence reside in more massive dmhs than narrow - line agns without radio emission and galaxies without the presence of agn activity ( see fig .
[ lowz_b ] ) .
neither @xcite nor @xcite detect a difference in the clustering properties within the radio - loud narrow - line agn sample as a function of radio luminosity .
@xcite compute the acf of optical sdss dr5 agns ( from the @xcite sample ) in the redshift range @xmath539 . as we do here , they divide their samples into first - detected and undetected sources
their fits ( over the range 3115 @xmath18 mpc ) suffer large uncertainties ; therefore , they fix the slope of the power law fit for all samples to @xmath540 .
they compute the acf of the first - undetected sdss agns .
when they compare this result to the ccf between first - detected and undetected sdss agns , they find a 2.5@xmath3 difference in @xmath541 , using only the diagonal elements of the covariance matrix .
fixing the slope and not using the full covariance matrix likely increases the systematic uncertainties in their results .
by contrast , our measurements test a much narrower redshift range , use the full covariance matrix , allow free @xmath541 and @xmath269 values for the power law fits , and use the hod modeling approach .
@xcite test much brighter agn samples at higher redshifts than our samples here .
the fact that they find a difference in the clustering of radio - detected versus non - detected agns and we do not may be a consequence of the samples having different luminosities or redshifts . if broad - line agns at @xmath7 indeed have no clustering dependence on radio emission , this may suggest that different physical mechanisms trigger radio emission in broad - line and narrow - line agns , as narrow - line agns do have a clustering dependence with radio emission ( @xcite ) at fixed stellar mass
. this would be somewhat surprising as significant radio emission in all agns is believed to be related to jet phenomena ( e.g. , @xcite ) . in papers
i and ii , we reported a possible x - ray luminosity - dependence in the agn clustering strength of rass / sdss agn in the redshift range @xmath16 ( see also fig .
[ xray_optical_b_hod ] , green and yellow green symbols ) .
the exclusion of radio - detected rass / sdss agns performed here tests whether the weak @xmath59-dependence of agn clustering observed is due to the presence of radio - loud agns in the higher @xmath59 sample ( see sect .
[ radio_rass ] ) . as shown in paper ii
, this luminosity - dependent clustering is more significant in the one halo term than in the two halo term .
using only larger scales of @xmath542 @xmath18 mpc for the hod modeling in this paper ( in order to derive bias parameter in a consistent way for all agn samples ) increases the error on the individual measurements .
this results in a decrease in the significance of the luminosity - dependent clustering .
we find that the 1.5@xmath3 ( 2.0@xmath3 using the power law bias of this paper ) clustering difference between the lower and higher @xmath59 agn samples remains roughly constant when we exclude radio - detect agns from both subsamples ( 1.7@xmath3 using the hod bias ; 2.4@xmath3 using the power law bias of this paper ) .
the difference in the clustering strength of the lower and higher @xmath59 rass / sdss agn samples in the redshift range of @xmath26 is 1.2@xmath3 ( hod model ) .
the error bars on the bias of these samples are high , due to the low number of objects .
the significance of the individual redshift measurements do not provide strong evidence for an x - ray luminosity dependence of the agn clustering strength .
other studies find similar weak trends at low redshifts ( fig .
[ lowz_b ] ) . @xcite compute the clustering signal for _
swift_/bat 1555 kev selected agns and find a clustering difference of 1.6@xmath3 for the higher @xmath59 agns
relative to the lower @xmath59 agns . comparing
the photometric galaxy density around spectroscopic sdss agns , @xcite and @xcite find that higher luminosity agns have more overdense galaxy environments compared to lower luminosity agns at scales smaller than 0.5 mpc .
these studies are not directly measuring agn clustering , rather they focus on the immediate environments of agns .
we do not detect an optical luminosity - dependence of the rass / sdss agn clustering . given the large sample size and low resulting errors in these samples ,
this is a constraining result .
previous clustering measurements of optically - selected agns using the 2df qso redshift survey ( 2qz , @xcite ) , 2df - sdss lrg and qso ( 2slaq , @xcite ) , and sdss also find little evidence for an optical luminosity dependence of the agn clustering strength ( e.g. , @xcite ; @xcite ; @xcite ) . @xcite
use ccfs between agns and lrgs and find some indication that bright sdss agns cluster less than faint 2slaq qsos , although the result is only marginally significant ( 1.6@xmath3 ) . on the other hand ,
@xcite detect a stronger clustering strength for the 10% most luminous sdss dr5 agns ( @xmath543 ) at the @xmath22@xmath3 level .
however , they caution that the dynamical range in luminosity probed is narrow and the sample size in the luminosity subsamples not large enough to yield constraining clustering measurements .
@xcite suggest that a luminosity dependence of the clustering may be more evident at @xmath544 , while @xcite find hints that the lower redshift ranges ( @xmath545 ) may show more dependence with luminosity .
the optical and x - ray luminosities of broad - line agns are connected via the optical to x - ray spectral index , which measures the ratio of the rest - frame luminosity density at 2500 to 2 kev .
although the relation has some scatter ( see @xcite for the optical to x - ray spectral index for rass / sdss agns ) , more x - ray luminous agns are , on average , also intrinsically brighter in the optical .
if indeed there is a weak x - ray luminosity dependence of the agn clustering strength , the very narrow @xmath107 range that the optical sdss agns span would hamper a detection . identifying agn activity using x - ray emission allows us to identify low - luminosity agns , where the optical light is dominated by the host galaxy .
this results in a wider luminosity range in x - ray emission than in the optical ( fig .
[ o_sdss_radio_x - ray ] , lower panel ) in the two lowest redshift ranges . at @xmath45 ,
the mean optical luminosity of the high @xmath107 sdss agn sample is only a factor 2 higher ( 0.75 mag ) than the low @xmath107 sdss agn sample , while the x - ray luminosities between the high and low @xmath59 samples differ by a factor of 4.4 .
this is also seen when considering the luminosity difference between the 90th percentile in the high luminosity samples and the 10th percentile in the low luminosity samples ( factor @xmath546 for optical sdss agns , @xmath547 for rass / sdss agns ) and covered luminosity range ( @xmath548 for optical sdss agns , @xmath549 for rass / sdss agns ) . hence , to detect a possible optical luminosity dependence in the broad - line agn clustering strength ,
as might be expected if there is a weak x - ray luminosity dependence , considering the optical to x - ray luminosity relation , a wider optical luminosity range has to be tested .
as these samples already include the brightest objects , this is only possible if one can include lower optical luminosities where broad - line agns are not effectively selected due to an increase in the host galaxy starlight fraction .
only much deeper surveys at larger redshifts may yield the dynamical range to test the luminosity dependence for optically selected broad - line agns .
the possible x - ray luminosity dependence of broad - line agn clustering detected at low redshifts ( in that more x - ray luminous agns are more clustered than less x - ray luminous agns ) may be difficult to reconcile with the result that , on average , low luminosity ( mainly narrow - line ) x - ray selected agns at higher redshifts are more clustered than luminous optically - selected broad - line agns ( see fig . [ xray_optical_b ] ) . if the x - ray luminosity dependence is real , this may suggest that different physical processes trigger agn activity at different cosmological times or at different luminosities .
although agn clustering measurements are currently not as constraining and the interpretation of the results is not as clear as for galaxy clustering measurements , some general findings have emerged in the last few years . at low redshifts ( @xmath550 ) , broad and narrow - line agn cluster similarly to inactive galaxies , occupying dmh masses of log @xmath1 ) \sim 12.0 - 13.5
this dmh mass range includes cases where the dmh is dominated by one @xmath551@xmath516 galaxy or a small galaxy group composed of multiple such galaxies ( @xcite ) .
independent of the selection method , the clustering strength of broad - line agns does not significantly change , while narrow - line agns show a significant increase in the clustering amplitude when radio - selected narrow - line agns are studied
. finally , more x - ray luminous broad - line agns may cluster more strongly than their lower luminosity counterparts . although the various agn samples have different luminosities , radio - loud , optically - selected narrow - line agns , very luminous x - ray agns , and red galaxies reside in somewhat similar high dmh masses .
lower luminosity x - ray agns , optical narrow - line agns with no radio emission , and blue galaxies tend to be found in lower dmh masses . at high redshifts ( @xmath552 ) , x - ray
selected agn samples appear to cluster more strongly than optically - selected agns .
the reason for this remains unclear .
possibly either low - luminosity or narrow - line ( x - ray absorbed ) agns cluster more strongly than very luminous broad - line optical agns . additionally ,
as some of the x - ray clustering studies significantly underestimate their systematic uncertainties it may turn out that these measurements are consistent with optical agn clustering measurements .
more high - z agn clustering measurements based on larger samples are needed to gain a clearer picture .
in this paper we use agn samples based on sdss . in the very near future
there is not another planned survey that includes photometry and a dedicated extensive spectroscopic follow - up program for agns over such large areas as that covered by sdss . as both large co - moving volumes and spectroscopic redshifts are essential for precise agn clustering measurements , major improvement in our x - ray and optically - selected low redshift agns are therefore not expected in the very near future .
boss ( @xcite ) and bigboss ( @xcite ) will detect high redshift agns at @xmath553 , which will improve agn clustering measurements at higher redshifts . in the coming years ,
the _ rosat _ successor _ erosita _ ( @xcite ) will perform an all - sky survey in the hard and soft x - rays , probing much fainter than rass , which is expected to detect up to @xmath23 million agns .
additionally , the large synoptic survey telescope ( lsst , @xcite ) is expected to identify @xmath22 million agns in optical bands .
_ erosita _ and lsst have the potential to significantly improve agn clustering measurements at low and high redshifts , though only if there are dedicated large spectroscopic follow - up programs .
this work presents agn clustering measurements at low redshifts in three independent redshift ranges : @xmath26 , @xmath16 , and @xmath27 . extending the use of the cross - correlation method of @xcite
, we infer the auto - correlation for both _
rosat _ all - sky survey ( rass ) and optically - selected sdss broad - line agns . as tracer sets we use sdss main galaxies ,
sdss luminous red galaxies , and very luminous red galaxies ( extended lrg sample ) .
we apply the hod model method ( @xcite ) directly to the measured ccfs to derive the bias parameter .
we study the impact of different agn selections on the clustering signal of broad - line agns , i.e. , by excluding radio - detected agns .
furthermore , we compute the clustering strength for rass - selected narrow - line agns .
we find no statistically convincing difference in the clustering of x - ray and optically - selected broad - line sdss agns at low redshifts ( @xmath7 ) .
different agn selections based on either x - ray ( rass ) , optical ( sdss ) , or radio ( first ) , and combinations of these , do not significantly change the clustering signal for broad - line agns .
this appears to be in contrast to other studies that find stronger clustering for radio - loud agns ( e.g. , @xcite ) .
however , those results are based on narrow - line , low luminosity agns , while our sample consists of more luminous broad - line agns . for the x - ray selected broad - line rass / sdss agns
we find hod bias values of 1.23@xmath285 , 1.30@xmath285 , and 1.02@xmath350 in the redshift ranges @xmath26 , @xmath16 , and @xmath27 , respectively , while the hod bias values for the optically selected broad - line sdss agns are 0.95@xmath361 , 1.29@xmath218 , and 1.33@xmath399 , respectively . the corresponding inferred typical dark matter halo masses hosting our broad - line agns are in the range of log @xmath1 ) \sim 12.5 - 13.2 $ ] and are consistent with those occupied by @xmath551@xmath516 galaxies at these redshifts .
we measure the clustering of rass selected narrow - line agns , which consists of a mix of nls1s , seyfert 1.5 , 1.8 , 1.9 , and seyfert 2 candidates .
we do not find a significantly lower clustering amplitude of rass narrow - line agns compared to broad - line agns , although these measurements are subject to large uncertainties .
in addition , we rule out that rass narrow - line agns cluster significantly more strongly than broad - line agns at low redshifts .
we show that the exclusion of radio - detected rass / sdss agns in @xmath16 does not change the weak x - ray luminosity dependence of the agn clustering strength that we find in paper i ( in that higher @xmath59 agns cluster more strongly than lower @xmath59 agns at @xmath554 ) .
we do not detect an optical luminosity dependence of the broad - line agn clustering in the same redshift range , though this result is not particularly constraining due to the narrow @xmath107 range that is covered .
we derive the bias parameter based on the best power law fit , the standard method used in literature , as well as by using hod modeling .
important differences between the two techniques are found for some agn samples . in particular , using a power law fit can underestimate the bias compared to hod modeling .
we show that hod model bias parameters are more reliable and more accurately reflect the large - scale clustering strength .
larger agn samples will be provided by future missions such as _ erosita_. as these samples will have lower statistical uncertainties , hod model bias parameters should be used to avoid introducting systematic errors that could exceed the statistical errors and thus possibly lead to a misinterpretation of clustering measurements .
we would like to thank richard rothschild , alex markowitz , slawomir suchy , and stephen smith for helpful discussions .
furthermore , we thank ryan hickox for providing the inferred acfs of their samples .
last but not least , we thank the referee for a very helpful report .
the _ rosat _ project was supported by the bundesministerium fr bildung und forschung ( bmbf / dlr ) and the max - planck - gesellschaft ( mpg ) . funding for the sloan digital sky survey ( sdss ) has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , princeton university , the united states naval observatory , and the university of washington . | this is the third paper in a series that reports on our investigation of the clustering properties of agns identified in the _ rosat _ all - sky survey ( rass ) and sloan digital sky survey ( sdss ) . in this paper
, we extend the redshift range to @xmath0 and measure the clustering amplitudes of both x - ray and optically - selected sdss broad - line agns with and without radio detections as well as for x - ray selected narrow - line rass / sdss agns .
we measure the clustering amplitude through cross - correlation functions ( ccfs ) with sdss galaxies and derive the bias by applying a halo occupation distribution ( hod ) model directly to the ccfs .
we find no statistically convincing difference in the clustering of x - ray and optically - selected broad - line agns , as well as with samples in which radio - detected agns are excluded .
this is in contrast to low redshift optically - selected narrow - line agns , where radio - loud agns are found in more massive halos than optical agns without a radio - detection .
the typical dark matter halo masses of our broad - line agns are log @xmath1 ) \sim 12.4 - 13.4 $ ] , consistent with the halo mass range of typical non - agn galaxies at low redshifts .
we find no significant difference between the clustering of x - ray selected narrow - line agns and broad - line agns .
we confirm the weak dependence of the clustering strength on agn x - ray luminosity at a @xmath22@xmath3 level .
finally , we summarize the current picture of agn clustering to @xmath4 based on three dimensional clustering measurements . |
in these notes , we mainly deal with the so - called @xmath0- and @xmath1-systems based on the lie algebra @xmath2 . these systems and their generalization to other lie algebras were introduced @xcite @xcite in the combinatorial study of the completeness of the bethe ansatz states for the diagonalization of generalized heisenberg quantum spin chains .
apart from their original representation - theoretic interpretation as recursion relations for kirillov - reshetikhin characters ( @xmath0-systems @xcite ) or @xmath3-characters ( @xmath1-systems @xcite @xcite ) , these equations have reappeared since in various combinatorial contexts @xcite @xcite @xcite .
we use here harmlessly renormalized variables that allow , by introducing sign flips , to connect the systems to cluster algebras with trivial coefficients , and by a slight abuse of language , we will still call the corresponding equations @xmath0- and @xmath1-systems .
let @xmath4 , we define the @xmath2 @xmath0-system : @xmath5 and the @xmath2 t - system : @xmath6 this involves quantities @xmath7 with fixed parity of @xmath8 ( which we take @xmath9 mod 2 ) .
the latter system is readily seen to reduce to the former , upon forgetting " the index @xmath10 .
one way to realize this is to view the solutions of the @xmath0-system as those of the @xmath1-system that are 2-periodic in the index @xmath10 , with the correspondence @xmath11 .
the @xmath1-system may also be viewed @xcite as a non - commutative generalization of the @xmath0-system , in which @xmath12 is no longer scalar , but an invertible operator acting on a hilbert space with basis @xmath13 . in this formulation ,
@xmath7 are the matrix elements of the operator @xmath14 in the basis .
for @xmath15 we define the actions : @xmath16 defining also the shift operator @xmath17 acting as @xmath18 and its formal inverse @xmath19 , the @xmath1-system is readily seen to be equivalent to the non - commutative @xmath2 q - system : @xmath20 in the following , we will also consider the fully non - commutative version of in the case of @xmath21 , namely the recursion relation @xmath22 for variables @xmath23 , a unital algebra , and which corresponds to for @xmath24 but with different boundary conditions @xmath25 , the unit of @xmath26 . to fix entirely their solutions , we must supplement the systems above with a suitable set of initial data .
-systems ( a ) , @xmath1-systems ( b ) and the non - commutative @xmath21 @xmath0-system ( c ) . in all cases we indicate a sample forward mutation . ] for the @xmath0-system , an obvious initial data set consists in fixing the values taken by @xmath27 for some fixed @xmath28 .
indeed , the system of recursion relations is a three - term relation in the variable @xmath29 , hence fixing all @xmath30 for two consecutive values of @xmath29 allows to determine all other @xmath31 s .
moreover the expression for @xmath30 in terms of the @xmath27 is the same as that for @xmath32 in terms of @xmath33 , by translational invariance of the equations .
we call @xmath34 the fundamental initial data set .
the most general admissible initial data sets are indexed by motzkin paths @xmath35 of length @xmath36 , with the property that @xmath37 , and they read @xmath38 ( see fig.[fig : initial ] ( a ) ) . for instance , the fundamental data set @xmath39 is coded by the null motzkin path @xmath40 .
the initial data sets @xmath41 are related via elementary moves , called mutations .
we define the action of the ( forward / backward ) mutation @xmath42 , @xmath43 on a motzkin path @xmath44 as @xmath45 , with @xmath46 , when @xmath47 is also a motzkin path ( see fig,[fig : initial ] ( a ) for an illustration ) .
assume @xmath48 is an admissible set of initial data , then @xmath49 differs from @xmath48 only by a substitution @xmath50 ( forward mutation @xmath51 ) or @xmath52 ( backward mutation @xmath53 ) .
these substitutions are done using the @xmath0-system relation for @xmath54 , which involves only other terms that are common to @xmath48 and @xmath49 .
it is easy to see that iterated mutations allow to connect the fundamental motzkin path @xmath55 to any other motzkin path @xmath44 . for the @xmath1-system , the obvious fundamental initial data set also consists of two layers with consecutive values of @xmath29 , hence @xmath56 , up to translation .
the most general admissible initial data sets are indexed by stepped surfaces " @xmath57 with @xmath58 and the conditions that @xmath59 and @xmath60 ( see fig.[fig : initial ] ( b ) ) .
the data set associated to the stepped surface @xmath61 is @xmath62 .
the fundamental data set @xmath63 corresponds to the surface @xmath64 with @xmath65 mod @xmath66 .
the initial data sets @xmath67 are related via elementary moves , still called mutations .
we define the action of the ( forward / backward ) mutation @xmath68 , @xmath43 , @xmath69 , on a stepped surface @xmath61 as @xmath70 , with @xmath71 , when @xmath72 is also a stepped surface ( see fig.[fig : initial](b ) for an illustration ) .
assume @xmath73 is an admissible set of initial data , then @xmath74 differs from @xmath73 only by a substitution @xmath75 .
these substitutions are done using the t - system relation for @xmath54 , which involves only other terms that are common to @xmath73 and @xmath74 .
it is easy to see that iterated mutations allow to connect the fundamental stepped surface @xmath64 to any other stepped surface @xmath61 . for the non - commutative @xmath21 @xmath0-system ,
admissible initial data have the form @xmath76 , and ( forward / backward ) mutations lead to @xmath77 ( see fig.[fig : initial ] ( c ) ) .
a common feature of all these systems is that they share the so - called positive laurent property , namely that their solutions , when expressed in terms of any of their admissible initial data , are laurent polynomials with non - negative integer coefficients . for the case of commuting variables
this is actually part of a more general structure known as cluster algebras @xcite .
a rank @xmath78 cluster algebra ( without coefficients ) is a dynamical system describing the evolution of data " vectors @xmath79 ( clusters ) of @xmath78 formal variables ( cluster variables ) , attached to the nodes of a cayley tree of degree @xmath78 , with edges cyclically labeled @xmath80 around each node .
this evolution is coded by a companion skew - symmetric @xmath81 matrix @xmath82 with integer entries called the exchange matrix , which evolves as well .
the pairs @xmath83 and @xmath84 at two adjacent nodes
@xmath85 connected by an edge with label @xmath29 are related via the ( involutive ) mutation @xmath86 , defined by @xmath87 with : @xmath88_+ } + \prod_{i=1}^n ( x_t)_i^{[b_{i , k}]_+}\right\ } & \hbox{otherwise}.\end{array}\right.\\ ( b_u)_{i , j}&=&\left\ { \begin{array}{ll } -(b_t)_{i , j } & \hbox{if $ i = k$ or $ j = k$ } ; \\ ( b_t)_{i , j } + { \rm sign}((b_t)_{i , k})[(b_t)_{i , k}(b_t)_{k , j}]_+ ) & \hbox{otherwise}.\end{array}\right.\end{aligned}\ ] ] here we use the notation @xmath89_+=$]max@xmath90 .
the exchange matrices @xmath91 are usually represented pictorially as quivers , namely graphs with oriented possibly multiple edges , with the condition that @xmath92number of edges @xmath93 when @xmath94 .
the finiteness restriction of rank @xmath78 may be removed , while keeping the same definitions , leading to the notion of cluster algebras of infinite rank .
the main property of cluster algebras is the so - called laurent property @xcite@xcite : every cluster variable of any node @xmath95 may be expressed as a laurent polynomial of the cluster variables at any other node @xmath96 .
the positivity conjecture states that this polynomial always has non - negative integer coefficients @xcite , and was only proved in a few cases so far ( e.g. in the so - called acyclic cases @xcite @xcite @xcite , or that of clusters arising from surfaces @xcite ) .
the above @xmath0- and @xmath1-systems have their variables belonging to specific cluster algebras in the following way @xcite@xcite .
the initial data sets turn out to be some particular clusters of a cluster algebra of rank @xmath97 ( @xmath2 @xmath0-system ) or of infinite rank ( @xmath1-system ) . to specify the cluster algebras , we only need to give the exchange matrix @xmath98 at the fundamental node @xmath99 of the cayley tree ,
while @xmath100 ( q - system ) or @xmath101 ( t - system ) is the suitably ordered fundamental set of initial data @xmath102 or @xmath103 defined above . for the @xmath2 @xmath0-system
, we have @xmath104 where @xmath105 is the cartan matrix of @xmath2 , @xmath106 . for the @xmath1-system
, @xmath98 is infinite .
it has the entries : @xmath107 with @xmath108 and @xmath109 .
the aim of these notes is to present a unified approach for proving laurent positivity in all three cases described above , and to extend it to more general cases .
this is essentially based on joint work with r. kedem @xcite @xcite @xcite @xcite .
the strategy consists of three steps . *
step 1 * : construct quantities conserved modulo the evolution equations ( discrete integrals of motion ) , and derive a linear recursion relation for some subset of solutions , from which all solutions can be obtained .
* step 2 * : rephrase the linear recursion relation in terms of a finite continued fraction expression for the corresponding generating function , which is manifestly positive in the initial data . in principle steps 1 and 2 can be repeated for each initial data set , but in practice , it is more efficient to only carry them out for one set of initial data ( say the fundamental one ) , and then use the * step 3 * : express mutations as rearrangements of the initial continued fraction , that preserve the manifest positivity .
note that continued fraction language immediately translates to that of weighted paths on some target graphs .
the laurent positivity results may all be rephrased in terms of partition functions of positively weighted paths on suitable graphs .
the mutations that implement changes of initial data ( rearrangements of the continued fractions ) may be interpreted as transformations of both the target graph and the weights .
this leads to theorem [ positq ] for the @xmath2 q - system ( section [ qsysec ] ) , a restricted version of theorem [ positall ] for the @xmath2 @xmath1-system ( section [ tsysec ] ) and theorem [ ncqpositaone ] for the fully non - commutative @xmath21 @xmath0-system ( section [ ncaonesec ] ) .
this approach is actually restrictive inasmuch as @xmath1-system solutions are concerned .
indeed , it only covers the case of translationally invariant stepped surfaces @xmath61 , with @xmath110 ( but arbitrary attached initial data @xmath111 ) .
the only allowed mutations are actually compound , i.e. infinite iterations of elementary mutations that map such surfaces to each - other . to include the most general case of stepped surfaces
, one must switch to a different point of view , by constructing for each stepped surface a weighted graph or network , from which the solutions can be computed explicitly .
we shall not cover this case in the present notes , and refer to @xcite for details .
the advantage of the present approach however is that it has a natural generalization to fully non - commutative settings , for which such notions as non - commutative continued fraction rearrangements exist .
we illustrate this further in sections [ ncranktwosec ] and [ ncaffinesec ] .
the general case corresponding to non - commutative weighted paths is investigated in section [ ncgensec ] .
* acknowledgments .
* we would like to thank especially r. kedem for a fruitful collaboration .
we also thank s. fomin for hospitality at the dept . of mathematics of the university of michigan , ann arbor , while this work was completed . we received partial support from the anr grant granma , the enigma research training network mrtn - ct-2004 - 5652 , and the esf program misgam .
we start with the @xmath21 @xmath0-system for @xmath24 , with @xmath112 : @xmath113 we have : the solution @xmath114 of the @xmath21 @xmath0-system , when expressed in terms of any admissible initial data @xmath115 , is a laurent polynomial with non - negative integer coefficients . we wish to express the general solution @xmath114 in terms of any admissible initial data @xmath115 .
obvious symmetries of allow to restrict ourselves to @xmath114 for @xmath116 and @xmath34 only .
indeed , introducing the two automorphisms @xmath117 such that @xmath118 and @xmath119 , while @xmath120 and @xmath121 , we immediately see that @xmath122 and @xmath123 for all @xmath124 , respectively from the invariances @xmath125 and @xmath126 of the relation .
if we have an expression @xmath127 , by applying to it @xmath128 we get an expression @xmath129 in terms of any @xmath130 . likewise ,
starting from @xmath127 , @xmath116 , and applying @xmath131 , we get @xmath132 which gives an expression for all @xmath114 , @xmath133 in terms of @xmath34 . moreover ,
if @xmath134 , @xmath116 is a laurent polynomial with non - negative integer coefficients , so are all the transformed expressions under iterations of @xmath135 or @xmath131 .
we now note that eq .
expresses that the discrete wronskian " @xmath136 therefore , we may write @xmath137 hence the columns of the resulting matrix must be proportional , and we write @xmath138 , a constant independent of @xmath78 . we compute : @xmath139 this gives a conserved quantity ( discrete first integral of motion ) for the relation .
it may also be rephrased into a linear recursion relation for the @xmath114 s : @xmath140 introducing the generating function @xmath141 , and defining the weights @xmath142 , @xmath143 and @xmath144 , we may use to get : @xmath145 the latter finite continued fraction expression has manifestly positive polynomial coefficients of @xmath146 in its series expansion in @xmath96 , and the theorem follows from the fact that @xmath146 are laurent monomials in the initial data @xmath34 .
the continued fraction expression above may be interpreted as the generating function for paths on the integer segment @xmath147 $ ] , with steps @xmath148 , starting and ending at @xmath99 and with step weights @xmath149 for steps @xmath150 , @xmath151 and @xmath152 for steps @xmath153 , @xmath151 , each path being weighted by the product of its step weights .
the two possible forward mutations of the cluster algebra associated to read : @xmath154 and @xmath155 , where the updates of initial data are made by use of the relation .
let us express explicitly the effect of the mutation @xmath156 on the continued fraction expression , by noting the two following rearrangement lemmas , both easily proved by direct calculation : [ reartone ] @xmath157 [ reartwo ] @xmath158 provided @xmath159 we write : @xmath160 where we have first applied lemma [ reartone ] and then lemma [ reartwo ] , with @xmath161 and : @xmath162 we conclude that @xmath163 is precisely the generating function @xmath164 where the @xmath114 s are expressed in terms of @xmath165 .
note also that @xmath166 , upon defining @xmath167 .
more generally , repeated application of the rearrangement lemmas [ reartone ] and [ reartwo ] leads to iterated mutations .
we now have all the ingredients necessary for the generalization to the @xmath2 @xmath0-system : integrals of motion , continued fractions , rearrangements .
the purpose of this section is to outline the proof the following : [ positq ] the solution @xmath168 of the @xmath2 @xmath0-system is a positive laurent polynomial of any admissible initial data @xmath41 , for any motzkin path @xmath44 of length @xmath36 . as before , the symmetries of the system allow to restrict ourselves to @xmath168 with @xmath116 and to the motzkin paths in a fundamental domain @xmath169 modulo global translation of all indices by @xmath149 .
the @xmath2 @xmath0-system possesses @xmath170 conserved quantities generalizing .
these are obtained by repeating the argument of the @xmath24 case . introducing the discrete wronskian @xmath171 we have the relation @xmath172 as a consequence of the desnanot - jacobi relation between the minors of any square matrix , also used by dodgson for his famous condensation algorithm @xcite .
moreover , we have @xmath173 and @xmath174 by definition .
we conclude that @xmath175 , the solution of the @xmath2 @xmath0-system . as a consequence
, we have the second ( @xmath2-)boundary condition : @xmath176 .
writing @xmath177 , we deduce the existence of a linear recursion relation @xmath178 with @xmath179 and where @xmath180 are @xmath170 conserved quantities modulo . in ref.@xcite , these were expressed explicitly in terms of the fundamental initial data @xmath102 corresponding to the null motzkin path @xmath55 , as partition functions for hard particles on some target graph .
introducing the generating function @xmath181 , the result of @xcite may be rephrased into the following continued fraction expression : @xmath182 where the weights @xmath183 are the following explicit laurent monomials of the initial data @xmath102 : @xmath184 [ rempath ] the continued fraction expression has an immediate path interpretation .
we consider the paths on the integer segment @xmath185 $ ] , from and to @xmath99 , with steps @xmath186 with respective weights @xmath149 for steps @xmath150 , @xmath187 for steps @xmath188 ( @xmath189 ) , @xmath190 for steps @xmath191 ( @xmath192 ) , and @xmath193 for the step @xmath194 , @xmath195 for the step @xmath196 .
then @xmath197 is the sum over all such paths of the product of their step weights . alternatively ,
this was interpreted in @xcite as the partition function for weighted paths on a graph @xmath198 with @xmath199 vertices labeled @xmath200 and adjacency matrix @xmath201 with non - vanishing elements @xmath202 ( @xmath203 ) and @xmath204 ( @xmath205 ) .
that allow to reach by iteration any motzkin path , up to global translation , from the null path @xmath55 . ]
the expression has manifestly positive polynomial coefficients of the @xmath206 s in its series expansion in @xmath96 , and the theorem [ positq ] is proved for @xmath207 and the initial data @xmath102 . to reach all the motzkin paths in the fundamental domain @xmath169
, we proceed by induction under ( forward ) mutation .
we note that all the motzkin paths in @xmath169 may be obtained from @xmath55 by iterated application of mutations @xmath208 , restricted only to the two following situations depicted in fig.[fig : twocases ] ( where by convention we must simply omit @xmath209 and @xmath210 ) : * * case ( i ) : * @xmath211 , then @xmath212 . * * case ( ii):*@xmath213 , then @xmath214 . in ref.@xcite
, it was found that the mutations may be implemented iteratively on the continued fraction @xmath215 , to reach any @xmath216 , @xmath217 , by application of the lemmas [ reartone ] and [ reartwo ] , according to the case @xmath208 at hand : * @xmath218 : apply lemma [ reartwo ] to @xmath219 . *
@xmath220 : apply first lemma [ reartone ] to obtain @xmath221 , and then lemma [ reartwo ] to @xmath222 .
each continued fraction thus obtained is manifestly positive , as the two rearrangements are manifestly positive .
note that under the rearrangement lemma [ reartwo ] the _ structure _ of the continued fraction changes , and in general we obtain multiply branching continued fractions .
remarkably however , the continued fraction @xmath223 may be expressed entirely in terms of a collection of weights @xmath224 called _
skeleton weights_. alternatively , the continued fraction expressions allow to interpret the ratio @xmath225 as the partition function for weighted paths on a rooted target graph @xmath226 , from and to the root , and with exactly @xmath78 _ descents _ , i.e. steps taken towards the root .
a compactified version of these graphs was found recently in @xcite , and allows to put all the continued fractions @xmath227 in the form of jacobi - type continued fractions , namely of the form @xmath228 and we have @xmath229 where for @xmath230 : @xmath231 we see that the price to pay for having a nice jacobi - type continued fraction structure for @xmath227 is that it is no longer manifestly positive , due to the subtractions in the definition of @xmath232 .
however , it is a straightforward exercise to rearrange into a manifestly positive , possibly multiply branching continued fraction , in which all the coefficients are laurent monomials of the @xmath233 s with coefficient @xmath96 ( we call this form _ canonical _ ) .
we simply apply iteratively from the bottom to the top of the jacobi - like continued fraction the lemma [ reartone ] in the form @xmath234 or its converse " : [ rearthree ] @xmath235 note that multiple branching will occur when applying lemma [ rearthree ] because the remaining part @xmath95 of the continued fraction is now duplicated .
moreover , the iteration of the procedure will produce new weights , which were called redundant weights " in @xcite , that are all laurent monomials of the skeleton weights .
note finally that we must apply at level zero " to @xmath236 itself , with @xmath237 , @xmath238 , @xmath239 ( i.e. lemma [ reartone ] ) .
the manifestly positive canonical form of @xmath236 can be directly interpreted @xcite as the partition function of paths on a rooted target graph @xmath226 , with step weights that are either trivial ( with value @xmath149 ) or laurent monomials of the @xmath233 s with coefficient @xmath96 .
the fraction corresponding to the motzkin path @xmath240 for @xmath241 reads : @xmath242 the first expression is the jacobi - like form .
the second expression is the canonical form , obtained by applying lemma [ rearthree ] first on the third level of the continued fraction , and then on the second , and by finally applying the lemma [ reartone ] .
it is manifestly positive , and introduces redundant " weights @xmath243 , @xmath244 , @xmath245 , all of which are laurent monomials of the skeleton weights @xmath246 with coefficient @xmath96 .
the fraction corresponding to the motzkin path @xmath247 for @xmath241 reads : @xmath248 we have applied the lemma [ rearthree ] to the third stage of the initial continued fraction , and then lemma [ reartone ] to its second and first stages .
we get a redundant weight @xmath245 .
finally , we get the following explicit expression for the skeleton weights in terms of the initial data @xmath41 , for any @xmath217 : the skeleton weights associated to the motzkin path @xmath249 read : @xmath250 this is proved by induction on mutations . first
, the formulas ( [ oddync]-[evenync ] ) hold for the fundamental case @xmath251 ( see ) .
assume they hold for some motzkin path @xmath44 .
as before , we only have to examine the cases ( i ) and ( ii ) . as a consequence of lemma [ reartwo ] , the mutation @xmath252 acts on the skeleton weights as follows : @xmath253 and @xmath254 in all the other cases . finally , it is straightforward to check that these relations are satisfied by the weights ( [ oddync]-[evenync ] ) , as a consequence of the @xmath0-system .
the theorem follows .
let us now consider @xmath236 in its manifestly positive canonical form , with possible redundant weights , which are all monomials of the skeleton weights , with coefficient @xmath96 .
the series expansion in @xmath96 has coefficients that are polynomials of the skeleton and redundant weights , hence @xmath255 is itself a positive laurent polynomial of the initial data @xmath41 .
this completes the proof of theorem [ positq ] for @xmath220 .
the case of @xmath218 follows from a slight generalization of the lindstrm - gessel - viennot theorem @xcite@xcite , which allows to interpret the discrete wronskian expression for @xmath256 as the partition function for families of strongly non - intersecting paths on the same target graph @xmath226 , with the same step weights ( see ref.@xcite for details ) .
the theorem [ positq ] follows .
in this section , we present the solution of the @xmath2 @xmath1-system when the boundary stepped surface is periodic .
it appears to be a non - commutative generalization of the @xmath0-system case , and to have the laurent positivity property .
we actually have the following more general result , proved in @xcite : [ positall ] for any stepped surface @xmath257 , the solution @xmath7 of the @xmath2 @xmath1-system is a positive laurent polynomial of the initial data @xmath67 . in this section ,
we restrict ourselves to initial data @xmath67 , with periodic stepped surfaces @xmath61 with @xmath110 for all @xmath258 .
an example is provided by the fundamental stepped surface @xmath64 .
these periodic stepped surfaces are related via compound mutations of the form @xmath259 , namely infinite products of elementary mutations , which are implemented by the use of the @xmath1-system relations for a fixed @xmath54 but for all @xmath260 simultaneously , in order to simultaneously substitute @xmath261 for all @xmath260 .
the periodicity property allows us to work within the framework of the non - commutative @xmath0-system formulation of the @xmath1-system ( [ actrt]-[ncqsysone ] ) .
the compound mutation @xmath42 indeed amounts to a substitution @xmath262 by use of the relation .
note that the operators @xmath14 are only mildly " non - commuting , as all the operators of the form @xmath263 act diagonally on the basis @xmath264 , hence they all commute , for all @xmath265 . in ref .
@xcite , the constructions of the @xmath0-system case are generalized to the case of the @xmath1-system ( conserved quantities , linear recursion relations , partition functions for hard objects , and finally path formulation ) . in particular , the desnanot - jacobi identity may still be applied to eliminate the @xmath7 for @xmath218 in terms of the @xmath266 s via the following discrete wronskian " formulas : @xmath267 we may reformulate the final result for the solution @xmath266 of the @xmath2 @xmath1-system as a function of the fundamental initial data @xmath103 as follows . the generating function @xmath268 has a continued fraction form analog to .
define the operators @xmath269 , @xmath270 by : @xmath271 then we have : @xmath272 where @xmath273 denotes the identity operator @xmath274 . picking the coefficient of @xmath275 in @xmath276 ,
this may be interpreted as a direct generalization of the @xmath0-system result : @xmath277 is the partition function for paths on the same graph @xmath198 as in the commuting case ( see remark [ rempath ] ) , but now with time "-
dependent weights @xmath278 associated to the operators @xmath279 , such that the paths start at time @xmath280 and end at time @xmath281 , and that the up " steps @xmath282 take no time ( i.e. do not increase the time counter ) and have weight @xmath149 , while the down steps @xmath283 and level steps @xmath99 take two units of time ( i.e. increase by @xmath66 the time counter ) and receive the weights @xmath284 , where @xmath96 is the value of the time counter after the step is completed .
this shows explicit laurent positivity for @xmath266 as a function of @xmath103 .
( to make contact with @xcite , we have to introduce conjugated weights @xmath285 - 1}{{\mathbf y}}_{{\alpha}}{{\mathbf d}}^{[{{{\alpha}}\over 2}]+1}$ ] for @xmath286 and @xmath287 , with the effect of distributing the time more evenly on the steps , namely steps @xmath288 take one unit of time , while steps @xmath99 take two . )
finally , let us show that , like in the commutative @xmath0-system case , we may implement mutations by non - commutative versions of the rearrangement lemmas [ reartone ] and [ reartwo ] .
we have the following two non - commutative rearrangement of non - commutative fractions , which preserve manifest positivity .
( in the following , and for later generalizations , we assume the continued fraction coefficients are elements of an algebra @xmath26 with unit @xmath289 ; moreover , any expression of the form @xmath290 may be formally expanded as @xmath291 . )
[ ncreartone ] let @xmath292 , we have : @xmath293 [ ncreartwo ] for @xmath294 , @xmath295 invertible , we have : @xmath296 provided @xmath297 we note that the periodic stepped surfaces @xmath61 may be indexed by motzkin paths of length @xmath36 .
indeed , the map @xmath298 defined by @xmath299min@xmath300 is a bijection between the set of periodic stepped surfaces and that of motzkin paths of length @xmath36 . by a slight abuse of notation
, we will denote by @xmath301 the function @xmath302 with @xmath303 .
starting from the operator continued fraction @xmath304 , we may now apply iteratively the above rearrangement lemmas [ ncreartone ] and [ ncreartwo ] , exactly in the same way as we applied lemmas [ reartone ] and [ reartwo ] for the commutative @xmath0-system .
the results are operator continued fractions @xmath305 , with operator coefficients , whose matrix elements in the basis @xmath264 are laurent monomials of the initial data @xmath67 with coefficient @xmath96 .
as the rearrangements preserve positivity manifestly , we deduce that the theorem [ positall ] holds for @xmath220 and all periodic stepped surfaces @xmath61 .
alternatively , the rearranged continued fractions @xmath301 generate time - dependent weighted paths on the _ same _ target graphs @xmath226 as in the commuting @xmath0-system case . for @xmath218 , we interpret the discrete wronskian formulas via the same generalization of the lindstrm - gessel - viennot theorem , as computing the partition function for families of strongly non - intersecting paths on @xmath226 , now with time - dependent weights .
the positivity and therefore theorem [ positall ] follows , for the case where the stepped surface @xmath61 is 2-periodic in the @xmath10 direction .
we have already seen how to interpret the @xmath1-system as a non - commutative @xmath0-system , in terms of variables obeying many special commutation relations . in this section
we investigate integrable relations involving fully non - commutative variables .
to prove the theorem , we must somehow repeat the steps of section [ warmupsec ] . as in the commuting case ,
the symmetries of the equation allow us to restrict to the case @xmath116 and to the initial data @xmath34 .
we still define anti - automorphisms @xmath307 that leave @xmath289 invariant , and such that @xmath308 , @xmath309 , and @xmath310 , @xmath311 , while @xmath312 and @xmath313 .
we deduce that @xmath314 and @xmath315 for all @xmath316 .
assuming we have a positive laurent polynomial expression for @xmath317 for all @xmath116 , we may obtain a positive laurent expression for @xmath318 as well by applying @xmath131 . finally from a positive laurent polynomial expression for @xmath317 for all @xmath124
, we may obtain one for @xmath319 for any @xmath320 , by applying @xmath128 .
introducing @xmath322 , and noting that @xmath323 commutes with the r.h.s .
of , we have @xmath324 , hence @xmath325 .
note that the conservation of @xmath326 implies a quasi - commutation relation : @xmath327 and that the @xmath21 @xmath0-system may be rewritten as : @xmath328 introducing @xmath329 and @xmath330 , we find that @xmath331 by use of the quasi - commutation , hence @xmath332 .
moreover , computing @xmath333 we deduce that @xmath334 , and the lemma follows . for @xmath96 a central element in @xmath26 , introducing the formal generating function @xmath338 , we easily compute : @xmath339 this may be expanded as a series in @xmath96 , whose coefficients are polynomials of @xmath340 with non - negative integer coefficients .
the theorem [ ncqpositaone ] follows from the definition of @xmath197 and the explicit form of the @xmath341 s .
finally , let us show that , like in the @xmath0 and @xmath1-system cases , we may implement mutations by use of the non - commutative rearrangement lemmas [ ncreartone ] and [ ncreartwo ] .
we write : @xmath342 where we have first applied lemma [ ncreartone ] and then lemma [ ncreartwo ] , with @xmath343 and : @xmath344 where we have used the @xmath0-system relation ( first line ) and the quasi - commutation relation ( second and third ) .
we conclude that @xmath345 is precisely the generating function @xmath346 where the @xmath323 s are now expressed in terms of @xmath165 . more generally , repeated application of the non - commutative rearrangement lemmas [ ncreartone ] and [ ncreartwo ] leads to iterated mutations .
the rank two affine cluster algebras have a fundamental skew - symmetrizable exchange matrix @xmath347 , with @xmath348 or @xmath349 .
the @xmath21 @xmath0-system of previous section reduces to the case @xmath350 when @xmath351 .
the two other cases also have non - commutative counterparts , namely the system @xmath352 and that obtained by interchanging even and odd indices . repeating the approach of previous section
, we note that the symmetries of the system allow to restrict to the case of @xmath116 and to the initial data @xmath34 and @xmath353 .
next , we identify two conserved quantities : @xmath354 we finally arrive at the following continued fraction expression for the generating function @xmath355 in terms of @xmath34 : @xmath356 where @xmath357 this proves positivity of all @xmath358 in terms of @xmath34 .
the case of @xmath353 is obtained by mutation / rearrangement of @xmath359 .
we write : @xmath360 where @xmath361
we get a series in @xmath96 with coefficients that are polynomial in @xmath362 , and as each of them is itself a positive laurent polynomial of the initial data we deduce positivity for all @xmath358 in terms of @xmath353 .
the case of the variables @xmath363 is easy , by directly checking that @xmath364 contains the term @xmath289 .
this completes the proof of theorem [ posionefour ] .
we now consider the systems @xmath366 with initial data @xmath367 for all @xmath320 .
the commutative version of ( [ rk3one]-[rk3two ] ) corresponds to mutations within the affine @xmath365 cluster algebra .
this is proved in the usual way .
we may restrict ourselves to @xmath116 and to the initial data @xmath34 .
we find the conserved quantity @xmath370 this allows to derive continued fraction expressions for @xmath371 and @xmath372 , @xmath373 : @xmath374 where @xmath375 we have manifest positivity in the @xmath376 s .
moreover , the @xmath341 s are defined recursively by @xmath377 and @xmath378 , @xmath379 , hence they form subsums of @xmath380 , and so do the @xmath381 s . finally all @xmath341 s contain the term @xmath382 , while all @xmath381 s contain @xmath383 , hence both @xmath384 and @xmath385 contain @xmath289 : we conclude that the @xmath376 s are all positive laurent polynomials of @xmath34 .
this completes the proof of theorem [ positrkthree ] . in view of the accumulated evidence for path models and continued fractions to play an important role in the discrete non - commutative integrable systems
, we now address the general case of paths on the graphs @xmath226 of the @xmath2 @xmath0-system solution , but with fully non - commutative weights . for any @xmath44 motzkin path of length @xmath36
, we define the family of non - commutative continued fractions @xmath301 as follows .
let @xmath386 be a given collection of non - commutative skeleton " weights .
we define the non - commutative jacobi - type fraction @xmath387 for a family of non - commutative weights @xmath388 by : @xmath389 we set @xmath390 where for @xmath230 : @xmath391 like in the commutative case , we may easily bring the continued fraction @xmath301 to a manifestly positive form where all non - trivial coefficients are laurent monomials of the @xmath341 s with coefficient @xmath96 ( we still call this the _ canonical _ form ) , and for which the coefficients of the formal series expansion in @xmath96 are therefore laurent polynomials of the @xmath341 s with non - negative integer coefficients .
this is readily performed by applying iteratively from the bottom to the top of the jacobi - type continued fraction the non - commutative rearrangement lemma [ ncreartone ] , in the form @xmath392 or its converse " : the net effect is to undo " the @xmath396 s , at the expense of creating branchings and new redundant " weights that are laurent monomials of the skeleton weights , all with coefficient @xmath96 , and therefore all manifestly positive . the manifestly positive canonical form of @xmath301 can now be directly interpreted as the partition function of non - commutative weighted paths on the _ same _ rooted target graph @xmath226 as for the commuting case .
the continued fractions @xmath301 are related via non - commutative mutations as follows . restricting like in the commutative case to the cases ( i ) and ( ii ) for @xmath44 and @xmath397
, we see that the forward mutation @xmath51 is implemented on @xmath301 written in the jacobi form via the following sequences of rearrangements : * * case ( i):*(a ) undo @xmath398 by applying lemma [ ncreartone ] ( b ) apply lemma [ ncreartwo ] to the piece containing @xmath399 and @xmath400 in factor ( c ) go back to the jacobi form by using in reverse * * case ( ii):*(a ) apply lemma [ ncreartwo ] to the piece containing @xmath401 ( b ) go back to the jacobi form by using and lemma [ ncrearthree ] in reverse this produces @xmath402 , in jacobi form as well .
note that when @xmath220 we have to apply at the level zero of the continued fraction , with @xmath403 , @xmath404 and @xmath405 .
so the transformation of the continued fraction @xmath301 reads : @xmath406 it is easy to follow the transformations of skeleton weights in the process .
the mutation @xmath252 acts on the skeleton weights as follows : @xmath407 while all other skeleton weights are left unchanged . for any square @xmath410 matrix @xmath411 with entries in @xmath26 , and any @xmath412
, we define the @xmath413-quasi - determinant @xmath414 as : @xmath415 whenever the r.h.s .
is well - defined .
the quasi - determinant has many interesting properties ( see @xcite for a detailed study ) . in the commutative case ,
it reduces to a ratio of determinants , namely @xmath416 , where @xmath417 stands for the matrix @xmath201 with row @xmath418 and column @xmath3 erased .
we now define the quasi - wronskians of the sequence @xmath419 to be : @xmath420 with the convention @xmath421 .
these quasi - wronskians satisfy a generalized desnanot - jacobi relation ( also called the non - commutative hirota equation @xcite ) : @xmath422 the @xmath2 boundary condition is replaced by @xmath423 expressing simply that the continued fractions are actually finite ( @xmath424 ) , or in other words that there exists a vanishing left linear combination of the @xmath323 s with @xmath425 constant coefficients ( the integrals of motion ) , causing the quasi - determinant to vanish ( this linear recursion relation for @xmath323 , @xmath116 is then used to extend the definition of @xmath323 to @xmath133 as well , as the unique solution to the linear recursion relation ) . introducing new variables
@xmath426 we may recast the above into the relations : @xmath427 [ fundstiel ] the weights for the fundamental motzkin path @xmath429 read : @xmath430 those for the ascending " motzkin path @xmath431 are : @xmath432 this is nothing but the non - commutative version @xcite of the stieltjes formula @xcite expressing the coefficients of the ordinary continued fraction expansion in terms of those of the series expansion . in a sense ,
our construction is a double generalization of that formula .
note that , due to the explicit value of @xmath433 , we have @xmath434 .
we conclude that for all motzkin paths @xmath44 of length @xmath36 , the quantity @xmath435 is the partition function for non - commutative weighted paths on @xmath226 .
all the weights are monic laurent monomials of the skeleton weights of theorem [ ncweights ] , hence the result is a positive laurent polynomial of the variables @xmath436 occurring in theorem [ ncweights ] . note that these are themselves laurent monomials of certain quasi - wronskians . except for the point @xmath437 of example [ fundstiel ] above , for which we have a clear laurent positive phenomenon in the variables @xmath438 ,
the classification of relevant initial data in terms of the variables @xmath439 for the equation is not clear at this stage .
all we have at our disposal is laurent positivity in terms of the non - commutative variables @xmath436 . a better understanding of what the good " dynamical variables are should lead to a notion of non - commutative cluster algebra , via the mutations thereof .
apart from the cases already treated in these notes , all of which may be viewed as particular cases of this section , an important application concerns the so - called quantum cluster algebras @xcite .
these are non - commutative versions of cluster algebras , in which the variables obey @xmath3-commutation relations .
a consequence of our construction is that the positive laurent phenomenon holds for all @xmath440 , @xmath116 for the corresponding @xmath2 quantum @xmath0-system @xcite : @xmath441 with the boundary condition @xmath442 , and where the variables within the same cluster ( or initial data set ) have the @xmath3-commutation relations for @xmath443 : @xmath444 we simply have to compute the quasi - wronskians : @xmath448 and the corresponding values of the @xmath436 variables : @xmath449 which are all laurent monomials of the initial data , with possibly some integer power of @xmath3 in factor .
the theorem follows immediately from the expressions of theorem [ ncweights ] .
e. frenkel and n. reshetikhin , _ the @xmath3-characters of representations of quantum affine algebras and deformations of @xmath450-algebras_. in recent developments in quantum affine algebras and related topics ( raleigh , nc 1998 ) , contemp .
* 248 * ( 1999 ) , 163205 .
a. n. kirillov and n. yu .
reshetikhin , _ representations of yangians and multiplicity of occurrence of the irreducible components of the tensor product of representations of simple lie algebras _ , j. sov . math .
* 52 * ( 1990 ) 3156 - 3164 .
a. kuniba , a. nakanishi and j. suzuki , _ functional relations in solvable lattice models .
i. functional relations and representation theory .
_ international j. modern phys . a * 9 * no .
30 , pp 52155266 ( 1994 ) .
stieltjes , _ recherches sur les fractions continues _ , in _
oeuvres compltes de thomas jan stieltjes _ , vol .
lxxxi , p. nordhoff , groningen , ( 1918 ) 402 - 566 ( see in particular eq.(1 ) p 402 and eq.(7 ) p 427 ) . | in this review article , we present a unified approach to solving discrete , integrable , possibly non - commutative , dynamical systems , including the @xmath0- and @xmath1-systems based on @xmath2 .
the initial data of the systems are seen as cluster variables in a suitable cluster algebra , and may evolve by local mutations .
we show that the solutions are always expressed as laurent polynomials of the initial data with non - negative integer coefficients .
this is done by reformulating the mutations of initial data as local rearrangements of continued fractions generating some particular solutions , that preserve manifest positivity .
we also show how these techniques apply as well to non - commutative settings . _
key words : _ cluster algebras , laurent phenomenon , positivity , integrable systems , non - commutative , continued fractions . _
2010 mathematics subject classification : _ 05e10 ; 13f16 ; 82b20 . |
in this note we apply a practical deterministic attack on the protocol proposed in @xcite .
this kind of attack introduced by the first author in @xcite , @xcite , @xcite and developed in @xcite works when the platform objects are linear .
it turns out that in this case , contrary to the common opinion ( and some explicitly stated security assumptions ) , one does not need to solve the underlying algorithmic problems to break the scheme , i.e. , there is another algorithm that recovers the private keys without solving the principal algorithmic problem on which the security assumptions are based .
the efficacy of the attack depends on the platform group , so it requires a specific analysis in each particular case . in general
one can only state that the attack is in polynomial time in the size of the data , when the platform and related groups are given together with their linear representations .
in many other cases we can effectively use known linear presentations of the groups under consideration .
a theoretical base for the decomposition method is described in @xcite where a series of examples is presented .
the monography @xcite solves uniformly many protocols based on the conjugacy search problem , protocols based on the decomposition and factorization problems , protocols based on actions by automorphisms , and a number of other protocols .
see also @xcite and @xcite where the linear decomposition attack is applied to the main protocols in @xcite , @xcite , and @xcite . in a series of works @xcite , @xcite and @xcite
( see also @xcite ) tsaban presented another general approach for provable polynomial time solutions of computational problems in groups with efficient , faithful representation as matrix groups . all along the paper
we denote by @xmath0 the set of all nonnegative integers , and by @xmath1 the set of all complex numbers .
in this section , we describe the andrecut s key exchange protocol proposed in @xcite .
firstly we introduce a necessary terminology . then we will give a cryptanalysis of this protocol .
let @xmath2 be a complex matrix of size @xmath3 , that is considered as a variable .
let @xmath4 be two complex polynomials in @xmath5 uniquely defined by the complex vectors of coefficients @xmath6 and @xmath7 * alice chooses the secret vectors @xmath8 and @xmath9 ( alice s private key ) . *
alice randomly generates and publishes the matrix @xmath10 ( alice s matrix public key ) .
* bob chooses the secret vectors @xmath11 and @xmath12 ( bob s private key ) . *
bob randomly generates and publishes the matrix @xmath13 ( bob s matrix public key ) .
* alice computes and publishes the matrix @xmath14 ( alice s public key ) .
* bob computes and publishes the matrix @xmath15 ( bob s public key ) . * alice calculates the secret matrix @xmath16 * bob calculates the secret matrix @xmath17 . *
the established secret key is @xmath18 it is assumed in @xcite , that the matrices @xmath19 and @xmath20 are different to give the non - commutativity assumption @xmath21 and @xmath22 . in general ,
even more strong assumption @xmath23 is not enough for this non - commutativity . also , there is a remark in @xcite that the following assumption @xmath24 should be satisfied in order to increase the security .
but by the classical cayley - hamilton theorem every matrix @xmath19 is a root of its own characteristic polynomial @xmath25 , where @xmath26 is the identity matrix of size @xmath3 .
the degree of @xmath27 is exactly @xmath28 .
then for any matrix @xmath10 and every matrix polynomial @xmath29 one has @xmath30 where @xmath31 is the remainder after division of the polynomial @xmath29 by @xmath32 hence , there is no sense to use in the protocol above polynomials of degrees @xmath33
we will provide two approaches to cryptanalysis of the protocol described in the previous section .
the first one is based on some simple facts from linear algebra and the second one is based on a linear decomposition attack . as we have seen in the previous section , there is no sense to use values @xmath34 .
thus we can assume that @xmath35 .
consider the linear systems @xmath36 of @xmath37 equations with @xmath38 unknowns @xmath39 and @xmath40 .
having a solution of the systems ( [ eq : systems ] ) one can compute the secret key as follows @xmath41 solution of a system of @xmath28 equations with @xmath28 unknowns using gauss elimination requires @xmath42 time .
thus solution of the systems ( [ eq : systems ] ) requires @xmath43 time . having precomputed values @xmath44 and @xmath45 ,
for @xmath46 , one can perform the step ( [ eq : key_computation ] ) in @xmath47 time .
so the overall time complexity for this approach is @xmath47 .
further we will describe the second approach based on a linear decomposition attack .
the algebra @xmath48 has a structure of a vector space over @xmath1 of dimension @xmath49 let @xmath50 be the semigroup generated by @xmath51 and let @xmath52 be the semigroup generated by @xmath53 a basis of the subspace sp@xmath54 can be effectively constructed as follows .
let @xmath55 be the sets of matrices considered as vectors .
define @xmath56 for @xmath57 the vector space spanned by the indicated set .
we choose a basis @xmath58 of @xmath59 , then extend @xmath60 to basis @xmath61 of @xmath62 , and so on . if for some @xmath63 we get @xmath64 then clearly @xmath65 is a basis of sp@xmath54 . by the cayley - hamilton theorem
one has @xmath66 in construction of @xmath67 we only use the gauss elimination process that is polynomial .
note , that we can do it offline , so we will call this phase the _ offline phase_. let @xmath68 be a basis of sp@xmath54 , where @xmath69 .
now we are ready to recover the secret key @xmath70 ( _ online phase _ ) .
now we will provide a rough estimate for the time complexity of the approach above .
observe that the number of the field operations in gauss elimination performed on a matrix of size @xmath74 is @xmath75 .
a basis of sp@xmath54 consist of at most @xmath37 elements , so it requires @xmath76 time to construct it .
computing @xmath77 in ( [ eq : b ] ) by solving a system of linear equations using gauss elimination requires @xmath43 time . having precomputed values @xmath78 and @xmath79 ,
for @xmath80 , one can perform the step ( [ eq : key_computation_2 ] ) in @xmath81 time .
so the offline phase could be done in @xmath82 time , the online phase could be done in @xmath43 time , and the overall time complexity is @xmath82 .
v. t. markov , a. v. mihalyov , a. v. gribov , p. a. zolotyh , and s. s. skazhenik , _ quasigroups and rings in coding and cryptoschemes constructing _ , prikladnaya discretnaya matematika , 4 ( 2012 ) , 3552 ( in russian ) . v. a. romankov , _ linear decomposition attack on public key exchange protocols using semidirect products of ( semi)groups _ , preprint , http://arxiv.org/abs/1501.01152v1[arxiv:1501.01152v1 [ cs.cr ] ] , 6 jan . 2015 .
x. wang , c. xu , g. li , h. lin , and w. wang , _ double shielded public key cryptosystems _ ,
cryptology eprint archive , report 2014/558 , version 20140718:185200 , 2014 , 114 , https://eprint.iacr.org/2014/558 .
m. habeeb , d. kahrobaei , c. koupparis , and v. shpilrain , _ public key exchange using semidirect product of ( semi)groups _ , in : _
acns 2013 _ , volume 7954 of _ lecture notes comp .
_ , p. 475486
, springer , 2013 .
d. kahrobaei , h. t. lam , and v. shpilrain , _ public key exchange using extensions by endomorphisms and matrices over a galois field _ , preprint , http://www.sci.ccny.cuny.edu/~shpil/semi_galois.pdf[http://www.sci.ccny.cuny.edu/shpil/semi_galois.pdf ] .
b. tsaban , _ the conjugacy problem : cryptanalytic approaches to a problem of dehn _ , minicourse , dsseldorf university , germany , july - august 2012 , http://reh.math.uni-duesseldorf.de/~gcgta/slides/tsaban_minicourses.pdf[http://reh.math.uni-duesseldorf.de/gcgta/slides/tsaban_minicourses.pdf ] .
b. tsaban , _ polynomial time solutions of computational problems in noncommutative - algebraic cryptography _ , journal of cryptology , 28 ( 2015 ) , p. 601622
, see also http://arxiv.org/abs/1210.8114v3[arxiv:1210.8114v3 [ cs.cr ] ] .
b. tsaban , _ practical polynomial time solutions of several major problems in noncommutative - algebraic cryptography ( preliminary announcement ) _ , iacr eprint 2014/041 .
version 20140115:201530 , jan . | we show that a linear decomposition attack based on the decomposition method introduced by the first author in monography @xcite and papers @xcite , @xcite , and developed in @xcite , works by finding the exchanging key in the protocol in @xcite . |
quantum spin liquids are magnetic systems in which large quantum mechanical zero - point spin fluctuations prevent the development of long - range order down to absolute zero temperature .
the search for real materials that display this phenomenology in two and three dimensions is a very active research topic in the field of condensed matter physics .
@xcite the interest in spin liquids stems from the expectation that they may host non - trivial quantum entanglement , topological order , as well as emergent fractionalized and deconfined low - energy excitations .
spin liquids have also been conjectured as progenitors of unconventional superconductivity at `` high temperature '' .
highly frustrated magnets are particularly apt at exhibiting large quantum spin fluctuations .
these magnets are realized in systems which consist of localized magnetic moments ( spins ) that reside on two and three dimensional lattices of corner - sharing triangles or tetrahedra and which interact with effective antiferromagnetic nearest - neighbor coupling .
@xcite such highly frustrated magnets display an exponentially large number of classical ground states .
@xcite this allows for quantum mechanical effects to be tremendously magnified compared to magnetic systems with conventional long - range magnetic order . @xcite the pyrochlore oxides , of chemical formula re@xmath0m@xmath0o@xmath1 , @xcite count a multitude of magnetic members . in re@xmath0m@xmath0o@xmath1 ,
re is a magnetic trivalent ( lanthanide , 4f ) rare - earth ion ( gd@xmath8 , tb@xmath8 , dy@xmath8 , ho@xmath8 , er@xmath8 , yb@xmath8 ) or a non - magnetic ion ( y@xmath8 , lu@xmath8 ) , while m is tetravalent , typically a transition metal ion , which can be either magnetic ( mo@xmath9 , mn@xmath9 ) or not ( ti@xmath9 , sn@xmath9 , zr@xmath9 , ge@xmath9 ) .
of relevance to the above discussion is the fact that the re and m ions reside on two distinct and interpenetrating three - dimensional lattices of corner - sharing tetrahedra . as a result ,
high geometric magnetic frustration ensues whenever the re - re or m - m interaction is effectively antiferromagnetic .
@xcite among the re@xmath0m@xmath0o@xmath1 family , the magnetic rare - earth oxides , in which re is magnetic and m is not , have been rather extensively studied .
@xcite these have revealed a number of fascinating phenomena such as long - range order induced by order - by - disorder , @xcite multiple-@xmath10 long - range ordered phase , @xcite spin liquid behavior @xcite and spin ice physics , @xcite the latter having attracted much interest .
@xcite the spin ice state displays a residual low - temperature magnetic entropy close to that found for common water ice , @xcite hence the name spin ice .
most recently , a renewed flurry of theoretical and experimental efforts have been directed at the study of spin ices for they have been argued to display an emergent coulomb phase @xcite accompanied at low energies by deconfined fractionalized magnetic charge excitations , or monopoles " .
@xcite notwithstanding the rich physics that re@xmath0m@xmath0o@xmath1 materials display , @xcite they have , until very recently , @xcite not attracted that much interest from the community of theorists and experimentalists searching for quantum spin liquids .
@xcite quantum fluctuations are expected to be , relatively speaking , more significant the smaller the spin quantum number @xmath11 . on the other hand , because of the intrinsic large spin - orbit coupling at play in the lanthanide 4f series , the total angular momentum @xmath12 , and not @xmath11 , is a good quantum number with @xmath13 being typically large across the whole lanthanide series ( e.g. @xmath14 and @xmath15 for free ho@xmath8 and gd@xmath8 ions , respectively ) .
this situation would seem unpromising to those seeking magnetic materials with potentially large quantum mechanical spin fluctuations and exotic quantum states of matter .
however , such a perspective is perhaps too pessimistic as we now discuss . in insulating magnetic compounds with 4f elements ,
the various inter - ion interactions , @xmath16 ( exchange , superexchange , virtual phonon exchange and magnetostatic dipole - dipole coupling ) are typically weak compared to the single - ion crystal - field interactions defined by a crystal - field hamiltonian @xmath17 . as a first approximation ,
one thus often proceeds by determining the energy level spectrum of @xmath17 for a fixed @xmath13 manifold .
@xcite the point group symmetry of the crystal dictates the allowed symmetry properties of @xmath17 .
these symmetries determine , in return , the spectral decomposition of the crystal - field states that derive from the original @xmath18 degenerate @xmath19 electronic ground state of the otherwise free re@xmath8 ion . in the simplest case ,
the resulting crystal field ground state is a magnetic doublet with wavefunctions @xmath20 and @xmath21 that have @xmath22 for spectral decomposition .
@xmath23 contains _ all _ the @xmath24 spectral components that transform similarly according to the point group symmetry operations .
consequently , there may or not be nonzero @xmath25 matrix elements .
this depends on the specific nature of the inter - ion couplings @xmath26 , a function of the components @xmath27 ( @xmath28 ) of angular momentum @xmath29 of ion @xmath30 , @xcite as well as the specific ion - dependent spectral decomposition of @xmath23 .
it is the nonzero @xmath25 matrix elements @xcite which determine whether significant quantum dynamics exist within the low - energy sector .
most importantly , quantum dynamics need not be ruled out despite the large @xmath31 of the isolated re@xmath8 since the crystal field hamiltonian @xmath17 entangles a superposition of the @xmath32 eigenstates of @xmath33 .
@xcite as a result , @xmath16 , by virtue of its lack of commutation with @xmath17 , can , in principle , have nonzero matrix elements between @xmath20 and @xmath21 and induce quantum dynamics within the low - energy hilbert space spanned by @xmath34 . in 4f ions with an even number of electrons ( i.e. non - kramers ion ) such as pr , tb and ho
, time - reversal symmetry imposes that @xmath35 .
@xcite consequently , in presence of solely bilinear interactions of the form @xmath36 with anisotropic @xmath37 couplings , non - kramers ions would display no quantum dynamics at low - energy and behave as effective classical ising spins s=1/2 , as in the well - studied lihof@xmath38 dipolar ising system .
@xcite in the ( pr , tb , ho)@xmath0(ti , sn , zr)@xmath0o@xmath1 materials , @xcite interactions beyond bilinear ones or consideration of the excited crystal field states are necessary to cause quantum dynamics in the low - energy sector .
@xcite in that context , multipolar interactions in pr@xmath0(sn , zr , ir)@xmath0o@xmath1 compounds @xcite and virtual crystal field excitations in the tb@xmath0(ti , sn)@xmath0o@xmath1 materials have been discussed .
@xcite conversely , odd - electron ( kramers ) ion systems ( e.g. gd , dy , er , yb ) are in principle symmetry - allowed to have nonzero @xmath39 matrix elements . the famous dy@xmath0ti@xmath0o@xmath1 spin ice compound , in which the dy@xmath8 ions are kramers ions and for which the excited crystal field states lie at @xmath40 k above the ground doublet , has been shown to be well - described by a dipolar spin ice model @xcite with classical ising spins .
@xcite this success very likely signals rather negligible interactions among the @xmath27 components beyond bilinear ones , concomitantly with the specific spectral decomposition of @xmath23 for dy@xmath8 in dy@xmath0ti@xmath0o@xmath1 .
@xcite on the other hand , er@xmath0ti@xmath0o@xmath1 and yb@xmath0ti@xmath0o@xmath1 have been known for some time @xcite to have predominant `` transverse '' @xmath39 matrix elements , @xcite along with a non - negligible @xmath41 `` longitudinal '' ( @xmath42 tensor ) component . yet
, the two compounds display quite different behaviors .
er@xmath0ti@xmath0o@xmath1 has overall antiferromagnetic interactions and develops long - range order at 1.2 k with zero propagation vector @xmath43 and zero magnetic moment per tetrahedron @xcite that is induced by order - by - disorder .
@xcite in contrast , yb@xmath0ti@xmath0o@xmath1 has overall ferromagnetic interactions and exhibits a phase transition at @xmath44 k. @xcite however , the nature of the long - range order below @xmath45 remains disputed @xcite and the high - sensitivity of the properties in the low - temperature regime ( @xmath46 mk ) on sample quality is just beginning to be understood .
@xcite it is here , within the yb@xmath0m@xmath0o@xmath1 family with overall ferromagnetic interactions and significant @xmath39 matrix elements , @xcite that the potential for an exotic class of quantum spin liquid arises a possibility that may have been casually dismissed from the naive perspective of `` there should be negligible quantum effects in such large @xmath31 systems '' .
the dy@xmath0(ti , sn , ge)@xmath0o@xmath1 and ho@xmath0(ti , sn , ge)@xmath0o@xmath1 materials are classical ising systems which may be viewed in their spin ice regime as collective paramagnets @xcite or , employing a more contemporary terminology , classical spin liquids .
@xcite as discussed above , highly frustrated magnetic systems , by virtue of their low - propensity to develop classical long - range order , are attractive candidates to search for quantum spin liquid behavior .
spin ice , reinterpreted as a classical spin liquid , @xcite may thus be viewed as a natural setting to explore how the addition of quantum dynamics may give rise to a quantum spin liquid state .
such a topic was originally explored by hermele , fisher and balents @xcite and castro - neto , pujol and fradkin @xcite a few years ago in the context of minimal theoretical toy models .
these two groups argued in their respective paper that the addition of quantum dynamics within a parent ( constrained ) classical spin ice manifold may promote the system to a u(1 ) quantum spin liquid .
such a state would be describable by a quantum field theory analogous to that of quantum electrodynamics ( qed ) in 3 + 1 dimensions . as a consequence
, this @xmath47 spin liquid state would display `` electric '' and `` magnetic '' charge excitations and an accompanying gauge boson , or `` artificial photon '' .
@xcite numerical simulations have , over the past few years , suggested that such phenomenology may indeed be at play in various lattice toy models .
@xcite such a u(1 ) quantum spin liquid state , which may be referred to as `` quantum spin ice '' , @xcite has been suggested to explain some of the properties of real materials such as tb@xmath0ti@xmath0o@xmath1 @xcite and pr@xmath0(sn , zr)@xmath0o@xmath1 .
@xcite most recently , it has been proposed that yb - based yb@xmath0m@xmath0o@xmath1 pyrochlore oxides @xcite may offer an exquisite class of systems to investigate the possible existence of a quantum spin ice state and where the complexities of virtual crystal field fluctuations @xcite and magnetoelastic coupling @xcite that complicate the tb@xmath0(ti , sn)@xmath0o@xmath1 compounds are avoided .
in a theory paper building on the original work of hermele _ et al .
_ , @xcite savary and balents have recently put forward a mean - field lattice gauge theory which identifies a number of possible phases , the most exotic ones being the aforementioned u(1 ) spin liquid as well as a novel coulomb ferromagnetic state .
@xcite this approach has been further extended to non - kramers ion systems .
@xcite finally , the question of how to probe the emergent photon in quantum spin ice via inelastic neutron scattering measurements has very recently been discussed .
@xcite one particularly interesting aspect of the search for quantum spin liquids in the yb - based pyrochlores , and perhaps pr - based pyrochlores as well , @xcite is that the microscopic hamiltonian is parametrized by a handful of independent anisotropic exchange parameters @xmath48 ( four for yb@xmath8 and three for pr@xmath8 ) @xcite between effective spin-1/2 degrees of freedom on each pyrochlore lattice site .
furthermore , it may be that the long - range dipolar interactions , so important in classical dy@xmath0(ti , sn)@xmath0o@xmath1 and ho@xmath0(ti , sn)@xmath0o@xmath1 spin ices with the large magnetic dy@xmath8 and ho@xmath8 magnetic moment , @xcite can perhaps be neglected as a first approximation . to make definite progress at this time ,
a determination of the effective anisotropic exchange from experiments and theory using controlled approximations is required in order to position a candidate spin ice material in the phase diagram of ref .
[ ] , or the corresponding phase diagram relevant to non - kramers ions .
@xcite in the case of yb@xmath0ti@xmath0o@xmath1 , the determination of those interactions from a series of experiments @xcite has led to different @xmath48 sets with very different numerical values .
perhaps most noteworthy , a determination of those parameters from a fit to spin waves in strong magnetic field measured via inelastic neutron scattering @xcite give values significantly different than those obtained by fitting the zero - field diffuse paramagnetic neutron scattering using a random phase approximation ( rpa ) method @xcite at a temperature @xmath49 k or a subsequent polarized neutron scattering version , also using rpa as the fitting procedure , but now very near @xmath45 .
@xcite encouragingly , however , recent numerical linked - cluster ( nlc ) calculations @xcite have found the zero - field magnetic specific heat data of yb@xmath0ti@xmath0o@xmath1 to be well described above 0.7 k by the @xmath48 set obtained from the inelastic neutron scattering , @xcite but not by the other sets determined from rpa fits to diffuse neutron scattering .
@xcite it is clearly desirable to fit bulk measurements to determine the @xmath48 parameters in cases where inelastic neutron scattering studies are impractical , and to corroborate such neutron studies and understand thermodynamic and bulk magnetic properties in a common framework with neutron studies whenever possible .
yet , the scarcity of controlled numerical methods readily available to calculate the thermodynamic properties of frustrated three - dimensional anisotropic quantum spin systems in a temperature regime where non - trivial correlations develop do not make this task straightforward . with the seeming success of previous nlc calculations applied to yb@xmath0ti@xmath0o@xmath1 , @xcite we are thus naturally led to ask : _ can one convincingly demonstrate that nlc does provide a controlled method to describe bulk data for a three - dimensional frustrated quantum ( spin ice ) system as it progressively enters its low - temperature strongly correlated regime ?
_ in this paper , we address this question by extending the work of ref .
[ ] by computing the thermodynamic properties of yb@xmath0ti@xmath0o@xmath1 in nonzero magnetic field using nlc and by comparing the results of such calculations with measurements above 2 k on single crystals .
we show , through a comparison with nlc , that the effective hamiltonian with its anisotropic exchange parameters @xmath50 previously determined by fitting spin waves in the field polarized paramagnetic state @xcite describes , with no adjustment , the temperature @xmath4 and magnetic field @xmath2 dependence of the magnetization , @xmath3 , of yb@xmath0ti@xmath0o@xmath1 . as a consequence ,
we demonstrate simultaneously the usefulness of the nlc method and further validate @xcite the quantitative merit of the effective spin hamiltonian of ref .
[ ] to describe yb@xmath0ti@xmath0o@xmath1 .
the rest of the paper is organized as follows . in the next section ,
we describe the effective spin-1/2 model for yb@xmath0ti@xmath0o@xmath1 along with the nlc method . in section
[ experimental ] , we discuss the details of the experimental method employed .
the nlc results are presented in section [ results ] while section [ confront ] provides a comparison between experiment and theory .
a brief discussion in section [ discussion ] concludes the paper .
symmetry considerations imply that nearest - neighbor bilinear exchange constants on the pyrochlore lattice can be parametrized in terms of @xmath51 distinct exchange parameters and two @xmath52-factors . in the local basis , this model is defined by the hamiltonian @xcite @xmath53\nonumber \\ & & + j_{z\pm } [ ( s_i^z(\zeta_{ij}s_j^+ + \zeta_{i , j}^*s_j^- ) + i \leftrightarrow j ] \ } .
\label{hqsi}\end{aligned}\ ] ] several notation conventions are possible for @xmath54 .
@xcite here , we adopt the one used in ref .
[ ] . in eq .
( [ hqsi ] ) , @xmath55 refers to nearest - neighbor sites of the pyrochlore lattice , @xmath56 is a @xmath57 complex unimodular matrix , and @xmath58 .
@xcite the @xmath59 quantization axis is along the local @xmath60 $ ] direction , @xcite and @xmath61 refers to the two orthogonal local directions .
the @xmath52 tensor takes values @xmath42 along the local @xmath60 $ ] cubic direction and @xmath62 perpendicular to it . in the presence of an applied external magnetic field @xmath63 , an additional zeeman interaction @xmath64 is added to @xmath54 , giving a total spin hamiltonian @xmath65 . here
@xmath66 is the @xmath52-tensor with eigenvalues @xmath67 .
@xcite as already stated above , there have been several experimental attempts to determine these @xmath68 effective exchange parameters leading to widely different results .
@xcite among them , ross _
@xcite used inelastic neutron scattering ( ins ) data obtained from measurements in high magnetic fields ( i.e. in the polarized paramagnetic state ) to determine the @xmath68 couplings for yb@xmath0ti@xmath0o@xmath1 : @xmath69 , @xmath70 , @xmath71 , and @xmath72 , all in mev and with @xmath73 and @xmath74 . in a recent study , @xcite it was found that the zero - field specific heat and entropy deduced from the data of blte _ et al .
_ @xcite are well described only by the exchange parameters obtained by ross _
_ @xcite a key goal of the present paper is to investigate further the validity of the hamiltonian ( [ hqsi ] ) as a description of the bulk properties of yb@xmath0ti@xmath0o@xmath1 with the @xmath48 exchange couplings from ref .
we do so by performing calculations of the thermodynamic properties of model ( 1 ) as a function of temperature @xmath4 and magnetic field @xmath2 and compare the numerical results with those obtained from experimental measurements .
we find an excellent agreement without any adjustment of the parameters determined by ross _
et al . _ , @xcite hence providing a strong validation to the @xmath50 parameters determined by ins .
as a corollary , our work confirms that fitting the ins spectra at high magnetic field is an excellent method to determine the exchange constants for pyrochlore oxides with well isolated magnetic ground doublets and which are well described by an effective spin - half model .
@xcite further implications of this agreement will be discussed in the concluding section [ discussion ] . to calculate the thermodynamic properties of model ( [ hqsi ] )
, we turn to numerical linked - cluster ( nlc ) expansions .
@xcite in this method , an extensive property @xmath75 ( such as heat capacity or magnetization ) of a thermodynamic system is calculated as a sum over contributions from different clusters embedded in the lattice : @xmath76 here @xmath77 is the count of the cluster per lattice site , defined as the number of ways to embed the cluster .
@xmath78 is the weight of the cluster which is obtained by calculating the property for a given cluster and subtracting the weight of all its subclusters .
@xmath79 here , the sum runs over all subclusters of the cluster @xmath80 .
a subcluster is defined as any cluster smaller than @xmath80 that can be embedded in cluster @xmath80 .
this scheme can be used to develop power series expansions , such as high temperature series expansions in powers of inverse temperature @xmath81 , or , some coupling constant expansion ( such as expansion in inverse field - strength ) .
it can also be used to numerically calculate properties for a given value of temperature and coupling constants , as we do in the present work .
the pyrochlore lattice is a lattice of corner - sharing tetrahedra and it proves useful to develop nlc in terms of clusters consisting of complete tetrahedra .
we have calculated temperature and field - dependent properties up to @xmath82 order , that is , including contributions from all clusters made of up to @xmath51 tetrahedra .
these nlc calculations are numerically exact in two limits .
they are so at high temperatures , since corrections to @xmath82 order nlc is of order @xmath83 in the high temperature series expansion for @xmath84 .
the nlc calculations are also exact at high magnetic field @xmath2 at all temperatures since corrections to @xmath82 order nlc for @xmath84 is of order @xmath85 at @xmath86 with exponentially small corrections @xmath87 at finite temperatures ( @xmath80 is some @xmath2 and @xmath4 independent constant ) .
the parameter region where nlc begins to lose accuracy is when the temperature and the applied field ( zeeman energy ) are both smaller than the exchange constants .
thus , as long as either the high temperature expansion or the high - field expansion converges , the nlc calculations should be highly accurate .
the reason for developing nlc in terms of complete tetrahedra comes from the fact that , for spin ice systems , the ` ice rule ' constraints at the origin of spin ice physics are local to tetrahedra .
thus , having clusters with only parts of a tetrahedron would cause wild oscillations in the calculated properties as the constraints can not be satisfied for such a cluster . having only clusters with full
tetrahedra allows the system to always satisfy the ice - rule constraint .
for example , in the case of the classical nearest - neighbor spin - ice exchange model , such a tetrahedra - based nlc was found to be highly accurate .
@xcite in fact , the first order nlc , which uses a single tetrahedron and is equivalent to pauling s approximation for the entropy of ice , gives quantities at all temperatures ( in zero external field ) that are accurate to a few percent .
@xcite however , in the quantum spin ice problem , nlc must break down at low temperatures due to the development of either long - range correlations or long - range entanglement .
we have found ( see results below and in ref . [ ] ) that the first signature of such a breakdown in the low temperature and low field region is an alternation of the thermodynamic property @xmath75 considered .
this reflects the fact that , as system sizes increase , the thermodynamic properties must approach their infinite size values in some way .
now , imagine that , at some order , the numerically calculated property is a bit too large , giving rise to a large cluster weight @xmath88 .
subcluster subtraction then causes the weight in the subsequent order to be too small .
this , in return , causes an alternation with the nlc order considered in the property obtained by restricting the sum to some order .
such an alternation is handled well by using the euler summation method , @xcite which ensures that an alternating piece is completely eliminated from the partial sums at all orders .
thus , the euler resummed properties are only missing the longer range correlations which are necessarily absent in finite order nlc . for yb@xmath0ti@xmath0o@xmath1
, we have found that in zero magnetic field @xcite the thermodynamic properties converge down to 2 k without the euler summation and down to about 1 k with euler summation ( the largest exchange constant for yb@xmath0ti@xmath0o@xmath1 , @xmath89 in eq .
( [ hqsi ] ) , is approximately 2 k @xcite ) .
nlc using the tetrahedral basis requires exact diagonalization of increasingly larger tetrahedral clusters . using a single intel^^ core^^ i7 - 920 processor and freely - available lapack linear algebra routines , diagonalizations for clusters of one tetrahedron ( four sites ) and two tetrahedra ( seven sites )
could be done in less than a second , while the three - tetrahedron ( 10-site ) cluster still required less than 10 seconds .
computing only the spectrum for a single four - tetrahedron ( 13-site ) cluster required approximately 25 minutes of cpu time and 2.1 gb of memory , or slightly over twice the memory required to store the full hamiltonian matrix ( @xmath90 complex numbers ) . generating the full set of eigenstates required between 4 and 8 gb of memory . with the computer resources available to us , full exact diagonalization of larger systems , and thus higher orders of nlc expansion ,
were prohibited by memory requirements .
( color online ) .
list of topological clusters with complete tetrahedra . in an applied field , each cluster must be treated as several distinct clusters as explained in the text .
the integer @xmath91 refers to the order at which the cluster arises and @xmath92 gives the cluster count in zero - field.,width=302 ] a list of different topological clusters with complete tetrahedra is shown in fig .
[ fig : clusters ] . a single site must be treated as a cluster as well , but is not shown . because the local quantization @xmath59 axis varies from sublattice to sublattice in the re@xmath0m@xmath0o@xmath1 pyrochlore oxides , @xcite in the presence of a magnetic field , each cluster with different types of sublattice site must be treated separately .
for example , each of the four single - site graphs ( in nlc order @xmath93 ) must be treated separately for a general field direction .
a single tetrahedron graph remains unique even in a field .
but , the graph with two tetrahedra joined at a point ( nlc order @xmath94 ) , is really 4 distinct graphs that depends on the type of site ( i.e. sublattice ) that is shared between the two tetrahedra .
similarly , all higher order graphs must be split into several distinct graphs to complete the calculation .
we use the eigenvalue spectrum @xmath95 of @xmath65 to compute the partition function @xmath96 : @xmath97 the expectation value for the internal energy is computed from the formula @xmath98 this quantity is used , in turn , to compute the entropy as @xmath99 to which the heat capacity is related as @xmath100 the average magnetization per site @xmath6 is calculated as a derivative of the free energy with respect to the applied magnetic field @xmath2 : @xmath101 in practice , an approximation of this derivative at a given field value requires two separate evaluations of the free energy separated by a small change of field strength , in this case @xmath102 t. we believe in the correctness of the results from our calculations of the field - dependent properties for the following three reasons .
first , the same results were obtained from two independently written computer programs .
second , we checked that in the limit @xmath103 , the zero - field results @xcite were reproduced .
third , we checked that at high temperatures the weights of the clusters were found to scale with the expected powers of inverse temperature @xmath81 .
these powers can be deduced from considerations of a high temperature expansion of the quantity of interest .
consider , for example , an expansion for @xmath84 .
in such an expansion either at least two powers of @xmath104 arise for each tetrahedron or at least one power of @xmath81 arises for each tetrahedron together with two powers of @xmath81 from placing the zeeman @xmath105 field - term on sites at the outside perimeter of the cluster , as needed to give a non - zero contribution to the trace in eq . [ eq : z ] .
thus , the weight of the 4-tetrahedra cluster is of order @xmath106 in zero field and of order @xmath83 in non - zero field .
these latter checks are non - trivial and demonstrate that all subgraphs have been properly subtracted within the nlc procedure .
nlc generates a sequence of property estimates @xmath107 with increasing order @xmath91 , where @xmath108 .
the convergence of such a sequence can be improved by euler summation .
@xcite in general , given alternating terms @xmath109 , the precise infinite - size lattice property @xmath110 is approached by the sum ( with @xmath91 even ) @xmath111,\ ] ] where @xmath112 is the forward difference operator @xmath113 usually , a small number of terms are computed directly , and the euler transformation , @xmath114 defined below , is applied to the rest of the series . in our case , where direct terms were available up to fourth order , we began the transformation after the second order , so that the third and fourth order euler - transformed property estimates , @xmath115 and @xmath116 respectively , are given by @xmath117
as we are aiming to establish a close connection between experiment and theory for the temperature and field dependence of the magnetization @xmath6 , it is necessary to perform adequate demagnetization corrections and this , in turn , necessitates measurements on single crystals .
magnetization experiments were thus carried out on three single crystal pieces .
the samples were cut from two large single crystals grown by the optical floating zone method , using a growth procedure similar to that previously employed for growing single crystals of tb@xmath0ti@xmath0o@xmath1 .
@xcite one crystal , which provided samples aligned along [ 110 ] and [ 100 ] , was grown at a rate of 6mm / h in oxygen pressure of 4atm .
the second crystal , from with a piece aligned along [ 111 ] was cut , was grown at 5mm / h in 2atm of oxygen .
the samples were aligned using x - ray diffraction to within 2 degrees of of each of the three high - symmetry directions ; [ 100 ] , [ 110 ] , and [ 111 ] .
two single crystal samples were cut into rectangular prisms ( `` needles '' ) measuring 0.74 mm @xmath118 0.74 mm @xmath118 3.74 mm and 0.69 mm @xmath118 0.65 mm @xmath118 2.67 mm , respectively , with [ 111 ] and [ 100 ] directions oriented along their long axes . the applied magnetic field was also oriented parallel to the long axis .
the third single crystal piece was cut and polished into a rounded triangular shape which we approximate , for the purposes of demagnetization corrections , as an ellipsoid with major axes @xmath119 = 1.85 mm , @xmath120 = 1.5 mm and @xmath80 = 0.8 mm .
the @xmath119 direction made an angle of 22@xmath121 with respect to both [ 110 ] and the applied field direction within the @xmath122 plane .
the magnetization of the same [ 110 ] sample was also reported in ross _
et al._,@xcite but without a demagnetization correction applied .
the data was collected with a quantum design mpms instrument , which uses a squid magnetometer to measure the dc magnetization in magnetic fields up to 5 t and temperatures as low as 1.8 k. the magnetization data for the [ 100 ] and [ 111 ] samples were corrected for demagnetization effects using an approximate formula for the demagnetization field in rectangular prisms .
@xcite the data for the [ 110 ] sample was corrected by approximating it as a very flat ellipsoid , with the field direction 22@xmath121 from the ( long ) @xmath119 direction . in general , the susceptibility , @xmath123 , can be corrected to account for the demagnetization field using the following formula , in si units , @xmath124 where @xmath125 is the actual susceptibility that we aim to compare with the results of nlc calculations .
@xmath126 is the apparent susceptibility ( given by the measured @xmath127 ) and @xmath128 is the demagnetization factor , which depends on the sample geometry .
( color online ) .
the figure illustrates the evolution of the magnetization divided by the field strength , @xmath123 , with nlc order @xmath91 and convergence towards experimental data .
panels a ) and b ) are for @xmath129 t and @xmath130 t , respectively , and with the field @xmath2 applied along the @xmath131 $ ] direction .
solid curves are for the nlc calculations and the number beside each curve corresponds to the order @xmath91 up to which the calculations were carried to .
pluses are the demagnetization corrected experimental @xmath6 results divided by @xmath2.,title="fig:",width=302,height=226 ] ( color online ) .
the figure illustrates the evolution of the magnetization divided by the field strength , @xmath123 , with nlc order @xmath91 and convergence towards experimental data .
panels a ) and b ) are for @xmath129 t and @xmath130 t , respectively , and with the field @xmath2 applied along the @xmath131 $ ] direction .
solid curves are for the nlc calculations and the number beside each curve corresponds to the order @xmath91 up to which the calculations were carried to .
pluses are the demagnetization corrected experimental @xmath6 results divided by @xmath2.,title="fig:",width=302,height=226 ] the temperature dependence of the magnetization @xmath6 divided by the strength of the applied magnetic field @xmath2 along the cubic @xmath131 $ ] direction and calculated using the nlc method is shown in fig .
[ fig : merit_nlc ] . panels a ) and b ) of fig .
[ fig : merit_nlc ] show the results for a field of 0.2 t and 1.0 t , respectively .
the experimental values of @xmath123 , after demagnetization corrections , are shown for the same applied field values and are marked by black pluses . the number beside each curve labels the nlc order at which @xmath6 was calculated ( see section [ model ] ) .
nlc-0 ( red curve , label `` 0 '' ) considers a single site cluster .
it therefore does not incorporate the effect of the interactions in eq .
[ hqsi ] between the yb@xmath8 ions and thus corresponds to the yb@xmath8 single - ion property with @xmath52-tensor components @xmath73 and @xmath74 .
the significant differences between nlc-0 and the experimental data emphasizes the importance of interactions and concurrent development of correlations below 10 k. the importance of interactions on causing a departure of @xmath123 from its single - ion value below @xmath7 k had previously been noted on the basis of a determination of the local susceptibility from polarized neutron measurements @xcite as well as a subsequent theoretical calculation of the local susceptibility @xcite using exchange parameters determined from rpa fits to the diffuse paramagnetic scattering .
@xcite however , none of these works @xcite had the quantitative descriptive power of the present nlc calculations . compared to nlc-0 , nlc-1 incorporates the effect of interactions and correlations at the scale of one tetrahedron ( see fig . [
fig : clusters ] ) .
focusing on the case @xmath129 t , one observes little difference between nlc-0 and nlc-1 above @xmath132 k , but a large increase in @xmath123 in going from nlc-1 to nlc-2 for @xmath133 k. the results for nlc-1 requires the exact diagonalization of the hamiltonian over a single tetrahedron and therefore , incorporates spatial spin - spin correlations extending only over a nearest - neighbor distance ( see panel for @xmath134 in fig .
[ fig : clusters ] ) .
in contrast , nlc-2 considers two tetrahedra ( see panel for @xmath94 in fig . [
fig : clusters ] ) and therefore considers correlations , and associated fluctuations , reaching out to third nearest neighbors . as first noted by ross _
et al . _ , @xcite and further expanded upon by applegate _
et al . _ , @xcite the fluctuations of the joining spin / site of two tetrahedra mediates an effective ferromagnetic third nearest - neighbor coupling ( @xmath135 in ref . [ ] ) .
this fluctuation - induced interaction promotes ferromagnetic correlations among the otherwise degenerate classical 2-in/2-out spin ice states , an effect that we believe important to induce ferrimagnetic correlations in yb@xmath0ti@xmath0o@xmath1 .
@xcite the incorporation of this fluctuation process and induced effective @xmath135 coupling only happens for nlc order @xmath136 and is thus absent for nlc-1 .
we believe that the large increase of @xmath123 in going from nlc-1 to nlc-2 results from the ability of clusters of two - tetrahedra to support that type of fluctuation physics while a single tetrahedron can not .
one may want to argue that the observed large increase of the calculated @xmath123 , when going from nlc-1 to nlc-2 , is further evidence that yb@xmath0ti@xmath0o@xmath1 , as described by the effective exchange parameters of ref .
[ ] , is either on its way or at the verge of developing spontaneous ferrimagnetic order at low - temperature .
we return to this point below in subsection [ field : cv ] when we discuss the magnetic entropy , @xmath137 , in nonzero magnetic field @xmath2 . considering the results of fig .
[ fig : merit_nlc]a ) , we observe that the nlc-2 , 3 and 4 results are quite close to each other down to @xmath138 k , the lowest temperature available for the experimental data .
most importantly , one should note that nlc-2 , 3 and 4 , using the effective anisotropic exchange parameters @xmath139 of ref .
[ ] already provide , without euler summation , and _ without any _ adjustment of the parameters defining @xmath54 and @xmath105 , a very good agreement with the experimental data .
a similarly good agreement is also found between the experimental data and the nlc results for the highest nlc orders ( @xmath140 ) for a field @xmath141 t ( see panel b ) of fig .
[ fig : merit_nlc ] ) . having demonstrated that ( i ) interactions do play a strong effect in renormalizing @xmath6 for @xmath7 k and that ( ii ) nlc orders @xmath142 and @xmath143 give a highly suitable description of the experimental data even for the weakest field @xmath129 t considered ( see the discussion in section [ model ] ) , we henceforth solely report results from the euler transformation method ( to order @xmath142 and @xmath143 , see eq .
( [ euler_trans ] ) ) to generate theoretical values of @xmath123 vs temperature and for various field @xmath2 along the three high - symmetry cubic directions ( @xmath131 $ ] , @xmath144 $ ] , and @xmath60 $ ] ) .
figure [ fig : euler_results ] shows @xmath123 , obtained using the @xmath82 order euler transformed nlc results , as a function of temperature @xmath4 for fields @xmath2 of strength 0.2 , 1 , 3 , and 5 t oriented parallel to the @xmath131 $ ] , @xmath144 $ ] and @xmath60 $ ] directions .
again , the calculation employed the microscopic hamiltonian appropriate to yb@xmath0ti@xmath0o@xmath1 as derived from inelastic neutron scattering data at high field .
@xcite figure [ fig : euler_results ] shows 4@xmath145 order nlc results for @xmath123 per yb@xmath8 ion vs temperature @xmath4 , with both quantities plotted on a log scale .
the calculated magnetization is relatively independent of direction and levels off at low temperatures , with the temperature at which it levels off being lower with smaller applied magnetic fields .
while the magnetization is only weakly dependent on the direction of the applied magnetic field , such differences in the magnetization as a function of direction of field are only evident at low temperatures .
( color online ) . calculated magnetizations as a function of temperature for different field strengths and directions
are shown using the 4@xmath145 order euler transformed results from the nlc expansion ( see .
( [ euler_trans ] ) ) .
the magnetic fields are applied along the @xmath131 $ ] , @xmath144 $ ] , and @xmath60 $ ] directions .
, width=302 ]
figure [ fig:3panels ] shows the experimentally - determined magnetization as a function of temperature on a semi - log plot .
data is shown for applied magnetic field directions parallel to @xmath131 $ ] , @xmath144 $ ] , and @xmath60 $ ] , from top to bottom , and for field strengths of 0.2 , 1 , 3 , and 5 t. once again @xmath123 is normalized per yb@xmath8 ion .
two sets of nlc expansion calculations using the euler transformation are shown in fig .
[ fig:3panels ] ; to order @xmath142 are shown as the solid lines , and to order @xmath143 are shown as the dot - dashed lines . here
again , the nlc expansions utilized the microscopic hamiltonian derived from the high field inelastic neutron scattering ( ins ) data @xcite with no adjustment in the parameters so determined .
several features are immediately clear from this comparison between theory and experiment .
first , the ( euler - transformed ) nlc expansion results to both @xmath142 and @xmath143 orders provide an excellent description of the magnetization for all field strengths and directions as well as all temperatures considered .
the fact that a remarkable degree of quantitative agreement between theory and experiment is achieved without adjustment of the microscopic hamiltonian appropriate to yb@xmath0ti@xmath0o@xmath1 is strong validation of the determination of the microcopic hamiltonian .
@xcite as the ins measurements were performed at @xmath146 k and applied magnetic fields of @xmath147 2 and 5 t with @xmath2 parallel to @xmath144 $ ] , and the magnetization measurements are performed for t @xmath148 1.8 k and fields @xmath149 5 t , we now have a very accurate description of yb@xmath0ti@xmath0o@xmath1 using the same microscopic hamiltonian over a remarkably large region of its @xmath150 phase diagram .
looking at the euler - transformed nlc expansion results for @xmath142 ( solid line ) and @xmath143 ( dot - dashed line ) in fig .
[ fig:3panels ] , one can see that , as expected , the two calculations are quite consistent with each other for @xmath151 1.8 k , but depart from each other at lower temperatures . there is also a better agreement at lower temperatures between @xmath142 and @xmath143 nlc expansions for higher fields , independent of the direction of the applied magnetic field , as anticipated on the basis of the general arguments presented in section [ model ] .
( color online ) . calculated magnetizations using the @xmath142 and @xmath143 euler - transformed @xmath123 nlc expansion results ( see eq .
( [ euler_trans ] ) ) are compared with the measured magnetization versus temperature for fields applied along the @xmath131 $ ] , @xmath144 $ ] and @xmath60 $ ] directions .
the parameters for @xmath65 are those determined form the spin wave results of ref .
appropriate demagnetization corrections have been applied to all experimental data .
departures between @xmath142 and @xmath143 nlc expansion results become noticible at temperatures @xmath152 k.,width=340 ] there have been over the last four years , apart from ref .
[ ] , a number of other studies @xcite combining experiment and theory and which were aimed at determining the strength of the anisotropic interactions in yb@xmath0ti@xmath0o@xmath1 .
we believe that most , if not all , including ref . [ ] that was co - authored by one of us , are beset by significant drawbacks compared to the in - field inelastic neutron scattering measurements of ross _
et al . _
@xcite in fact , one might have wondered whether the anisotropic exchange parameters @xmath48 extracted in ref .
[ ] in magnetic fields of 5 tesla might have suffered from large renormalization due to magnetoelastic effects . the good agreement between experimental and nlc results for zero magnetic field specific heat @xcite and the ones presented in this paper for the temperature and magnetic field dependent magnetization , @xmath153 , provide compelling evidence that such renormalization , if it exists , is well within the experimental uncertainty of the estimated @xmath48 parameters .
@xcite that being said , we now discuss each of the other works .
cao _ et al .
_ used polarized neutron diffraction to extract the local susceptibility tensor of a number of re@xmath0ti@xmath0o@xmath1 pyrochlores , including yb@xmath0ti@xmath0o@xmath1 , in an applied external field of 1 tesla along the [ 100 ] crystallographic direction .
@xcite their analysis was based on a mean - field theory that ignores the sublattice nature of the pyrochlore lattice and incorporates the effect of the local mean - field only via two independent coupling constants referred to as @xmath154 and @xmath155 . by construct ,
such an approach makes it difficult to reconstruct the microscopic exchange parameters of the spin hamiltonian .
furthermore , as they do not comment on this , it is not clear that their data analysis took into account demagnetization effects .
_ reinvestigated the description of the bulk and local susceptibility of re@xmath0ti@xmath0o@xmath1 compounds , but now starting from a microscopic formulation , and incorporating an adequate sublattice structure in their model as well as including demagnetization corrections .
@xcite in the case of yb@xmath0ti@xmath0o@xmath1 , they reported being unable to describe the longitudinal site susceptibility @xmath125 of cao _ et al .
_ by using a single set of crystal field parameters .
it is perhaps important to note that malkin _
_ assumed that the bilinear anisotropic interactions between the magnetic moment operators were symmetric , which amounts to neglecting dzyaloshinskii - moriya ( dm ) -like interactions .
as was found in the work of ross _
, @xcite the @xmath156 interactions , which originate from the dm interactions , are the second largest ones in @xmath54 , and of almost the same magnitude as the largest @xmath89 coupling .
hence , the neglect of the antisymmetric interactions with coupling @xmath157 for yb@xmath0ti@xmath0o@xmath1 is bound to cause difficulty in obtaining a quantitative description of the properties of this material . consequently , because of the aforementioned caveats , the results obtained in refs . [ ] can not not provide a quantitative account of the microscopic interactions at play in yb@xmath0ti@xmath0o@xmath1 .
et al . _
@xcite were the first to consider a microscopic theory which incorporates the single - ion crystal field hamiltonian , @xmath17 , for yb@xmath8 in yb@xmath0ti@xmath0o@xmath1 , the four symmetry - allowed bilinear nearest - neighbor interactions , @xmath37 between the @xmath29 angular momentum operators along with the long - range magnetostatic dipole - dipole interactions .
@xcite in order to determine the @xmath37 , thompson _
_ used a random phase approximation ( rpa ) to calculate the diffuse ( energy - integrated ) neutron scattering intensity @xmath158 in the @xmath159 $ ] scattering phase and compare with experimental measurements at a temperature @xmath160 k. @xcite with the curie - weiss temperature of yb@xmath0ti@xmath0o@xmath1 given by @xmath161 k @xmath61 0.2 k , it would have seemed that rpa , which is in essence a mean - field scheme , may be a priori reasonably quantitatively accurate at a temperature of 1.4 k with a frustration / fluctuation level set by @xmath162 . forced by construction to describe short - range diffuse scattering , the rpa fit of ref . [ ]
provided a set of couplings @xmath37 allowing for a good fit of @xmath158 and , by `` retroactive consistency '' , a mean - field critical temperature @xmath163 k. when translating the determined @xmath37 parameters in the @xmath139 notation of ross _
, one finds a large difference between the two sets .
the authors of ref .
[ ] commented in the supplementary material section of their paper that a possible reason for the failure of the model of thompson _ et al . _
@xcite to agree with the results of ross _
et al . _
@xcite is that the former work neglected multipolar interactions between the @xmath29 operators beyond the considered bilinear ones .
however , as discussed in the supplementary material of ref .
[ ] , such critique @xcite would appear of little merit for the following reason .
thanks to the essentially total isolation of the kramers crystal field doublet of yb@xmath8 from the excited states , the ground crystal field doublet acts for an almost exact _
invertible _ unitary transformation of the @xmath164 interactions , hence providing a one - to - one correspondence between the bilinear @xmath37 couplings and the @xmath48 effective anisotropic exchange between effective spin-1/2 operators of ross _
et al . _
@xcite in other words , the bilinear @xmath164 model of ref .
[ ] can be viewed as a `` high energy '' model whose projection ( with ` correct ' values of the @xmath37 couplings ) gives the @xmath165 interactions in the model of eq .
( [ hqsi ] ) [ ] . the difficulty with thompson _
et al . _ s results
is readily understood on the basis of the mean - field theory results presented in ref .
the mean - field @xmath166 using the @xmath48 parameters from ref .
[ ] is approximately 3.5 k. such a high @xmath166 compared to the 1.4 k rpa fit of @xmath158 means that the extracted @xmath37 parameters suffer from a very significant and uncontrolled renormalization from thermal ( and possibly quantum ) fluctuations .
concerns that 1.4 k might be too low for a quantitative rpa fit of @xmath158 of yb@xmath0ti@xmath0o@xmath1 and that fits at the higher temperature of 9.1 k data @xcite might have been more appropriate had been expressed in ref .
[ ] . unfortunately , the diffuse signal at 9.1 k proved too weak to proceed .
as demonstrated by applegate _
_ in their nlc calculations , @xcite the anisotropic exchange parameters of ref .
[ ] provide a poor description of the zero - field magnetic specific heat @xmath167 of yb@xmath0ti@xmath0o@xmath1 .
chang _ et al .
_ most recently reported their own estimate of the exchange parameters for yb@xmath0ti@xmath0o@xmath1 .
@xcite they also employed an rpa scheme to fit the ( polarized ) neutron scattering intensity of the compound , but using an effective spin-1/2 model rather than the bilinear @xmath164 interactions plus @xmath17 of thompson _ et al . _
@xcite perhaps believing in the incorrectness of the latter description ( see discussion two paragraphs above ) , they passed over in the body of their paper the opportunity to offer a critique of thompson _ et al .
_ s model .
that said , with the published evidence from ref .
[ ] that @xmath166 for yb@xmath0ti@xmath0o@xmath1 may be as high as 3.5 k @xcite , and that the rpa fits of thompson _ et al .
_ at 1.4 k may thus be of questionable quantitative merit , it is surprising that chang _
_ did nevertheless proceed to use rpa to fit the polarized neutron scattering at a temperature as low as 0.3 k , which is a mere @xmath168 percent higher than the experimental @xmath169 k. it is also not clear how the authors of ref . [ ] find a similarity between their @xmath48 values and those of ross _
et al . _ , especially after evidence had been reported by applegate _
_ , @xcite that the @xmath48 of ref .
[ ] fail to describe @xmath167 , while those of ross _
provide a quite adequate description of @xmath167 .
@xcite chang _
_ propose a rationalization of the low - temperature state of yb@xmath0ti@xmath0o@xmath1 on the basis of a higgs - like phase and suggest positioning this material in a @xmath48 parameter space near a parent classical spin ice . yet , at the same time , they overlook to comment on the inability of their microscopic model with its rpa - determined set of @xmath48 to describe @xmath167 .
@xcite as in the case of thompson _ et al .
_ s rpa analysis,@xcite chang _
s fit to the neutron scattering data provides , by the very consequence of using rpa at a temperature @xmath170 , for a set of coupling parameters which are significantly and uncontrollably renormalized downward compared to the bare @xmath139 .
this is illustrated in fig .
[ fig : chang_no_inter ] where it is shown that nlc-1 and nlc-4 using the @xmath48 parameters of chang _
et al . _ do not describe @xmath6 for temperatures @xmath7 k , where correlations have barely started to develop .
@xcite in fact , down to @xmath171 k , there is little difference between either nlc-1 and nlc-4 with the single - ion magnetization with all interactions turned off and given by the nlc-0 results . to conclude this discussion , it thus appears that only ross _
_ s set of microscopic parameters @xcite consistently describe yb@xmath0ti@xmath0o@xmath1 for fields @xmath172 tesla and temperatures @xmath173 k. ( color online ) .
the figure shows that @xmath6 from nlc calculations ( orders @xmath174 ) using the @xmath50 parameters reported by chang _
et al . _
@xcite fail dramatically to describe the experimental @xmath6 results ( plus symbols ) .
see text.,width=302 ] ( color online ) . the calculated molar specific heat using nlc
is shown as a function of temperature and magnetic field with field applied along , from top to bottom , the @xmath131 $ ] , @xmath144 $ ] and @xmath60 $ ] directions .
these results were calculated using the @xmath142 and @xmath143 euler transformation nlc results using the @xmath50 parameters of ref .
see the legends in top panel of fig .
[ fig : entropy ] for values of the field strengths considered and legend for nlc order employed.,width=340 ] ( color online ) .
the calculated molar entropy using nlc is shown as a function of temperature and magnetic field with field applied along , from top to bottom , the @xmath131 $ ] , @xmath144 $ ] and @xmath60 $ ] directions .
these results were calculated using the @xmath142 and @xmath143 euler transformation nlc results using the @xmath50 parameters of ref .
[ ] . , width=340 ] while we are not aware of in - field specific heat , @xmath175 , measurements on yb@xmath0ti@xmath0o@xmath1 single crystals , we expect these to be soon carried out given the interest devoted to this compound . one perspective as to why such measurements might be of interest is the following .
the er@xmath0ti@xmath0o@xmath1 pyrochlore antiferromagnet displays a transition to long - range order at @xmath176 k. @xcite the application of a magnetic field along @xmath144 $ ] ultimately destroys that order at @xmath86 , giving rise to a quantum phase transition at @xmath177 t. @xcite specific heat measurements in non - zero field have been shown useful to characterize the evolution of this system in and out of the long - range ordered phase at @xmath178 t and @xmath179 k. @xcite measurements of @xmath175 in yb@xmath0ti@xmath0o@xmath1 for @xmath180 k may well be very interesting
. this would be especially true if , once the sample dependence of @xmath45 has been understood and @xmath181 k indeed turns out to be a phase transition to a ferrimagnetically - ordered state , as suggested by some experiments and theory , but not all experiments .
@xcite thus , in anticipation of such measurements in yb@xmath0ti@xmath0o@xmath1 , we have used nlc expansion to calculate @xmath175 .
the specific heat , @xmath182 , calculated using the euler - transformed nlc expansion to order @xmath143 and the microscopic hamiltonian determined for yb@xmath0ti@xmath0o@xmath1 from inelastic neutron scattering ( ins ) @xcite is shown in fig .
[ fig : spec_heat ] .
we present results for @xmath182 for a field parallel to @xmath131 $ ] , @xmath144 $ ] , and @xmath60 $ ] from top to bottom , respectively . in all cases , calculations are presented for field strengths ranging from 0.2 t to 5 t. the trend is very similar for all three field directions .
a broad peak is observed for a field strength @xmath2 above @xmath183 2 t , with the peak position moving to lower temperatures as @xmath2 is decreased . while the nlc calculation is known to be inaccurate for temperatures less than @xmath183 1 k @xcite and for appropriately low field strengths , the calculated molar specific heat at the lowest applied magnetic fields is qualitatively consistent with that observed experimentally in zero field , @xcite displaying a sharp anomaly at very low temperaures as well as a broad shoulder near 2 k. even though , at first sight , the dependence of thermodynamic properties looks independent of field direction , there are subtle differences at low temperatures and fields , which could be taken as evidence of an ultimate spontaneous ferrimagnetic ordering with @xmath184 along [ 100 ] in zero field .
@xcite on close inspection , one finds that for fields of 1 t and lower , @xmath175 is higher at temperatures around 1 k for the [ 100 ] direction and , as a result , the entropy @xmath137 is substantially lower .
in fact , as shown in fig .
[ fig : entropy ] , the entropy along [ 100 ] appears to become vanishingly small at temperatures of order 1 k , whereas there remains a residual entropy along the other two directions .
the residual entropy remains largest along the [ 111 ] direction .
this observation is consistent with the picture of a zero - field ordering in this system , characterized by @xmath185 along one of the @xmath186 crystallographic directions .
@xcite in that case , applying a field along [ 100 ] selects one ordered state and the system merely exhibits a crossover to a paramagnetic state at a temperature of order 1 k. the entropy then displays an activated temperature dependence at low temperatures .
in contrast , for small fields along [ 110 ] two degenerate ordered states remain and for small fields along [ 111 ] , three degenerate states remain .
thus , there should continue to be a low temperature phase transition in sufficiently weak magnetic fields along either @xmath144 $ ] or @xmath60 $ ] , with only below the transition the entropy going to zero . by its nature
, nlc does not allow for our theoretical calculations to converge at low enough temperatures and magnetic fields to fully confirm this picture .
it would thus be interesting to invesgitate its validity via @xmath175 experimental measurements . to summarize , the rather large anisotropic exchange interactions at play in yb@xmath0ti@xmath0o@xmath1 lead to the development of a collective paramagnetic state at @xmath187 k with @xmath188 , being thus somewhat `` hidden '' and difficult to quantitatively describe using standard textbook ( e.g. rpa and other mean - field like ) methods .
in this material , and perhaps in other candidate quantum spin ice materials , the fit of bulk thermodynamic data such as @xmath175 and @xmath6 using nlc may provide a useful alternative .
in this paper we have compared the results from numerical linked - cluster ( nlc ) calculations of the temparature and magnetic field dependent magnetization of yb@xmath0ti@xmath0o@xmath1 with those obtained from experimental measurements on this material in a temperature range @xmath189 $ ] k and magnetic field range @xmath190 $ ] t. the nlc calculations were performed on a hamiltonian describing the interactions between pseudospins @xmath191 and characterized by effective anisotropic exchange couplings determined via inelastic neutron scattering ( ins ) measurements in the polarized paramagnetic state of yb@xmath0ti@xmath0o@xmath1 .
@xcite the overall agreement between the nlc and the experimental results were found to be excellent ( see fig . [ fig:3panels ] ) .
conversely , nlc magnetization results obtained using the exchange couplings determined from a random phase approximation analysis of the diffuse paramagnetic neutron scattering @xcite were found to be in significant disagreement with the experimental magnetization measurements .
the excellent argreement between nlc and experimental measurements of @xmath6 shows that ( i ) the proposed nearest - neighbor exchange model @xcite is quantitatively accurate to describe yb@xmath0ti@xmath0o@xmath1 , that ( ii ) high - field inelastic scattering in the polarized paramagnetic state is a reliable method for extracting effective exchange parameters and that ( iii ) long - range dipolar interactions appear to not play an important role in the energetics of the system .
the latter conclusion is perhaps a bit surprising given that long - range dipolar interactions play a key role in the physics of classical ( dipolar ) spin ices .
@xcite one obvious difference is that , relative to the @xmath89 effective exchange interaction , the nearest - neighbor contribution of the magnetostatic dipole - dipole interactions in yb@xmath0ti@xmath0o@xmath1 is approximately one order of magnitude smaller than in the classical dipolar spin ices .
@xcite at the very least , one may consider that the nearest - neighbor contribution of the dipolar interactions are incorporated in the @xmath139 effective anisotropic exchange .
then , it thus appears that the perturbative long - range ( beyond nearest - neighbor ) part of the dipolar interactions plays no dramatic role in a field greater than 0.2 t and down to 1.8 k ( this work ) or , in zero field , down to approximately 0.7 k as found in ref .
it may be that , if yb@xmath0ti@xmath0o@xmath1 does possess a ferrimagnetically ordered state with a magnetization along of the @xmath192 axes , that the sole role of dipolar interactions is to weakly renormalize the critical temperature and the level of quantum fluctuations , along with inducing ferrimagnetic domains . yet , perhaps one should not be too expedient in assuming a generic irrelevance of long - range dipolar interactions of order @xmath193 compared to @xmath89 for any candidate quantum spin ice material .
one can imagine that dipolar interactions may play an important role , possibly inducing novel phases , in a material that would , based solely on its effective anisotropic exchange couplings @xmath139 , find itself at the boundary between the various semi - classical and intrinsically quantum phases identified in mean - field lattice gauge theories of quantum spin ices .
@xcite we believe that it would be , at this time , very interesting and most useful to carry out similar studies that combine inelastic neutron scattering and thermodynamic bulk measurements for other candidate quantum spin ice materials , with yb@xmath0sn@xmath0o@xmath1 and yb@xmath0zr@xmath0o@xmath1 being obvious choices . from the lessons learned in the present work , as well as from ref .
[ ] , we anticipate that nlc will contribute to developing a quantative parametrization of microscopic models of quantum spin ices and help pave the way for an ultimate understanding of these fascinating systems .
however , for this program to be successful , the availability of large high resolution inelastic neutron scattering data sets will likely prove to be essential . in parallel , in - field single crystal measurements which properly account for demagnetization effects will also be necessary . to end on yb@xmath0ti@xmath0o@xmath1 , we believe that , notwithstanding the evidence presented here and elsewhere @xcite , it is by no means certain @xcite at this time that long - range ferrimagnetic order develops below 0.26 k @xcite in this very interesting material .
more experiments on well - characterized samples are needed to settle this question .
we acknowledge useful discussions with p. dalmas de rotier , b. javanparast , p. mcclarty , j. thompson and a. yaouanc . this work is supported in part by nsf grant number dmr-1004231 , the nserc of canada and the canada research chair program ( m.j.p.g . ,
tier 1 ) and by the perimeter institute for theoretical physics .
research at the perimeter institute is supported by the government of canada through industry canada and by the province of ontario through the ministry of economic development & innovation .
the pairwise interactions themselves could be ferromagnetic but be frustrated by strong local single - ion anisotropy with local axis that form an angle larger than 90@xmath194 , as arises in spin ice compounds . in that case , the _ effective _ interactions can be said to be antiferromagnetic .
( see ref . [ ] ) .
j. d. m. champion _
et al . _ , m. j. harris , p. c. w. holdsworth , a. s. wills , g. balakrishnan , s. t. bramwell , e. imr , t. fennell , j. s. gardner , j. lago , d. f. mcmorrow , m. orend , a. orendov , d. mck .
paul , r. i. smith , m. t. f. telling , and a. wildes , phys .
b * 68 * , 020401 ( 2003 ) .
l. savary , k. a. ross , b. d. gaulin , j. p. c. ruff , and leon balents , phys .
. lett . * 109 * , 167201 ( 2012 ) ; m. e. zhitomirsky , m. v. gvozdikova , p. c. w. holdsworth , and r. moessner , phys
* 109 * , 077204 ( 2012 ) .
j. s. gardner , s. r. dunsiger , b. d. gaulin , m. j. p. gingras , j. e. greedan , r. f. kiefl , m. d. lumsden , w. a. macfarlane , n. p. raju , j. e. sonier , i. swainson , and z. tun , phys .
lett . * 82 * , 1012 ( 1997 ) .
s. t. bramwell , m. j. harris , b. c. den hertog , m. j. p. gingras , j. s. gardner , d. f. mcmorrow , a. r. wildes , a. l. cornelius , j. d. m. champion , r. g. melko , and t. fennell , phys .
lett . * 87 * , 047205 ( 2001 ) . the pyrochlore lattice can be described by a fcc space lattice with a four - site ( tetrahedron ) sublattice basis .
the @xmath59 quantization axis is along the local [ 111 ] cubic direction at each of the four sublattice site , and @xmath61 ( or equivalently , @xmath195 and @xmath196 ) refers to the two orthogonal local directions .
the `` transverse '' ( to the local [ 111 ] direction @xcite ) @xmath197 matrix elements determine the zeeman energy splitting for an hypothetical magnetic field applied perpendicular to the local @xmath60 $ ] at a given site and which determines the @xmath62 component of the @xmath52-tensor .
h. r. molavian , m. j. p. gingras , and b. canals , phys .
rev . lett . * 98 * , 157204 ( 2007 ) . h. r. molavian , p. a. mcclarty , and m. j. p. gingras , arxiv:0912.2957 .
s. onoda and y. tanaka , phys .
b * 83 * , 094411 ( 2011 ) .
j. a. hodges , p. bonville , a. forget , a. yaouanc , p. dalmas de rotier , g. andr , m. rams , k. krlas , c. ritter , p. c. m. gubbens , c. t. kaiser , p. j. c. king , and c. baines , phys . rev
. lett . * 88 * , 077204 ( 2002 ) .
y. yasui , m. soda , s. iikubo , m. ito , m. sato , n. hamaguchi , t. matsushita , n. wada , t. takeuchi , n. aso , and k. kakurai , j. phys .
japan * 72 * , 3014 ( 2003 ) .
j. s. gardner , g. ehlers , n. rosov , r. w. erwin , and c. petrovic , phys .
b * 70 * , 180404 ( 2004 ) .
chang , s. onoda , y. su , y .- j .
tsuei , y. yasui , k. kakurai , and m. r. lees , nat .
* 3 * , 992 ( 2012 ) .
unlike tb@xmath0(ti , sn)@xmath0o@xmath1 , @xcite the energy gap to the first excited crystal field state in pr@xmath0(sn , zr)@xmath0o@xmath1 and yb@xmath0ti@xmath0o@xmath1 is large compared to the interactions , @xmath16 .
this allows one to consider a spin-1/2 effective low - energy theory that contains a one - to - one mapping of the @xmath198 matrix elements ( see discussion in supplementary material of ref . [ ] ) without concerns about interaction - induced admixing between the crystal - field levels ( see refs .
[ ] ) j. d. thompson , m.sc . thesis , university of waterloo ( 2011 ) . some comments in this document as per the merit of using rpa to fit the diffuse neutron scattering of yb@xmath0ti@xmath0o@xmath1 at 1.4 k as opposed to the other data set at 9.1 k are of relevance to the present discussion . citing an extract from page 49 : `` we choose to use the data collected at @xmath160 k rather than the data collected at @xmath199 k as the signal to noise ratio of this data is higher . in principal [ sic ]
the @xmath200 k data would have better suited our purpose , as at this temperature yb@xmath0ti@xmath0o@xmath1 is situated much more firmly in the paramagnetic phase , avoiding any potential problems with yb@xmath0ti@xmath0o@xmath1 being in a collective paramagnetic phase , and correlation effects that appear due to the presence of a nearby phase transition . '' the quantitative failure of a quantum spin ice hamiltonian @xmath54 with anisotropic exchange couplings obtained from a random phase approximation ( rpa ) fit of diffused neutron scattering , as reported in both ref .
[ ] and ref .
[ ] , to describe the zero field magnetic specific heat had already been pointed out in ref .
[ ] .
we note , however , that no magnetic bragg peaks were found in the neutron diffraction pattern of the powder sample studied in ref .
[ ] , but which nevertheless displays a fairly sharp specific heat feature at @xmath44 k ( see fig . 4 in p. dalmas de rotier , v. glazkov , c. marin , a. yaouanc , p.c.m .
gubbens , s. sakarya , p. bonville , a. amato , c. baines , and p.j.c .
king , physica b * 374 - 375 * , 145 ( 2005 ) , albeit not as sharp as that found in another powder sample studied in ref . | the thermodynamic properties of the pyrochlore yb@xmath0ti@xmath0o@xmath1 material are calculated using the numerical linked - cluster ( nlc ) calculation method for an effective anisotropic - exchange spin-1/2 hamiltonian with parameters recently determined by fitting the neutron scattering spin wave data obtained at high magnetic field @xmath2 .
magnetization , @xmath3 , as a function of temperature @xmath4 and for different magnetic fields @xmath2 applied along the three high symmetry directions [ 100 ] , [ 110 ] and [ 111 ] , are compared with experimental measurements on the material for temperature @xmath5 k. the excellent agreement between experimentally measured and calculated @xmath6 over the entire temperature and magnetic field range considered provides strong quantitative validation of the effective hamiltonian .
it also confirms that fitting the high - field neutron spin wave spectra in the polarized paramagnetic state is an excellent method for determining the microscopic exchange constants of rare - earth insulating magnets that are described by an effective spin-1/2 hamiltonian .
finally , we present results which demonstrate that a recent analysis of the polarized neutron scattering intensity of yb@xmath0ti@xmath0o@xmath1 using a random phase approximation ( rpa ) method [ chang _ et al .
_ , nat . comm .
* 3 * , 992 ( 2012 ) ] does not provide a good description of @xmath6 for @xmath7 k , that is in the entire temperature regime where correlations become non - negligible . |
one of the most impressive occurrences of painlev transcendents in non - linear mathematical physics stems from their applicability in random matrix theory . in this
field painlev functions describe , for instance ( cf .
@xcite ) , eigenvalue distribution functions for classical finite @xmath10 ensembles , they appear in the description of universal distribution functions in the large @xmath10 limit and are also used in the computation of gap probabilities in the large @xmath10 limit .
one concrete example for the last two cases is given by the celebrated tracy - widom distribution @xcite : the distribution function for the largest eigenvalue @xmath11 of a @xmath12 random matrix drawn from the gaussian unitary ensemble ( gue ) in the large @xmath10 limit equals @xmath13,\ \
\ x\in\mathbb{r},\ ] ] where @xmath14 is the hastings - mcleod solution of the second painlev equation , @xmath15 this special solution was first analyzed in the works of hastings and mcleod @xcite who showed that the boundary value problem has a unique , monotonically decreasing , smooth solution .
in addition , @xmath16 although @xmath17 in is written as a distribution function , it is equivalent to a gap probability for the rescaled eigenvalues @xmath18 , and as such , expressible as fredholm determinant .
more precisely , @xmath19 where @xmath9 is the trace - class operator in @xmath20 with kernel @xmath21 and @xmath22 the classical airy function . besides the gap probability other spectral properties of large random matrices from the gue can also be computed through painlev transcendents , for instance for @xmath23 we have ( cf .
@xcite ) @xmath24\bigg|_{\gamma=1}\ ] ] and @xmath25,\ \ \gamma\geq 0.\ ] ] here , @xmath26 is the ablowitz - segur solution of the second painlev equation , @xmath27 this one - parameter family of solutions has different analytical and asymptotical properties depending on the values of @xmath28 : namely , for the values of @xmath29 fixed , the boundary value problem has a unique , bounded , smooth solution with oscillatory behavior as @xmath3 , compare below . on the other hand ,
if @xmath30 , smoothness is destroyed at finite @xmath31 and the solution blows up , see . historically , solutions of the boundary value problem for fixed @xmath29 were first analyzed , both analytically and asymptotically , in the late 1970 s , cf . @xcite . in particular for
the bounded solutions , ablowitz and segur solved the important connection problem , i.e. determine the complete asymptotic description of @xmath32 as @xmath3 provided the same description is given as @xmath33 ( or vice versa ) .
the same problem was subsequently also solved for the unbounded solution @xmath34 of , see @xcite . however , the singular asymptotic structure for @xmath30 as @xmath3 remained unknown until the work of kapaev @xcite in 1992 .
our modest goal in this paper is to place ablowitz - segur ( for @xmath29 or @xmath35 ) and hastings - mcleod solutions on equal asymptotic footing as @xmath3 .
this problem is of interest to the asymptotics of the fredholm determinant @xmath36 as @xmath3 and thus to the asymptotics of the tracy - widom distribution itself . for the gap probability let us use the refined asymptotic behavior ( compare remark [ deirem ] below ) ,
@xmath37 and the identity @xmath38 from which we learn in that @xmath39|x|^{-\frac{1}{8}}c_0\left(1+\mathcal{o}\left(x^{-3}\right)\right),\ \ \ x\rightarrow-\infty\ ] ] with some constant @xmath40 .
the constant factor was first derived by deift , its and krasovsky in @xcite using scaling limits for a finite laguerre ensemble combined with universality results for the airy kernel , @xmath41,\ ] ] with @xmath42 the riemann - zeta function . shortly afterwards
baik , buckingham and difranco @xcite gave another derivation of , based on exact integral representations of @xmath17 which , opposed to , involve an integration from @xmath43 to @xmath31 .
in either approach the regularization of the integral in is important and thus the asymptotics relevant as they are needed in the evaluation of a large gap probability in random matrix theory . the very same asymptotic question for the behavior of @xmath44 with fixed @xmath45 is still open and our focus does not lie on this interesting problem .
however , considering the different qualitative behaviors of @xmath32 as @xmath3 ( see , below for further detail ) , @xmath46 @xmath47 @xmath48 it becomes evident that @xmath7 experiences a phase transition near @xmath49 as @xmath3 : for @xmath29 the determinant will be strictly positive , decaying exponentially fast but with a slower decay rate than . opposed to that , for @xmath35 the determinant will display oscillations with decreasing amplitudes and its zeros are accumulating at @xmath50 .
this qualitative change leads to a challenging problem : determine the asymptotic behavior of @xmath51=\det\big(i-\gamma k_{\textnormal{ai}}\big)\big|_{l^2(x,\infty)}\ \ \ \ \textnormal{as}\ \ x\rightarrow-\infty\ \ \
\textnormal{and simultaneously}\ \ \gamma\rightarrow 1.\ ] ] as a direct application of theorem [ res:4 ] below on the transition asymptotics of @xmath32 we will prove a short result which shows that to leading order also describes one particular case of the transition asymptotics for @xmath44 .
[ nice ] as @xmath3 and @xmath52 , with @xmath53 as in , @xmath54|x|^{-\frac{1}{8}}c_0\big(1+o(1)\big),\ \ ] ] uniformly for @xmath55 for the values of @xmath31 and @xmath28 such that @xmath56 , our results in theorem [ bet:1 ] and [ res:5 ] do in general not allow us to state an analogous expansion ; the integration in simply becomes too complicated .
still we can make a qualitative prediction _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the transition asymptotics of @xmath7 as @xmath8 and @xmath56 is modeled by quasi - periodic functions , to leading order .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ recentpap ] we would like to point out that an asymptotic analysis similar to was recently carried out for the case of another prominent fredholm determinant in random matrix theory , @xmath57 it is well known @xcite that @xmath58 equals a gap probability : the probability of finding no eigenvalues in the interval @xmath59 for a random matrix chosen from the gue , in the bulk scaling limit with mean spacing one .
similar to the airy kernel determinant , @xmath60 admits a representation in terms of painlev functions , this time involving a special solution of the fifth painlev equation @xcite .
building on previous work of dyson @xcite , the analysis of @xmath60 as @xmath61 , was essentially completed in @xcite , however without using painlev asymptotic analysis .
this lead to results completely analogous to corollary [ nice ] and the qualitative prediction , compare theorems @xmath62 and @xmath63 as well as corollary @xmath64 in @xcite .
we now move ahead and present the setup for our results on the transition asymptotics as @xmath3 between @xmath32 and @xmath34 .
opposed to introducing painlev ii functions as solutions to boundary value problems as in and , we choose to follow a riemann - hilbert point of view . in this approach
@xcite solutions @xmath65 to @xmath66 are parametrized through the monodromy data of an associated linear system of ordinary differential equations in the complex plane .
the details are as follows , compare @xcite : given six complex numbers @xmath67 which satisfy the relations @xmath68 we introduce the triangular matrices @xmath69 and the sectors @xmath70 with oriented boundary rays @xmath71 as shown in figure [ figure1 ] below .
consider now the following riemann - hilbert problem ( rhp ) .
[ masterrhp ] determine the piecewise analytic @xmath72 matrix - valued function @xmath73 such that * @xmath74 is analytic for @xmath75 and for every @xmath76 , the function @xmath77 has a continuous extension on the closure @xmath78 .
* the boundary values @xmath79 ( resp .
@xmath80 ) from the left ( resp .
right ) side of the oriented contour @xmath81 are related via the jump condition @xmath82 where @xmath83 *
as @xmath84 tends to infinity , the function @xmath74 is normalized as follows , @xmath85 the first two of the relations in ensure consistency of the jump conditions with the continuity of @xmath86 on @xmath78 .
indeed , conditions can be rewritten as @xmath87 and thus @xmath88 but every solution @xmath74 of the rhp [ masterrhp ] is invertible since @xmath89 is an entire function normalized to unity at infinity . hence yields upon cancellation of @xmath90 the cyclic constraint @xmath91 which is equivalent to the first two conditions in .
we also mention that by triangularity of @xmath92 , we have @xmath93 as @xmath94 along the rays @xmath81 , i.e. is consistent with .
the connection between the rhp [ masterrhp ] and is as follows .
it is known @xcite that for any set of parameters @xmath95 satisfying the rhp for @xmath74 is meromorphically ( with respect to @xmath31 ) solvable .
moreover its solution determines the painlev ii transcendent via @xmath96,\ ] ] and in addition we have @xmath97 .
conversely every real for real @xmath31 solution of has a unique riemann - hilbert representation for suitable @xmath98 and we therefore adopt the notation @xmath99 , indicating the parametrization of solutions to pii equation in terms of the data @xmath98 . [
connect ] the previously considered cases and are special cases of , @xmath100 the data @xmath98 is equal to the monodromy data of the associated linear @xmath72 matrix ode @xmath101\psi\ ] ] where @xmath102 satisfies . in the language of classical monodromy theory of differential equations ( cf .
@xcite ) the entire functions @xmath103 and @xmath104 form the seven canonical solutions of the ode and the triangular matrices @xmath92 in are the stokes matrices characterizing the stokes phenomenon of the irregular singular point @xmath105 .
the matrices @xmath92 are @xmath31-independent and thus the second painlev transcendent @xmath106 describes the isomonodromy deformations of system with respect to the ( isomonodromic ) deformation parameter @xmath31 .
as we are interested in the asymptotic behavior of real solutions @xmath107 as @xmath3 , it will be useful to recall the following three different cases which have been singled out over the past @xmath108 years . 1 .
if @xmath4 is fixed , then , as @xmath3 , @xmath109 with @xmath110 and @xmath111 is the euler gamma function . expansion appeared first in @xcite and was partially proven by hastings , mcleod and clarkson in @xcite using gelfand - levitan type integral equations . in the late @xmath112 s ,
its , kapaev , suleimanov and kitaev @xcite , using the isomondromy method " , obtained the full leading term .
another derivation of was given by deift and zhou @xcite in the mid @xmath113 s based on a direct asymptotic analysis of the rhp [ masterrhp ] with the help of the deift - zhou nonlinear steepest descent method @xcite .
we refer the reader to the monograph @xcite for a detailed exposition of the methods used in the rigorous derivation of . 1 .
if @xmath5 , then with @xmath114 ( compare here ) , as @xmath3 , @xmath115 [ deirem ] the leading order expansion in is originally due to hastings and mcleod @xcite but was later on also derived by deift and zhou @xcite , who extended to a full asymptotic series in which all coefficients can be computed recursively . 1 .
if @xmath6 is fixed , then , as @xmath3 , @xmath116 with @xmath117 where the error terms are uniform outside some neighborhoods of the singularities of the trigonometric function appearing in the denominator of the leading order .
the singular asymptotic formula was first obtained by kapaev in @xcite through the isomonodromy method . in @xcite ,
the authors rederived based on nonlinear steepest descent techniques applied to the rhp [ masterrhp ] . in this paper
we derive new asymptotic expansions of @xmath107 as @xmath3 with the help of a nonlinear steepest descent analysis applied to rhp [ masterrhp ] .
these expansions explain how the leading oscillatory power - like decay in case ( a ) is transformed to the leading power - like growth in case ( b ) as @xmath118 .
we shall refer to the transition between regimes ( a ) and ( b ) as _ regular transition_. in addition , we will also explain the transformation of the leading singular oscillatory power - like growth in case ( c ) to the leading power - like growth asymptotics in ( b ) as @xmath119 .
it is natural to refer to the transition between ( b ) and ( c ) as _ singular transition_. the results are as follows .
let @xmath120 and @xmath121 be implicitly determined from the transcendental equation @xmath122,\ ] ] in which @xmath123 and @xmath124 are standard complete elliptic integrals @xmath125 with modulus @xmath121 and complementary modulus @xmath126 .
it is shown in proposition [ modex ] that for any @xmath127 , equation determines @xmath128 uniquely .
in addition , set @xmath129;\hspace{1cm}\tau=\tau(\varkappa)=2{\mathrm{i}}\frac{k}{k'},\ ] ] and let @xmath130 denote the jacobi theta functions with nome @xmath131 .
the theta functions in turn determine the jacobi elliptic function @xmath132 and we list properties of the elliptic integrals and theta functions which are relevant to our analysis in appendices [ appa ] and [ app : theta ] .
the main result concerning the regular transition asymptotics of @xmath107 as @xmath3 is formulated in the following theorem .
[ bet:1 ] for any fixed @xmath133 , there exist positive constants @xmath134 and @xmath135 such that @xmath136 with @xmath137 and @xmath138 theorem [ bet:1 ] describes the transition of the leading order asymptotics between case ( a ) for fixed @xmath4 given in and case ( b ) for @xmath5 in . indeed , using the limiting behavior of @xmath139 and @xmath140 in the jacobi elliptic function as @xmath141 and @xmath142 ( compare corollary [ cor1 ] below ) , we obtain directly [ mat ] as @xmath3 , @xmath143 uniformly for @xmath144 . in addition , as @xmath145 , @xmath146 uniformly for @xmath147 .
our second result addresses the asymptotic regime @xmath145 for @xmath148 with fixed @xmath149 . in this case
the leading behavior is already of type and therefore connects the boutroux behavior , completely to the hastings - mcleod asymptotics for @xmath5 .
[ res:4 ] given @xmath149 , there exist positive constants @xmath150 and @xmath151 such that @xmath152 with @xmath153 theorems [ bet:1 ] and [ res:4 ] describe the behavior of @xmath107 as @xmath3 for the values of @xmath154 in @xmath155\bigsqcup\left[\frac{2}{3}\sqrt{2}-\frac{f_2}{t},\infty\right)\subset(0,\infty).\ ] ] the corresponding regions in the double scaling diagram , i.e. _ boutroux region _ for theorem [ bet:1 ] and _ hastings - mcleod region _ for theorem [ res:4 ] , are shown in figure [ diareg ] .
another region , the _ stokes region _ , corresponds to @xmath156 we will not provide a full asymptotic description of @xmath107 for the latter values of @xmath154 as the nonlinear steepest descent analysis becomes increasingly difficult .
instead we focus on the scale @xmath157 which eventually motivates the term _ stokes region _ : for @xmath158 the leading order behavior of @xmath107 turns out to be unchanged from , however once we cross the _ stokes lines _
@xmath159 once we go deeper into , additional @xmath160 terms will contribute to the leading behavior .
this is in sharp contrast to theorems [ bet:1 ] and [ res:4 ] where the fixed choices of @xmath161 and @xmath149 had no effect on the leading orders . in the present paper
, we shall prove the following estimation .
[ res:5 ] given @xmath162 there exists positive constants @xmath163 and @xmath164 such that @xmath165 with @xmath166 this rigorously clarifies the appearance of the first stokes line @xmath167 : fix @xmath158 in , then @xmath168 and subsequently by geometric progression , as @xmath145 , @xmath169 in this case we also work with the double scaling parameter @xmath170 but notice that opposed to , we have now @xmath171 furthermore we choose an upper constraint @xmath172 which is important to our analysis below , but at the same time is also consistent with the anticipated transition asymptotics for @xmath119 . , the leading order behavior of @xmath107 as @xmath3 is already known through . ] the module @xmath121 is again determined via and we require @xmath173 and @xmath174 as in .
however , instead of the jacobi elliptic function @xmath175 in , we need @xmath176 the analogue of theorem [ bet:1 ] for the singular transition is contained in the following theorem .
[ bet:2 ] for any fixed @xmath177 , there exist positive constants @xmath178 and @xmath179 such that @xmath180 with @xmath181 provided @xmath182 is uniformly bounded away from the discrete set @xmath183 on the other hand , for any fixed @xmath161 , there exist constants @xmath184 and @xmath185 such that holds true with @xmath186 and no further constraint placed on @xmath182 . here , theorem [ bet:2 ] describes the transition to leading order between case ( c ) for fixed @xmath6 in and case ( b ) with @xmath5 in
. indeed , through the limiting behavior of @xmath187 and @xmath188 as @xmath141 and @xmath142 in we obtain [ singmat ] as @xmath3 , @xmath189 uniformly for @xmath144 and away from the zeros of the trigonometric function appearing in the denominator of the leading order .
in addition , as @xmath190 , @xmath191 uniformly for @xmath147 . our last result is the direct analogue of theorem [ res:4 ] to the singular transition .
[ res:9 ] given @xmath149 , there exist positive constants
@xmath150 and @xmath151 such that @xmath192 with @xmath193 the regions in the @xmath194-plane captured by theorems [ bet:2 ] and [ res:9 ] are again shown in figure [ diareg ] and we use the same terminology , _ boutroux _ and _ hastings - mcleod_. we do not address the stokes region for the singular case , a full analysis for all stokes lines in both regular and singular transition will be postponed to a forthcoming publication .
the transition asymptotics are derived through an application of the nonlinear steepest descent method @xcite to the initial rhp [ masterrhp ] .
this analysis depends crucially on the values of the double scaling parameter @xmath195 for instance , the proof of theorem [ bet:1 ] consists of several steps . 1 .
first , the master rhp [ masterrhp ] is analyzed with the constraint @xmath196 in place .
all details are worked out in sections [ sec : reg ] and [ relax ] , which include a @xmath197-function transformation , the construction of local model problems via jacobi theta and airy functions and iterative solutions of singular integral equations .
this part resembles in some of its aspects the approach carried out in @xcite .
2 . second , we analyze rhp [ masterrhp ] subject to the constraint @xmath198 with @xmath199 .
the techniques used now are very different from ( 1 ) , in fact they are close to the ones we would use in the asymptotic analysis of rhp [ masterrhp ] for fixed @xmath4 , i.e. no need for a @xmath197-function transformation but different model problems involving parabolic cylinder functions .
we work out the necessary details in section [ flower ] and combining error estimations from ( 1 ) and ( 2 ) , we obtain with estimation . in addition , expansion follows as well .
third , estimations and are derived by applying modular transformations to the jacobi theta functions used in ( 1 ) and introducing a new model function in a vicinity of the origin .
the details are presented in section [ upextension ] .
after that , we address the regime @xmath200 in theorem [ res:4 ] . here
the nonlinear steepest techniques are again different from the ones used in the derivation of theorem [ bet:1 ] , we use a new @xmath197-function and different model problems , compare section [ near ] .
it is worth mentioning that these techniques resemble the ones used in the derivation of with @xmath5 .
our final result for the regular transition is contained in theorem [ res:5 ] which we derive in section [ stokes ] .
the appearance of stokes lines is already observed earlier in section [ upextension ] , see remark [ stokefirst ] , however we give a quantitative analysis of this phenomenon only subject to the constraint @xmath201 in @xcite , theorem @xmath63 , the authors also discovered a stokes region while analyzing @xmath60 as @xmath202 , recall remark [ recentpap ] . starting from the known large @xmath98 asymptotics of the eigenvalues of @xmath203
all stokes lines were classified by an application of lidskii s theorem to the fredholm determinant @xmath60 .
we will now argue that we can reverse this approach and derive asymptotics of the eigenvalues @xmath204 of @xmath9 via theorem [ res:5 ] and identity .
first , we state a slight improvement of corollary [ nice ] .
[ eig0 ] as @xmath3 and @xmath52 , with @xmath53 as in and @xmath205 in , @xmath206|x|^{-\frac{1}{8}}b(\gamma)\left(1+\frac{1}{2^{\frac{9}{4}}\sqrt{\pi}}\frac{e^{\sigma t}}{t^{\frac{1}{2}}}\right)\left(1+\mathcal{o}\left(t^{-\min\{\frac{2}{3},\frac{3}{2}-f_3\}}\right)\right),\ ] ] uniformly for @xmath207 where @xmath208 is positive , bounded in @xmath28 , such that @xmath209 this result is a direct consequence of theorem [ res:5 ] and we state its proof in section [ eigproof ] . compared to corollary [ nice ] , an additional contribution to the leading order appears in and the extra factor is in general not close to unity .
in other words , also the fredholm determinant displays an asymptotic stokes phenomenon , and we have identified the first stokes line , @xmath210 now back to the spectrum of @xmath9 , we let @xmath204 denote the eigenvalues of the trace class operator @xmath211 . these have been , for instance , analyzed in @xcite , were it was proven that the spectrum of @xmath9 is simple , we have @xmath212 and for fixed @xmath213 , @xmath214 in fact , tracy and widom in loc .
. also gave an asymptotic formula , @xmath215 but the derivation is not fully rigorous .
we will now outline a proof for the stated behavior , using corollary [ eig0 ] . choose @xmath216 such that @xmath217 apply lidskii s theorem , @xmath218 with @xmath219 and @xmath220 is the projection onto the eigenspace of @xmath9 with eigenvalues @xmath221 .
this allows us to compare , to , i.e. as @xmath145 subject to , @xmath222 we multiply though with the second summand in the first factor in the right hand side of the last estimation , @xmath223 and here , both factors in the left hand side of are positive .
but for @xmath216 such that @xmath224 all summands in the right hand side of are bounded .
thus , by positivity , all summands in the left hand side of have to be bounded .
so , using also , @xmath225 since @xmath226 , we deduce from the first estimation , @xmath227 which is consistent with for @xmath228 .
in fact , allows us to deduce the inequality , @xmath229 but in order to achieve equality our analysis in the stokes region has to be extended ; so far only gives @xmath230 which is not sufficient yet to deduce equality in . summarizing , provided theorem [ res:5 ]
is extended to the full stokes region and simultaneously also corollary [ eig0 ] , the results on the transition asymptotics of @xmath107 would enable us to derive rigorously .
the derivation of theorem [ bet:2 ] follows largely its regular counterpart . 1 .
the master rhp [ masterrhp ] is analyzed asymptotically subject to the constraint @xmath196 in sections [ singsec:1 ] and [ singsec:2 ] .
opposed to the regular case @xmath231 we require a different outer parametrix which has the singular structure and corresponding exceptional set encoded .
2 . after that
, we address the scale @xmath198 following @xcite .
the details are summarized in section [ singsec:3 ] .
3 . again
, by modular transformations , plus an additional argument related to the singular structure , we complete the proof of theorem [ bet:2 ] in section [ singsec:4 ] and in addition obtain corollary [ singmat ] .
section [ singsec:5 ] concludes the manuscript by deriving theorem [ res:9 ] and we list a few identities for complete elliptic integrals and jacobi theta functions in appendices [ appa ] and [ app : theta ] .
we start with the scaling transformation @xmath232 and are lead from the initial rhp [ masterrhp ] to a rhp for the function @xmath233 .
this problem is formulated on the same jump contour @xmath234 with jump @xmath235 and jump exponent @xmath236 next we deform the original jump contours , to obtain a rhp for a function @xmath237 as shown in figure [ figure2 ] - at this stage the endpoints @xmath238 are yet to be determined .
the function @xmath237 has the following analytical properties : * @xmath237 is analytic for @xmath239\cup\bigcup_1 ^ 6\gamma_k)$ ] * we have the following jumps , compare figure [ figure2 ] , @xmath240;\\ z_+(\lambda ) & = z_-(\lambda)e^{-t\vartheta(\lambda)\sigma_3}\sigma_2 s_3s_4s_5\sigma_2e^{t\vartheta(\lambda)\sigma_3 } , \ \
\lambda\in[0,\lambda^{\ast}],\end{aligned}\ ] ] where in fact @xmath241 * as @xmath94 , the function @xmath237 is normalized as @xmath242 next , referring to the domains @xmath243 shown in figure [ figure4 ] , we set @xmath244 and are lead to a rhp for @xmath245 with jumps ( compare figure [ figure4 ] for the jump contours ) @xmath246 and @xmath247 for @xmath248 $ ] and @xmath249 $ ] in the last two cases . notice that @xmath250 the presence of the factors @xmath251 on the diagonal motivates the use of a @xmath197-function in order to address the transition asymptotics @xmath252 .
subsequently we shall first work out the necessary details in case of the regular transition .
we introduce the double scaling parameter @xmath255 and keep @xmath256 $ ] with @xmath254 fixed throughout . based on the cyclic constraints the parameter @xmath154 allows us to derive expansions of the stokes multipliers in terms of @xmath154 which are used later on . first observe from and the exact identity @xmath257 .
second , by definition , @xmath258\rightarrow[0,\pi],\ ] ] and we obtain for @xmath259 , as @xmath260 with @xmath261 , @xmath262 and @xmath263 uniformly for @xmath256 $ ] with @xmath254 fixed .
let us now move ahead and introduce one of the key ingredients in the nonlinear steepest descent analysis , the @xmath197-function .
set @xmath264 where @xmath121 is determined implicitly in , @xmath265 = \left(\frac{2}{1+\k^2}\right)^{\frac{3}{2}}\int_{\k}^1\sqrt{(1-\mu^2)(\mu^2-\k^2)}\,{{\mathrm d}}\mu.\ ] ] the following proposition addresses the bijective correspondence between @xmath127 and @xmath121 .
[ modex ] the modulus @xmath266 is uniquely determined via in case @xmath127 .
moreover , as @xmath141 , @xmath267 and , as @xmath142 , with @xmath268 , @xmath269 consider the function @xmath270,\ ] ] where @xmath271 with the help of standard expansions for the complete elliptic integrals as @xmath272 and @xmath273 ( compare appendix [ appa ] ) , we get @xmath274,\ \ \
\k\downarrow 0,\label{i : k1}\\ \mathrm{i}(\k)&=&\frac{\pi}{4}(1-\k)^2\left[1+(1-\k)+\frac{11}{32}(1-\k)^2+\mathcal{o}\left((1-\k)^3\right)\right],\ \ \
\k\uparrow 1.\label{i : k2}\end{aligned}\ ] ] thus @xmath275 but @xmath276 is real analytic in a neighborhood of @xmath277,\varkappa\in[0,\infty))$ ] with the first partial derivatives equal to @xmath278>0,\\ \mathrm{f}_{\varkappa}(\k,\varkappa)&=&1.\end{aligned}\ ] ] hence the implicit function theorem guarantees existence of a unique real analytic solution @xmath128 of the equation @xmath279 near the point @xmath280 .
we use and to derive and . besides the expansions and for @xmath128 itself
, we will later on also require expansions of the frequency @xmath173 and module @xmath174 introduced in .
these follow directly from and and are summarized in the corollary below .
[ cor1 ] as @xmath141 , @xmath281 secondly , as @xmath282 , @xmath283 and in the same limit also @xmath284 we now define the @xmath197-function , @xmath285\ ] ] where @xmath286 and the contour of integration is chosen in the simply connected domain @xmath287 $ ] .
moreover , we fix @xmath288 and thus , @xmath289 is single - valued and analytic in @xmath287 $ ] . as @xmath290 tends to infinity @xmath291 where @xmath292{{\mathrm d}}\mu.\ ]
] further steps in the analysis require certain analytical properties of @xmath289 .
[ gprop ] for @xmath293 introduce @xmath294 the functions @xmath295 and @xmath296 are real - valued on the real line , in fact @xmath297 moreover @xmath298 we first go back to the rhp for @xmath245 as defined in with jump contour @xmath299 shown in figure [ figure4 ] .
now fix the endpoint @xmath300 and employ the following transformation , @xmath301 \cup\bigcup_{k=1}^6 \gamma_k\cup\hat{\gamma}_2\cup\hat{\gamma}_5\cup\tilde{\gamma}_2\cup\tilde{\gamma}_5\right),\ ] ] with @xmath302 as in and @xmath303 in .
since all jumps in the @xmath304-rhp display the structure @xmath305 we are lead to the following rhp determine the @xmath72 matrix valued function @xmath306 such that * @xmath307 is analytic for @xmath308 where the jump contour @xmath309 is identical to the contour @xmath299 shown in figure [ figure4 ] . *
the function @xmath307 satisfies the jump condition @xmath310 * as @xmath94 , @xmath311 recall at this point : for @xmath312 we have thus @xmath313 with @xmath154 from .
but ( compare proposition [ gprop ] ) @xmath314 and @xmath315 in the right slit .
hence , using also the expansions , , @xmath316,\ \ 0<\delta<\frac{1}{3}\sqrt{2}\ ] ] uniformly in @xmath84 chosen from any compact subset of the right slit @xmath317 . in the left slit @xmath318 a similar situation occurs , @xmath319 but here @xmath320 and also @xmath321 hence combined with , , @xmath322,\ 0<\delta<\frac{1}{3}\sqrt{2}\ ] ] uniformly in @xmath84 chosen from any compact subset of the left slit @xmath318 .
next , we consider the jump contours which extend to infinity , i.e. @xmath323 along these contours the jumps in the @xmath324-rhp are given by @xmath325 hence , by triangularity and the sign chart of @xmath326 ( compare figure [ figure6 ] ) , the jumps on the infinite contours approach the identity matrix exponentially fast in the limit @xmath145 with @xmath256,\delta>0 $ ] , provided we stay away from the endpoints @xmath327 and the origin @xmath328 . in the remaining gap @xmath329
two cases need to be distinguished : first for @xmath330 , @xmath331 and secondly for @xmath332 , @xmath333 in and all off - diagonal entries are fast oscillating as @xmath3 , see proposition [ gprop ] .
we now transform this behavior on @xmath334 to exponential decay with the help of contour deformations ( _ opening of lens _ ) tailored to the factorizations written in and .
the key to this explicit transformation is the following proposition .
[ openup ] introduce for @xmath335 the functions @xmath336 with @xmath337 as in proposition [ gprop ] .
then @xmath338 admits local analytical continuation into a neighborhood of the gap @xmath334 into the upper half - plane such that @xmath339 similarly , the function @xmath340 admits local analytical continuation into a neighborhood of the gap @xmath334 into the lower half - plane such that @xmath341 we follow the standard line of argument .
first @xmath342 and thus @xmath343 which implies the claim via the cauchy - riemann equations .
suppose @xmath344 denotes the lens shaped regions shown in figure [ figure6 ] . with the help of the analytical continuations of @xmath338 and @xmath340 as discussed in proposition [ openup ]
, we set @xmath345 so that @xmath346 solves the following rhp [ openrhp ] determine the @xmath72 piecewise analytic function @xmath346 such that * @xmath346 is analytic for @xmath347\cup[m , m]\cup\gamma_1^{\pm}\cup\gamma_2^{\pm}\cup\sigma_{s_{\infty}}\right)$ ] * we have the following jump behavior , with orientation as indicated in figure [ figure6 ] , @xmath348 * as @xmath94 , @xmath349 after employing the explicit transformation , it is now time to focus on the local model problems near the points @xmath327 and @xmath328 as well as on the slit segment @xmath350 .
the rhp associated to the outer parametrix is motivated by , and thus consists in [ oup:2 ] find a @xmath72 matrix - valued piecewise analytic function @xmath351 such that * @xmath352 is analytic for @xmath353 * along the branch cuts , with orientation as in , @xmath354 * @xmath352 is square integrable on @xmath355 * as @xmath94 , @xmath356 observe that the function @xmath357 solves a similar rhp as the one posed for @xmath352 , but with jumps on @xmath358 given by @xmath359 this is precisely the same jump behavior which appeared in @xcite , section @xmath360 .
the solution method is therefore analogous , we only summarize the relevant steps and refer to @xcite for further details .
the solution to rhp [ oup:2 ] is derived in terms of jacobi theta functions defined on the elliptic curve @xmath361 of genus one .
we view @xmath362 as two sheeted covering of the riemann sphere , glued together in the standard way . for definitness ,
let @xmath363 as @xmath364 on the first sheet and @xmath365 in the similar limit on the second sheet .
we fix a homology basis for @xmath362 as indicated in figure [ figure5 ] , and let @xmath366^{-1 } = \frac{{\mathrm{i}}}{2}\frac{m}{k'}\ ] ] be the unique holomorphic one form on @xmath362 with normalization @xmath367 and @xmath368-period ( compare ) @xmath369 also , define the abel ( type ) map @xmath370\rightarrow\mathbb{c},\ \
z\mapsto u(z)=\int_m^z\omega = u(\infty)-\frac{c}{z}+\mathcal{o}\left(z^{-3}\right),\ z\rightarrow\infty\ ] ] and collect the following properties [ pout:1 ] the abelian integral @xmath371 is single - valued and analytic for @xmath372 $ ] , we have in addition @xmath373 and @xmath374 equation is a simple identity between complete elliptic integrals , @xmath375 our construction requires furthermore the functions @xmath376 defined and analytic for @xmath372 $ ] such that @xmath377 for @xmath378 . to complete the derivation , we use the jacobi theta function @xmath379,\ \ z\in\mathbb{c},\ ] ] and define @xmath380 where ( see ) @xmath381 with this , a solution to rhp [ oup:2 ] is given in the next proposition .
[ pout:3 ] the function @xmath382 is single - valued and analytic in @xmath383 .
its jumps are stated in and , furthermore , as @xmath94 , @xmath384 our next move focuses on the construction of model functions near the branch points and the origin . near the origin
, we first observe that @xmath385 thus the jumps on the quasi - vertical segments @xmath386 displayed in figure [ figure6 ] approach the identity matrix exponentially fast along the entire segment , more precisely @xmath387 using also and we shall consider the following rhp . find a @xmath72 piecewise analytic function @xmath388 such that * @xmath388 is analytic for @xmath389 where @xmath390 and we fix @xmath391 . * the boundary values are related via the equations ( compare figure [ figure6 ] for orientation ) @xmath392 * @xmath388 is bounded at the origin @xmath328 . *
as @xmath145 with @xmath256,\delta\in(0,\frac{1}{3}\sqrt{2})$ ] fixed , we have @xmath393 uniformly for @xmath394 . the solution to this problem
is very elementary : assemble @xmath395 with @xmath396 ( see proposition [ gprop ] ) and observe that @xmath397 solves a bare " rhp * @xmath397 is analytic for @xmath398 * along the four rays oriented from zero to infinity , @xmath399 * @xmath397 is bounded as @xmath400 . * as @xmath401 , we have @xmath402 uniformly in a full neighborhood of @xmath403 . referring to the locally analytic change of variables , @xmath404 the origin parametrix is then given by @xmath405 with @xmath352 as in .
since @xmath352 is analytic in the disk @xmath406 , compare proposition [ pout:3 ] , we check directly that the jumps of @xmath388 are indeed as in and with the orientation of the contours near the origin like in figure [ figure6 ] . note
also that the four rays in the bare rhp can always be locally deformed to match the contours @xmath407 . furthermore with and we obtain from the desired matching between the model functions : as @xmath408 such that @xmath256 $ ] is fixed , @xmath409 uniformly for @xmath410 .
this completes the construction of the origin parametrix . our construction for the model function near the right inner branch point
@xmath411 is motivated by the local expansions @xmath412 where @xmath413 we fix @xmath414 and the function @xmath415 is defined for @xmath416 with the branch fixed by the requirement @xmath417 for @xmath418 .
this motivates the use of airy functions and our construction follows from now on @xcite , section @xmath419 . with minor modifications .
we define @xmath420 where @xmath421 denotes the unimodular entire function @xmath422 which is constructed with the help of the airy function @xmath423 , the unique solution to the boundary value problem @xmath424 through the standard properties of @xmath423 ( cf .
@xcite ) , the model function @xmath425 in has jumps on the contour shown in figure [ airy1 ] , more precisely * @xmath425 is analytic for @xmath426 * the boundary values are related via the equations @xmath427 * as @xmath401 , we have in a full neighborhood of infinity , @xmath428e^{\frac{2}{3}\z^{\frac{3}{2}}\sigma_3}.\ ] ] with the ( locally conformal ) change of variables @xmath429^{\frac{2}{3}}\sim\left(4t\sqrt{2m(m^2-m^2)}\,\right)^{\frac{2}{3}}(\lambda - m),\hspace{0.5 cm } |\lambda - m|<r,\ ] ] we define @xmath430 which involves the ( locally ) analytic function @xmath431 analyticity of @xmath432 near @xmath411 follows from the simple observation that for @xmath433 , @xmath434 thus @xmath432 can only have an isolated singularity at @xmath411 . but
this singularity is at worst of square root type , hence has to be removable . as we are allowed to locally deform the jump contours in and match those in the initial @xmath435-rhp , the local parametrix @xmath436 has jumps in @xmath437 as shown in figure [ figure7 ]
moreover , with the help of , @xmath438e^{-{\mathrm{i}}\frac{\pi}{4}\epsilon\sigma_3}\nonumber\\ & = & \left[i+n(\lambda)\left\{\frac{1}{48\z^{\frac{3}{2}}}\begin{pmatrix } -1 & -6\epsilon\\ 6\epsilon & 1\\ \end{pmatrix } + \mathcal{o}\left(\z^{-\frac{6}{2}}\right)\right\}\big(n(\lambda)\big)^{-1}\right]n(\lambda)\label{mat:1 } \end{aligned}\ ] ] so that ( as @xmath439 with @xmath256 $ ] fixed ) @xmath440 uniformly for @xmath441 .
the parametrix near the remaining inner branch point @xmath442 is constructed similarly : introduce @xmath443 where @xmath444 with @xmath445 and collect the following properties * @xmath446 is analytic for @xmath447 * the jumps are as follows , compare figure [ airy2 ] for orientation , @xmath448 * as @xmath401 , valid in a full neighborhood of infinity , @xmath449e^{\frac{2}{3}{\mathrm{i}}\z^{\frac{3}{2}}\sigma_3 } \end{aligned}\ ] ] with the change of variables @xmath450^{\frac{2}{3}}\sim\left(4t\sqrt{2m(m^2-m^2)}\,\right)^{\frac{2}{3}}(\lambda+m),\ \ \ |\lambda+m|<r,\ ] ] we move ahead and define the parametrix near @xmath442 , @xmath451 where @xmath452 with @xmath453 .
again , the multiplier @xmath454 is analytic at @xmath442 , since @xmath455 moreover , we check that @xmath456 ( after employing a local contour deformation ) has jumps inside @xmath457 as shown in figure [ figure8 ] and through , @xmath458e^{-{\mathrm{i}}\frac{\pi}{4}\epsilon\sigma_3}\\ & = & \left[i+n(\lambda)\left\{\frac{{\mathrm{i}}}{48\z^{\frac{3}{2}}}\begin{pmatrix } 1 & -6\epsilon\ , e^{2\pi{\mathrm{i}}tv}\\ 6\epsilon\ , e^{-2\pi{\mathrm{i}}tv } & -1\\ \end{pmatrix}+\mathcal{o}\left(\z^{-\frac{6}{2}}\right)\right\}\big(n(\lambda)\big)^{-1}\right]n(\lambda )
. \end{aligned}\ ] ] thus for @xmath439 such that @xmath256,\delta\in(0,\frac{1}{3}\sqrt{2})$ ] , @xmath459 uniformly for @xmath460 .
this completes the construction of the parametrices at the inner branch points @xmath461 .
once more airy functions are used in the constructions , first near @xmath462 : define the bare parametrix , @xmath463 with @xmath464 this model function has the properties listed below * @xmath465 is analytic for @xmath466 * we have the following jump behavior on the contour shown in figure [ figure9 ] , @xmath467 * as @xmath401 , @xmath468e^{-\frac{2}{3}{\mathrm{i}}\z^{\frac{3}{2}}\sigma_3}\ ] ] this time , we use the change of variables , @xmath469^{\frac{2}{3}}\sim\left(4t\sqrt{2m(m^2-m^2)}\right)^{\frac{2}{3}}(\lambda - m),\ \ \ |\lambda - m|<r\ ] ] and the local parametrix is defined as @xmath470 here , @xmath471 is once more analytic near @xmath462 .
the relevant jump properties of @xmath472 are displayed in figure [ figure10 ] and with , @xmath473e^{-{\mathrm{i}}\frac{\pi}{4}\epsilon\sigma_3}\\ & = & \left[i+n(\lambda)\left\{\frac{{\mathrm{i}}}{48\z^{\frac{3}{2}}}\begin{pmatrix } -1 & 6\epsilon\\ -6\epsilon & 1\\ \end{pmatrix}+\mathcal{o}\left(\z^{-\frac{6}{2}}\right)\right\}\big(n(\lambda)\big)^{-1}\right]n(\lambda ) .
\end{aligned}\ ] ] hence for @xmath439 such that @xmath256,\delta\in(0,\frac{1}{3}\sqrt{2})$ ] , @xmath474 uniformly for @xmath475 .
we are left with the parametrix near the left most branch point @xmath476 : assemble @xmath477 where @xmath478 and , as before , @xmath479 .
we have * @xmath480 is analytic for @xmath481 * fixing orientations as shown in figure [ figure11 ] below , the relevant boundary values are related as follows @xmath482 + * as @xmath401 , @xmath483e^{-\frac{2}{3}\z^{\frac{3}{2}}\sigma_3 } \end{aligned}\ ] ] and with the change of variables @xmath484^{\frac{2}{3}}\sim\left(4t\sqrt{2m(m^2-m^2)}\,\right)^{\frac{2}{3}}(\lambda+m),\ \ |\lambda+m|<r,\ ] ] the required local parametrix is given by @xmath485 where @xmath486 and @xmath487 as @xmath94 . since @xmath488 is analytic at @xmath476 , the model function @xmath489 has the jumps shown in figure [ figure12 ] and with , @xmath490e^{-{\mathrm{i}}\frac{\pi}{4}\epsilon\sigma_3}\\ & = & \left[i+n(\lambda)\left\{\frac{1}{48\z^{\frac{3}{2}}}\begin{pmatrix } 1 & 6\epsilon\,e^{2\pi{\mathrm{i}}tv}\\ -6\epsilon\,e^{-2\pi{\mathrm{i}}tv } & -1\\ \end{pmatrix}+\mathcal{o}\left(\z^{-\frac{6}{2}}\right)\right\}\big(n(\lambda)\big)^{-1}\right]n(\lambda ) .
\end{aligned}\ ] ] this implies for @xmath439 such that @xmath256,\delta\in(0,\frac{1}{3}\sqrt{2})$ ] , @xmath491 uniformly for @xmath492 .
in this final step , we use the explicit model functions @xmath388 in , @xmath436 in , @xmath456 in , @xmath472 in , @xmath489 in and @xmath352 in .
we transform the previous rhp [ openrhp ] for @xmath346 by setting @xmath493 where @xmath494 and @xmath495 remain fixed .
definition implies that the rhp for @xmath346 shown in figure [ figure6 ] is transformed to the rhp for @xmath496 posed on the contour @xmath497 shown in figure [ figure13 ] .
* @xmath496 is analytic for @xmath498 where @xmath499\cup[m , m]\cup\gamma_1^{\pm}\cup\gamma_2^{\pm}\cup c_0\cup c_{r_1}\cup c_{r_2}\cup c_{\ell_1}\cup c_{\ell_2}\cup\sigma_{s_{\infty}}$ ] * for the oriented contour @xmath497 shown in figure [ figure13 ] , the jumps read as follows .
first inside the five circles @xmath500 , @xmath501 where @xmath502 and ( inside the far most left circle @xmath503 ) , @xmath504 followed by ( inside the left circle @xmath505 ) , @xmath506 as well as for the corresponding circles on the right , @xmath507 and @xmath508 inside the remaining circle centered at the origin , @xmath509 recalling our previous expansions for @xmath510 in , , we see that for @xmath511 inside the circles with radii @xmath512 and @xmath513 , we have @xmath514,\ ] ] which is uniform with respect to the chosen radii .
but since @xmath515 is bounded on the same contours , we obtain the estimation @xmath516 where @xmath517 denotes any of the five open disks centered at @xmath328 or @xmath327 .
secondly we estimate the jumps on the remaining finite branches of @xmath497 : for the horizontal line segments , first for @xmath518 , followed then by @xmath519 , @xmath520 notice that for @xmath519 , @xmath521 with a similar estimate also holding for @xmath518 . as the outer parametrix @xmath522 remains bounded on the line segments for fixed radii , we have with universal @xmath523 ( using again , ) , @xmath524 + although we have chosen fixed radii in we shall always indicate the dependency of the estimations of @xmath525 on the latter .
this will be useful later on when we partially drop the constraint on @xmath256 $ ] .
+ next , on the lens boundaries , for @xmath526 , @xmath527 and therefore , with fixed radii which ensures the boundedness of @xmath352 , @xmath528 the pieces of @xmath497 now left are all infinite branches and circle boundaries , more precisely @xmath529 @xmath530 and @xmath531 we conclude with , , , and , ( fixed radii so that @xmath352 is bounded ) @xmath532 also , for the infinite branches , @xmath533 where @xmath534 depends only on the stokes multipliers and @xmath534 are piecewise constant triangular matrices . due to this triangularity and the sign chart of @xmath535 , compare figure [ figure6 ] , we have that @xmath536 * as @xmath94 , @xmath537 the latter rhp for @xmath496 can be solved iteratively as @xmath439 provided @xmath256 $ ] .
we collect estimations - , so that @xmath538,\,\,\,0<\delta<\frac{1}{3}\sqrt{2}.\ ] ] here we used that for the values of @xmath256 $ ] we only work with fixed radii in . since the rhp at hand is equivalent to the singular integral equation @xmath539 estimation guarantees ( cf .
@xcite ) the unique existence of an asymptotic solution of the rhp in @xmath540 , this solution satisfies moreover @xmath541
we now extract the asymptotics through , i.e. we apply @xmath542\ ] ] and for this have to trace back the sequence of transformations @xmath543 first , we get that @xmath544=2\sqrt{-x}\lim_{z\rightarrow\infty}\big[zx(z)\big]_{12}=2\sqrt{-x}\lim_{z\rightarrow\infty}\big[zt(z)\big]_{12}\\ & = 2\sqrt{-x}\lim_{z\rightarrow\infty}\left[ze^{t\ell\sigma_3}s(z)e^{-t(g(z)-\vartheta(z))\sigma_3}\right]_{12}=2\sqrt{-x}\lim_{z\rightarrow\infty}\left[ze^{t\ell\sigma_3}l(z)e^{-t(g(z)-\vartheta(z))\sigma_3}\right]_{12}\\ & = 2\sqrt{-x}\lim_{z\rightarrow\infty}\left[ze^{t\ell\sigma_3}n(z)r(z)e^{-t(g(z)-\vartheta(z))\sigma_3}\right]_{12}.\end{aligned}\ ] ] but since @xmath545 we have @xmath546 also , as @xmath364 , @xmath547 and @xmath548 with ( compare proposition [ pout:3 ] ) , @xmath549 where we used and , i.e. @xmath550 and the well known transformation identities between jacobi theta functions , compare appendix [ app : theta ] .
so far , we have thus the exact identity , @xmath551,\ ] ] with @xmath552\ ] ] in order to further simplify expansion , we note that [ helpful ] @xmath553 let us treat @xmath554\,{{\mathrm d}}\mu\\ & -2{\mathrm{i}}\int_{-m}^m\sqrt{\left(m^2-\mu^2\right)\left(m^2-\mu^2\right)}\,{{\mathrm d}}\mu,\hspace{0.7 cm } m = m(m)=\sqrt{\frac{1}{2}-m^2}\end{aligned}\ ] ] as a function of one variable @xmath555 $ ] .
it is not hard to verify that @xmath556.\ ] ] but from the identity @xmath557 , compare , we see that the last difference of integrals vanishes , hence @xmath558 is in fact @xmath559 independent .
letting @xmath560 , we have @xmath561 and thus the stated identity follows .
summarizing our previous simplifications and noting that @xmath562 , we have @xmath563 and [ error:1 ] for any given @xmath177 there exist positive constants @xmath564 and @xmath565 such that @xmath566 the leading term in , although currently expressed in terms of ratios of jacobi theta functions , can be rewritten with the help of the jacobi elliptic function using appendix [ app : theta ] , @xmath567,\ ] ] and the identity ( see ) @xmath568 hence , @xmath569 and the error term @xmath570 is estimated in proposition [ error:1 ] .
we have already control over the error in the domain @xmath571 but in order to complete the proof of theorem [ bet:1 ] this domain will be extended in the following sections .
define @xmath891 where @xmath892 and @xmath682 remain fixed throughout.this leads to the rhp shown in the figure below .
* @xmath893 is analytic for @xmath894 where @xmath895\cup c_0\cup c_{r_3}\cup c_{\ell_3}\cup\sigma_{s_{\infty}}$ ] * the jumps are described as follows : inside the two circles @xmath896 resp . @xmath897
we simply copy the corresponding expressions from section [ ratiosec1 ] , namely the jumps inside @xmath601 resp . @xmath503 with the proper identification of the @xmath197-function and the replacement @xmath898 .
the same applies to the infinite branches @xmath386 outside @xmath778 as well as for @xmath899 , with the replacement @xmath900 as determined through . on
the line segments @xmath901 , observe that @xmath902 and inside @xmath778 , first for @xmath903 , secondly for @xmath904 , @xmath905 notice that for @xmath906 , @xmath907 and for @xmath908 , @xmath909 but since @xmath910 is bounded on the full segment @xmath628 $ ] and @xmath911 inside @xmath846 , we obtain that @xmath912)}\leq d_{11}e^{-d_{12}t},\ \
t\rightarrow\infty,|s_1|\uparrow 1:\ \
\varkappa\geq \frac{2}{3}\sqrt{2}-\frac{f_2}{t},\ \ f_2\in\mathbb{r}.\ ] ] the jumps along @xmath386 inside @xmath778 are again exponentially close to the unit matrix , hence we now list the contributions on the circle boundaries @xmath913 and @xmath778 , @xmath914 and @xmath915 from , with fixed radius @xmath513 , @xmath916 on the other hand with and again fixed radius @xmath693 , @xmath917 * as @xmath94 we have that @xmath918 .
we solve the rhp for @xmath893 iteratively .
the following estimation is obtained as in section [ ratiosec1 ] , we only have to keep in mind , and , @xmath919 thus the ratio problem has a unique asymptotic solution in @xmath920 which satisfies @xmath921
assume that both @xmath572 and @xmath573 are sufficiently large such that @xmath574 from and , we see that @xmath575 as @xmath576 for some @xmath577 . hence
, in the definition of the ratio problem for @xmath496 , we can not keep the radii @xmath578 and @xmath513 fixed since @xmath579 .
this issue can be resolved by scaling both radii with @xmath580 , in fact we shall choose in @xmath581 in order to estimate the jumps in the ratio problem with the latter choice in place , we need to derive estimations for the outer parametrix @xmath352 and @xmath582 on the jump contours .
first with corollary [ cor1 ] , expansion , @xmath583 uniformly for @xmath290 chosen from compact subsets of @xmath584 .
next , with the contracting choice of radii , @xmath585 uniformly for @xmath586 .
the same estimations for @xmath587 are also valid on the infinite branches @xmath588 as well as on all remaining finite branches outside the five disks @xmath517 centered at @xmath328 and @xmath327 .
together@xmath589 uniformly on the circle boundaries and all other finite or infinite branches outside @xmath590 .
we can now simply go back to - and substitute , and into the estimations .
we obtain immediately @xmath591 and thus , repeating the steps in the previous section ( with adjusted error terms ) , @xmath592 uniformly as @xmath145 such that @xmath593 for any @xmath594 . in estimation
, we obtained control over the error term through the a - priori knowledge @xmath595 and a direct estimation of @xmath596 we can get a better error estimation by explicitly computing the first terms in @xmath597 two of the four contour integrals along the circles @xmath598 have been explicitly computed in @xcite , section @xmath599 .
it was shown in loc .
cit that the aformentioned two contour integrals are in fact @xmath600 even with contracting radii .
the same result also applies to the contour integrals over @xmath503 and @xmath601 and without reproducing the lengthy computations of @xcite , we simply conclude @xmath602 [ bettererror ] even without referring to the explicit computations in @xcite we see directly from , say and , that the evaluation of the contour integral @xmath603 by residue theorem leads to an asymptotic series in reciprocal nonnegative integer powers of @xmath580 . for further improvement , we use that we have slightly better @xmath604-estimations than , @xmath605 so that by general small norm theory of @xcite , @xmath606 back to , for any @xmath607 , @xmath608 and we have therefore derived with an error term @xmath570 such that [ imp:1 ] for any given @xmath609 there exist positive constants @xmath610 and @xmath611 such that @xmath612 notice that for @xmath576 such that @xmath144 we obtain from and , @xmath613 with @xmath614 also from , as @xmath144,@xmath615 but from stirling s approximation , @xmath616 since @xmath617 . using in addition , @xmath618 expansion ,
combined in gives therefore @xmath619-\frac{\pi}{2}(1+\epsilon)+\mathcal{o}\left((-x)^{-\frac{3}{10}}\right),\ ] ] where ( compare ) @xmath620 substituting back into the expansion of the jacobi elliptic function , we use the addition theorem for @xmath621 and the identity @xmath622 so that finally @xmath623 uniformly for @xmath144 .
this short computation shows that the jacobi elliptic function leading term in reproduces the known oscillatory leading behavior of for the values of @xmath194 such that @xmath576 and @xmath144 .
however in the same region of the double scaling diagram we do not yet have control over the error @xmath570 , compare proposition [ imp:1 ] .
this will be achieved below using different nonlinear steepest descent techniques applied to the rhp [ masterrhp ] .
suppose throughout that @xmath572 is sufficiently large such that@xmath624 with this constraint in place we employ a different approach to the nonlinear steepest descent analysis of rhp [ masterrhp ] .
in fact we now choose steps which are close in certain points to the ones used in the derivation of for fixed @xmath4 , see @xcite , chapter @xmath625 , @xmath626 .
start from the riemann - hilbert problem for @xmath237 with its jump contour shown in figure [ figure2 ] or figure [ lower1 ] below and where we fix @xmath627 the jump matrix on the full segment @xmath628 $ ] , compare , is factorized @xmath629 and we observe that with , @xmath630 next define ( compare figure [ lower1 ] ) , @xmath631 and obtain the following rhp * @xmath632 is analytic for @xmath633 * the boundary values are related via the following jump conditions @xmath634 \end{aligned}\ ] ] * the function @xmath632 is normalized , @xmath635 notice that the jumps on the infinite branches @xmath636 for @xmath637 are exponentially close to the unit matrix in the double scaling limit as long as we stay away from @xmath238 . on the remaining infinite branches we have for @xmath638 with our choice @xmath639 , @xmath640 with a universal constant @xmath577 .
thus recalling , we could choose a contracting radius @xmath641 , compare section [ rloweresti ] below , and obtain again sufficiently fast decay for the jumps on the latter infinite branches .
we will work out all necessary details after the following two subsections .
the required outer parametrix is given by @xmath642;\hspace{1cm}\nu=\frac{\varkappa t}{2\pi{\mathrm{i}}}\ ] ] and appeared in @xcite , chapter @xmath625 , @xmath626 .
its relevant analytical properties are summarized below * @xmath643 is analytic for @xmath644 $ ] * the boundary values are related via the equation @xmath645 * @xmath643 is square integrable on the closed interval @xmath628 $ ] * as @xmath94 , we have @xmath646 near the end points @xmath238 the required parametrices are constructed in terms of parabolic cylinder functions , see again @xcite , chapter @xmath625 , @xmath626 .
we briefly summarize the results : let @xmath647 denote the unique solution to weber s equation @xmath648 subject to the boundary condition @xmath649 observe that @xmath650 is entire in @xmath651 .
we define for @xmath651 the wronskian type matrix @xmath652 and assemble @xmath653 where @xmath654 and @xmath655 using standard properties of parabolic cylinder functions , cf .
@xcite this setup leads to the following rhp for the bare parametrix @xmath656 * @xmath656 is analytic for @xmath657 * the jumps along the contours shown in figure [ cylindbare ] are as follows @xmath658 * as @xmath401 , valid in a full neighborhood of infinity , @xmath659\\ & \times\,e^{(\frac{\z^2}{4}-(\nu+\frac{1}{2})\ln\z)\sigma_3 } \end{aligned}\ ] ] where we used the full asymptotic series @xmath660 combined with the identity @xmath661 next , we employ the locally conformal change of variables @xmath662 and define the edge point parametrix near @xmath663 as @xmath664 where @xmath665 is analytic at @xmath663 . after a local contour deformation
we check that the jumps of @xmath666 near @xmath663 are identical to the ones in the initial @xmath667-rhp , i.e. ( compare figure [ parabo ] ) @xmath668 where @xmath669 .
moreover , with the previous asymptotic expansion of @xmath656 , @xmath670\z^{-\frac{1}{2}\sigma_3}\left(-\frac{h_1}{s_3}\right)^{\frac{1}{2}\sigma_3}\big(\beta(\lambda)\big)^{-\sigma_3}\psi^d(\lambda)\nonumber\\ & = \bigg[i+\frac{\sigma_-}{\z(\lambda)}\big(\beta(\lambda)\big)^{-2}\left(-\frac{h_1}{s_3}\right)\sum_{m=0}^{\infty}(\nu+1)_{2m}\frac{\z(\lambda)^{-2m}}{m!\,2^m}+\sigma_+\big(\beta(\lambda)\big)^2\left(-\frac{s_3}{h_1}\right)\nonumber\\ & \times\,\sum_{m=1}^{\infty}(-)^m\big((-\nu)_{2m}-(-\nu-1)_{2m}\big)\frac{\z(\lambda)^{-2m+1}}{m!\,2^m}+\sum_{m=1}^{\infty}\begin{pmatrix } ( \nu)_{2 m } & 0\\ 0 & ( -)^m(-\nu)_{2m}\\ \end{pmatrix}\frac{\z(\lambda)^{-2m}}{m!\,2^m}\bigg]\psi^d(\lambda),\label{mup:1}\end{aligned}\ ] ] as @xmath671 ( and thus @xmath672 ) .
here we used the abbreviation @xmath673 expansion establishes the matching relation between the model functions @xmath666 and @xmath643 valid as @xmath674 for @xmath675 .
near the other edge point @xmath676 we can use the symmetry of the problem and define the required local parametrix as @xmath677 which replaces the matchup by @xmath678\psi^d(\lambda).\label{mup:2}\end{aligned}\ ] ] collecting the model functions @xmath679 and @xmath680 we move on to the final transformation .
assemble @xmath681 with @xmath682 which we will choose more specifically below .
this leads to the rhp formulated on the contours as shown in figure [ ratiolowerfig ] * @xmath683 is analytic for @xmath684 * the following jumps are valid : @xmath685 along the infinite branches @xmath686 with piecewise constant matrices @xmath687 which are given in the @xmath667-rhp .
there are no jumps inside the circles and along the line segment @xmath628 $ ] since we constructed exact " local parametrices @xmath679 and @xmath680
. however on the circle boundaries , @xmath688 * as @xmath94 , we have that @xmath689 for the jumps on the infinite branches @xmath690 we obtain immediately @xmath691 with universal constants @xmath692 . if we choose a ( in general ) contracting radius @xmath693 such that @xmath694 then by for @xmath198 , @xmath695 hence all contributions from the infinite branches are approaching the identity matrix exponentially fast , i.e. @xmath696 for @xmath198 . for the circle boundaries we use and , for instance for @xmath697 we use @xmath698 as well as @xmath699 which follows from stirling s approximation
thus for @xmath697 , @xmath700 where @xmath701 bounds for the error term @xmath702 can be found for instance in @xcite : there exists @xmath703 and constants @xmath704 such that @xmath705 for the latter estimation to hold for @xmath706 , we have to adjust the radius @xmath707 with a sufficiently large , but @xmath708 independent , constant @xmath709 .
notice that with completely similar estimations will also hold for @xmath710 when @xmath711 .
thus , summarizing @xmath712 the latter estimation combined with , we obtain from general theory @xcite the unique solvability of the ratio problem for the values of @xmath194 satisfying .
moreover its unique solution satisfies @xmath713 we need to trace back the transformations @xmath714 in order to obtain with , @xmath544=2\sqrt{-x}\lim_{z\rightarrow\infty}\big[z\,\phi(z)\big]_{12 } = 2\sqrt{-x}\lim_{z\rightarrow\infty}\big[z\psi^d(z)\chi(z)\big]_{12}\\ & = 2\sqrt{-x}\left[\nu\sigma_3+\frac{{\mathrm{i}}}{2\pi}\int_{\sigma_{\chi}}\chi_-(w)\big(g_{\chi}(w)-i\big)\,{{\mathrm d}}w\right]_{12}\\ & = \frac{{\mathrm{i}}\sqrt{-x}}{\pi}\oint_{c_r\cup c_{\ell}}\big(g_{\chi}(w)-i\big)_{12}\,{{\mathrm d}}w+\mathcal{o}\left((-x)^{-\frac{2k-7}{2(k+1)}}\right),\ \
k\in\mathbb{z}_{\geq 4}\end{aligned}\ ] ] where we already neglected all exponentially small contributions from the infinite branches , compare .
also the latter error estimation follows from and the following computation : observe that by residue theorem and as well as , @xmath715 recalling the definition of @xmath716 and applying standard symmetrization techniques , we obtain with @xmath717 , @xmath718 uniformly as @xmath3 subject to the scale with @xmath717
. estimation appears in corollary [ mat ] , expansion .
moreover , with , we have now shown that @xmath719 where [ imp:2 ] there exists positive constants @xmath720 and @xmath721 such that @xmath722 propositions [ imp:2 ] and [ imp:1 ] combined together ( for @xmath723 in the latter ) form the content of theorem [ bet:1 ] , estimation . in order to complete the proof of theorem [ bet:1 ] we have to derive the outstanding estimation by extending the region @xmath724 at the upper end .
this will be achieved in the next section .
in the analysis of the lower end extension presented in section [ relax ] we used the crucial fact that for @xmath141 , we have @xmath725 , compare . thus at the lower end , the oscillations in the jacobi theta functions become strictly periodic and their amplitudes vanish exponentially fast . in turn
we obtained @xmath726 which lead us to the central estimation . at the upper end
the oscillations in the theta functions still vanish , however this time not through vanishing amplitudes but increasing periods : as @xmath727 we get with , @xmath728 this implies the necessity of using modular transformations for the jacobi theta functions appearing in the construction of the outer parametrix , i.e. we use ( cf .
@xcite ) @xmath729 going back to the transformation identities imply for @xmath353 , @xmath730 where @xmath731 notice that from , for @xmath732 $ ] chosen from compact subsets , @xmath733,\ \ \sigma\downarrow 0\ ] ] and for the same values of @xmath84 we have at the same time @xmath734 since with , @xmath735 along the line segment @xmath736 $ ] , by direct computation @xmath737\cup[m+r , m),\ \ \ 0<r<\frac{1}{2}\ \ \ \textnormal{fixed},\ ] ] and hence all together @xmath738 for any fixed @xmath739 , provided we ensure that @xmath740 we shall in fact impose a stronger condition on @xmath154 than implied by : suppose subsequently that @xmath741 and @xmath742 are sufficiently large such that @xmath743 subject to this constraint we obtain directly with , @xmath744 uniformly for @xmath745
. moreover [ ou : match ] as @xmath145 subject to we have @xmath746 uniformly for @xmath84 chosen from @xmath747 for any @xmath739 . here
, @xmath748 equals @xmath749 where @xmath750 is defined and analytic on @xmath751 $ ] . at this point
let us get back to the nonlinear steepest descent analysis .
we start from the initial rhp [ masterrhp ] and use the transformation sequence @xmath752 just like it occurred in section [ sec:3 ] and parts of section [ sec : reg ] in the analysis for @xmath256 $ ] .
however , instead of using the opening of lens " transformation we will not split the jump contours near @xmath334 but instead introduce a new model problem on the ( contracting ) line segment @xmath753 with @xmath754 in more detail [ nmodel ] find a @xmath72 matrix - valued piecewise analytic function @xmath755 such that * @xmath756 is analytic for @xmath757 * the limiting values of @xmath756 are related via the jump condition @xmath758 where @xmath759 has been explicitly computed in proposition [ gprop ] .
also , @xmath760 are bounded on @xmath761 . *
as @xmath145 subject to , @xmath762 uniformly for @xmath763 and @xmath748 has been introduced in .
we choose a solution to this model problem in the following form , @xmath764 where @xmath765 is locally analytic , @xmath766 let us quickly verify that indeed satisfies the properties of rhp [ nmodel ] .
first , analyticity of the multiplier @xmath765 follows from the jump of @xmath748 , @xmath767 thus the jump behavior of @xmath756 is a consequence of the plemelj - sokhotskii formula , @xmath768 secondly , as @xmath145 with , @xmath769 uniformly for @xmath763 .
here we used a laplace - type argument for the two remaining integrals based on the behavior @xmath770 as well as the expansions @xmath771 thus @xmath772 as @xmath439 subject to uniformly for @xmath763 with @xmath578 as in . instead of the ratio transformation , introduce for , @xmath773 where @xmath774 and @xmath775 remains fixed .
the model functions @xmath776 and @xmath489 are ( as before ) given in , and .
the rhp for @xmath777 is formulated on the contour shown in figure [ figureratio3 ] .
compared to the ratio problem for @xmath496 the only difference in the jump behavior occurs inside the circle @xmath778 centered at the origin , @xmath779,\ ] ] for @xmath780 as well as @xmath781 and on the circle boundary @xmath747 , @xmath782 notice that ( since @xmath783 ) @xmath784\ ] ] subject to as well as @xmath785 the latter combined with , , , using also the boundedness of @xmath582 away from the three disks @xmath517 centered at @xmath328 and @xmath786 and recalling at the same time our estimations for the jumps given in section [ ratiosec1 ] , we obtain @xmath787 opposed to the this @xmath788-estimation we can obtain better @xmath604-estimations through @xmath789 and @xmath790 the last estimation is required for the @xmath604-estimation of the jump on @xmath747 : @xmath791 we summarize , @xmath792 and combining , this implies from general theory @xcite the unique solvability of the @xmath793-rhp in @xmath794 , moreover its unique solution satisfies @xmath795 in the same way as we derived , we obtain @xmath796\ ] ] with @xmath797\big(g_{r_u}(w)-i\big)_{12}\,{{\mathrm d}}w+\mathcal{o}\left(\frac{1}{t^{\frac{1}{3}}\ln t}\right).\ ] ] however , by direct computation , @xmath798 and thus @xmath799 with [ imp:3 ] for any given @xmath800 there exist positive constants @xmath801 and @xmath802 such that @xmath803 together with propositions [ imp:1 ] and [ imp:2 ] , proposition [ imp:3 ] completes the proof of theorem [ bet:1 ] . in order to derive in corollary [ mat ] , we choose @xmath216 such that @xmath804 .
since @xmath805 we get with at once @xmath806 similarly , @xmath807 moving ahead ( recall ) , @xmath808 uniformly as @xmath145 such that @xmath804 . in view of proposition [ imp:3 ] , the latter estimation completes the proof of corollary [ mat ] .
[ stokefirst ] the introduction of @xmath756 in together with bounds on @xmath582 in , allowed us to control the error @xmath570 in the region .
the analysis of @xmath570 inside the region @xmath809 which is not covered by propositions [ imp:1 ] , [ imp:2 ] and [ imp:3 ] is much more difficult . in this case is violated in general and thus @xmath582 may become unbounded .
for instance , if we were to choose @xmath810 then from and , using again , @xmath811 hence we only restore the leading hastings - mcleod asymptotics for @xmath812 . for the values of @xmath813
more terms will contribute to the leading order , or equivalently , the leading asymptotics changes once the stokes line " @xmath814 is crossed .
we provide further detail on this interesting feature in section [ stokes ] below .
in this section we assume that both @xmath741 and @xmath816 are sufficiently large such that @xmath817 again , we use the transformation sequence @xmath818 but subsequently the nonlinear steepest descent analysis of the @xmath304-rhp in is very different from the one used in sections [ sec : reg ] and [ flower ] .
in fact , the steps carried out below are a modification of the ones used in the analysis for @xmath5 , see @xcite , chapter @xmath819 , @xmath626 .
the details are as follows .
we fix @xmath820 in and define for @xmath821 with @xmath822\cup\hat{\gamma}_2\cup\hat{\gamma}_5\cup\tilde{\gamma}_2\cup\tilde{\gamma}_5\bigcup\gamma_k$ ] , @xmath823 this transforms the @xmath304-rhp to a rhp with jumps on @xmath824\cup[0,\lambda^{\ast}]$ ] such that @xmath825\label{s:1}\\ \varpi_+(\lambda ) & = & \varpi_-(\lambda)\begin{pmatrix } ( 1-s_1s_3)e^{t\pi(\lambda ) } & -s_3\\ s_1+s_1(1-s_1s_3 ) & ( 1-s_1s_3)e^{-t\pi(\lambda)}\\ \end{pmatrix},\hspace{0.5cm}\lambda\in[0,\lambda^{\ast}]\label{s:2}\end{aligned}\ ] ] where @xmath826 for the @xmath827-entries in and we use and observe that @xmath828 hence the strict inequality for @xmath829 will fail on the ( shrinking ) interval @xmath830 $ ] .
we thus face the necessity to consider an additional model problem at the origin . besides this new feature , we however still expect ( and will prove ) that the outer model problem is given by a problem in which we neglect the diagonal entries in , and apply estimations and for the off - diagonal , i.e. on the full line segment @xmath628 $ ] we require a function @xmath748 such that @xmath831 furthermore , along the remaining infinite branches , i.e. along @xmath832 the jumps in the @xmath324-rhp are once more of the form @xmath833 hence by triangularity and the sign of @xmath326 these jumps are asymptotically close to the unit matrix away from the endpoints @xmath238 .
we proceed now with the analysis of the relevant model problems .
the outer problem consists in finding a @xmath72 piecewise analytic function @xmath834 such that * @xmath748 is analytic for @xmath835 $ ] * if the cut is oriented from @xmath836 to @xmath837 , the boundary values of @xmath748 are related via @xmath838 * @xmath748 is square integrable on the closed interval @xmath839 $ ] * as @xmath94 , @xmath840 a solution to this _ quasi - permutation _ rhp appeared already in proposition [ ou : match ] , formula .
note also that @xmath841 our construction is motived by the local expansion @xmath842 and solves the following problem .
determine a @xmath72 piecewise analytic function @xmath843 such that * @xmath843 is analytic inside the disk @xmath844 for some fixed @xmath845 * orienting the segment @xmath846 from left to right , we have the jump condition @xmath847 * as @xmath145 subject to we have @xmath848 uniformly for @xmath849 .
here , @xmath748 is the outer model function introduced in section [ outer:2 ] .
the difference between this problem and rhp [ nmodel ] is only marginal : recall that @xmath850 and set @xmath851 we require the following lemma .
[ princip ] as @xmath401 , we have @xmath852 assume first that @xmath853 , hence @xmath854=-\frac{\sqrt{\pi}}{2\pi{\mathrm{i}}\z}\left(1+\mathcal{o}\left(\z^{-1}\right)\right),\ \ \z\rightarrow\infty.\ ] ] secondly for @xmath855 we apply plemelj formula @xmath856 the principal value integral can be simplified as follows : let @xmath857 denote the contour shown in figure [ box ] with @xmath858 . from cauchy s theorem @xmath859 on the other hand @xmath860 as well as @xmath861 and @xmath862
thus @xmath863 which completes the proof
. the solution to the model problem is now given by @xmath864 where we introduce the ( locally analytic ) multiplier @xmath865 more detail on the local behavior of @xmath866 near @xmath328 will be needed later on , first @xmath867 followed by @xmath868 and thus @xmath869 we directly verify the required jump condition , and as @xmath674 ( subject to ) , @xmath870 uniformly for @xmath871 with @xmath872 fixed .
the construction is almost identical to the one presented in section [ outpara ] .
more precisely for the right edge point we take @xmath873 where @xmath465 was introduced in and we have @xmath874 with the locally conformal change of variables @xmath875 the reader easily checks that @xmath876 defined in has the same jump behavior as shown in figure [ figure10 ] with @xmath559 replaced by @xmath877 and @xmath878 as in .
moreover @xmath879\upsilon(\lambda),\ \ \z\rightarrow\infty\ ] ] so that for @xmath439 such that @xmath148 , @xmath880 uniformly for @xmath881 . near the remaining edge point , we choose @xmath882 with @xmath883 as in and @xmath884 where we use the change of variable @xmath885 the jump behavior of is depicted in figure [ figure12 ] with @xmath886 replaced by @xmath887 and the @xmath197-function as in .
we note that @xmath888\upsilon(\lambda),\ \
\z\rightarrow\infty,\ ] ] hence , as @xmath439 such that @xmath148 , @xmath889 uniformly for @xmath890 .
this completes the parametrix construction and we can move ahead with the ratio transformation .
through the transformation sequence @xmath922 we obtain @xmath923=2\sqrt{-x}\lim_{z\rightarrow\infty}\left[z\xi(z)\upsilon(z)e^{-t(g(z)-\vartheta(z))\sigma_3}\right]_{12}.\ ] ] since @xmath924 we deduce from the integral equation for @xmath893 the exact identity @xmath925,\ ] ] where ( as @xmath260 such that @xmath926 ) , @xmath927 and the last contour integral was computed by residue theorem , compare . hence , as @xmath145 , @xmath928 with [ imp:4 ] for any given @xmath149 there exist positive @xmath150 and @xmath151 such that @xmath153 this completes the proof of theorem [ res:4 ] and at the same time allows us to derive corollary [ nice ] .
suppose both @xmath580 and @xmath929 are sufficiently large such that @xmath930 in this case @xmath931 is bounded away from zero and hence can be improved to @xmath932 compare .
therefore , @xmath933 and gets replaced by @xmath934 with @xmath31 and @xmath28 independent coefficients @xmath935 .
these could in principle be computed by residue theorem in using and , but it is easier to work with the differential equation itself : as @xmath145 , @xmath936 uniformly for @xmath937 and the asymptotic expansion can be differentiated with respect to @xmath31 .
the modulus @xmath938 will appear explicitly in , but only contained in exponentially small error terms .
all power like terms are @xmath938-independent and in fact identical to the ones derived in @xcite , theorem @xmath939 .
we go back to the fredholm determinant and note that for fixed @xmath940 , @xmath941 hence with remark [ connect ] and after integration , @xmath942 where @xmath943 is @xmath31 independent and there exist universal @xmath944 and @xmath945 such that @xmath946 now keep @xmath576 fixed and let @xmath52 in , so that with , @xmath947 this , together with , completes the proof of corollary [ nice ] .
we assume that @xmath741 and @xmath816 are sufficiently large such that @xmath948 working with this scale , several steps in the nonlinear steepest descent analysis of section [ near ] can still be applied . in fact , the only difference appears in the matching relation .
subject to , we have that @xmath949 and thus the two model functions @xmath843 and @xmath748 are in general no longer asymptotically close to each other on @xmath950 . in order to deal with this feature
, we choose @xmath951 with @xmath952 and notice that for @xmath853 , @xmath953 which implies with lemma [ princip ] , as @xmath674 subject to , @xmath954 and the last term is small because of our assumption @xmath955 .
we now factorize , @xmath956 and go back to section [ near ] . after the explicit transformations @xmath957 we use
another : with and figure [ ratio2 ] , let @xmath958 so that by definition , the function @xmath959 solves the following ( singular ) rhp .
[ singrhp ] determine a @xmath72 matrix valued function @xmath959 such that * @xmath959 is analytic for @xmath960 * along @xmath961 we have jumps , @xmath962 , with @xmath963 * @xmath959 has a pole of order @xmath964 at @xmath328 provided @xmath965 .
in order to be more precise about the singular structure , notice that for @xmath965 , @xmath966\big(e(\lambda)\big)^{-1},\ ] ] with @xmath967 , is analytic at @xmath328 .
thus with and comparison , we can express the full singular part @xmath968 in terms of the first @xmath969 terms of the regular part of @xmath959 .
this is done explicitly for @xmath970 , @xmath971 but becomes much more involved in case @xmath972 ; we shall postpone the general case @xmath972 to a forthcoming publication . * as @xmath94 , we have @xmath973 .
the jump matrices in rhp [ singrhp ] are asymptotically close to the unit matrix , however @xmath959 has a pole at the origin . such type of problems have been analyzed in nonlinear steepest descent literature , see e.g. @xcite , at least for the case of a first order pole .
following loc .
, we first note that is equivalently stated as @xmath974 where @xmath975 is analytic at @xmath328 .
hence , @xmath959 is indeed characterized uniquely by the properties stated in rhp [ singrhp ] .
secondly we can resolve the singular structure for @xmath965 via another transformation , the _ dressing transformation _
, @xmath976 where @xmath977 are @xmath84-independent .
the function @xmath978 is characterized as follows . [ dressrhp ] determine a @xmath72 matrix valued function @xmath978 subject to the following conditions * @xmath978 is analytic for @xmath894 * we have that @xmath979 for @xmath980 , with @xmath981 given in .
* as @xmath94 , @xmath982 notice that the unknowns @xmath983 are algebraically determined from the singular structure in the @xmath984-rhp and , i.e. for @xmath970 we have @xmath985 or for more general @xmath972 equations for the remaining unknowns in term of @xmath986 and higher derivatives thereof .
observe that all jumps in the @xmath987-rhp are close to the identity since @xmath988 uniformly for @xmath260 such that @xmath989 and @xmath990 .
the rhp [ dressrhp ] is iteratively solvable and we have for its solution @xmath991 notice that from a direct computation , by tracing back the transformations , @xmath992 we get @xmath993,\ \ q\in\mathbb{z}_{\geq 0}:\ \ b_{-2}\equiv 0.\ ] ] we will now continue with the special case @xmath970 , i.e. we compute @xmath994 . \ ] ] notice that @xmath995 and thus @xmath996.\ ] ] from this it follows in that @xmath997 and for any given @xmath162 there exist positive @xmath163 and @xmath164 such that @xmath998 if we were to choose @xmath158 , then @xmath999 and we restore the leading order hastings - mcleod asymptotics , @xmath1000 in other words , @xmath1001 forms the first stokes line and we have completed the proof of theorem [ res:5 ] .
observe that and allow us to derive a slightly better expansion than the one given in theorem [ res:5 ] , as @xmath145 , @xmath1002\ ] ] where , uniformly for @xmath1003 , @xmath1004\\ \max\{t^{-\frac{3}{2}}e^{\sigma t},t^{-3}\},&f_3\in(\frac{1}{2},\frac{7}{6 } ) .
\end{cases}\ ] ] next , we use identity and combine it with , @xmath1005 and from , noting that @xmath1006 with @xmath929 fixed , @xmath1007 however , by straightforward computation , @xmath1008 and @xmath1009 satisfies also .
hence , we have shown that for @xmath145 , @xmath1010 uniformly for @xmath1003 with some @xmath31-independent constants @xmath1011 .
these can be computed as in the proof of corollary [ nice ] , i.e. we use @xmath941 together with to conclude @xmath1012 and the limit @xmath52 , with @xmath576 fixed , yields @xmath1013
several steps of the analysis carried out below have their natural counterpart in the regular analysis .
one major difference however lies in the construction of a new outer parametrix which naturally involves the exceptional set @xmath1014 as defined in below .
first we let @xmath1015 and keep @xmath256 $ ] with @xmath254 fixed . also here
, the following expansions will prove useful .
note that @xmath1016 and @xmath1017 moreover , @xmath1018 so that as @xmath1019 with @xmath261 , @xmath1020 @xmath1021 uniformly for @xmath256 $ ] with @xmath254 fixed . the bijective correspondence between @xmath127 and @xmath121 has already been established in proposition [ modex ] , and all local expansions in corollary [ cor1 ] can directly be applied to the singular transition analysis . with this in mind
we define ( formally just as in ) , @xmath1022\ ] ] where @xmath1023 and @xmath128 is determined through .
the relevant analytical properties of @xmath289 are summarized in proposition [ gprop ] and we choose in the @xmath304-rhp ( as before ) , @xmath1024 after that , introduce , @xmath1025 where we shall use the same notation @xmath307 for the transformed function as in the regular transition analysis .
observe that , for @xmath312 , @xmath1026 so that ( as before ) @xmath1027,\ \delta>0.\ ] ] similarly , for @xmath1028 , @xmath1029 hence ( as before ) @xmath1030,\ \delta>0.\ ] ] the only difference occurs in the _ gap _ , first for @xmath330 , @xmath1031 and secondly for @xmath332 , @xmath1032 here , @xmath1033 and @xmath1034 have appeared in , and @xmath1035 .
hence , we can use proposition [ openup ] in the given context as well and open lens : compare figure [ figure6 ] and define @xmath1036 this leads to the following rhp determine a @xmath72 matrix valued function @xmath346 such that * @xmath346 is analytic for @xmath1037\cup[-m , m]\cup[m , m]\cup\gamma_1^{\pm}\cup\gamma_2^{\pm}\cup\sigma_{s_{\infty}}\big)$ ] * the jumps are as follows , see figure [ figure6 ] for orientations , @xmath1038 * as @xmath94 , @xmath1039 the jumps in this problem along the lens boundaries @xmath407 and along the infinite contours are exponentially close to the unit matrix in the ( singular ) double scaling limit as long as we stay away from the branch points @xmath327 and @xmath328 .
compared to the @xmath435-rhp in the regular transition analysis , the outer parametrix will now also have a jump in the gap @xmath334 . in case of the singular transition analysis the outer model problem is posed on the full segment @xmath1040 and reads as follows determine a @xmath72 matrix - valued piecewise analytic function @xmath351 such that * @xmath352 is analytic for @xmath1041 $ ] * we have jumps @xmath1042 * @xmath352 has @xmath604 boundary values on @xmath1043 $ ] * as @xmath94 , @xmath1044 notice that the function @xmath1045\ ] ] has jumps on @xmath1040 described as follows @xmath1046 up to the presence of the diagonal jump in the gap @xmath334 , the jumps are thus identical to the ones which we generated in proposition [ pout:3 ] . to compensate for the diagonal part
, we slightly modify our construction of section [ outer1para ] : let @xmath371 denote again the abel ( type ) map @xmath370\rightarrow\mathbb{c},\ \
z\mapsto u(z)=\int_m^z\omega = u(\infty)-\frac{c}{z}+\mathcal{o}\left(z^{-3}\right),\ \ z\rightarrow\infty;\ \ c = c(\varkappa)=\frac{{\mathrm{i}}}{2}\frac{m}{k'}\ ] ] for which we summarized certain properties in proposition [ pout:1 ] .
we require @xmath1047 where ( formally as before , see also ) , @xmath1048 and @xmath379,\ \ \ z\in\mathbb{c}\ ] ] is the third jacobi theta function . with this
, we have the following analogue to proposition [ pout:3 ] for the current situation - . [ pout:4 ]
the function @xmath1049 is single - valued and analytic in @xmath287 $ ] .
its jumps are stated in , and , furthermore , as @xmath94 , @xmath1050 provided @xmath182 are uniformly bounded away from the discrete set @xmath1051 the stated jump behavior follows from a direct computation using proposition [ pout:1 ] and the properties of jacobi theta functions . for the normalization ,
we take into account the shift by the half period appearing in the numerators in , hence we have to guarantee that @xmath1052 which is ensured for the values @xmath182 away from the set @xmath1014 given in as the simple roots of @xmath1053 are located at @xmath1054 . the latter proposition is in sharp contrast to proposition [ pout:3 ] . there
we do not require the half - period shift and thus we always have that @xmath1055 since @xmath1056 , i.e. there is no exceptional set . from now on we shall always assume that @xmath182 is bounded away from the set @xmath1014 .
the required model functions near @xmath328 and @xmath461 slightly differ from the ones used in the case of the regular transition .
near @xmath328 we analyze determine a @xmath72 piecewise analytic function @xmath388 such that * @xmath388 is analytic for @xmath1057 with @xmath391 .
* for the boundary values , @xmath1058 * as @xmath1019 such that @xmath256 $ ] , uniformly for @xmath394 , @xmath1059 a solution to the latter problem is derived in two steps .
first define @xmath1060 with @xmath1061 from and where we orient all six rays from zero to infinity .
second , the required parametrix itself : with @xmath352 from , @xmath1062 where @xmath1063 the required jump behavior follows from a direct computation using a local contour deformation , moreover we have @xmath1064 as @xmath1065 such that @xmath256 $ ] uniformly for @xmath1066 . near @xmath411 we have to modify .
consider @xmath1067 with @xmath421 as in .
this leads to the following bare rhp * @xmath425 is analytic for @xmath1068 * we have jumps on the contours sketched in figure [ airy1sing ] , @xmath1069 * as @xmath401 , @xmath1070e^{\frac{2}{3}\z^{\frac{3}{2}}\sigma_3}\ ] ] now define @xmath1071 where ( as before ) @xmath1072^{\frac{2}{3}}\sim\left(4t\sqrt{2m(m^2-m^2)}\right)^{\frac{2}{3}}(\lambda - m),\ \ |\lambda - m|<r\ ] ] and @xmath1073 is analytic near @xmath411 .
this leads to the jump behavior of @xmath436 as indicated in figure [ figure7sing ] , moreover we have the matching relation , @xmath1074n(\lambda)\ ] ] and thus , as @xmath1065 such that @xmath256 $ ] with @xmath254 fixed , @xmath1075 uniformly for @xmath441 . for the remaining endpoint @xmath442 , we put @xmath1076 with @xmath1077 as in .
we obtain that * @xmath446 is analytic for @xmath1078 * along the curves shown in figure [ airy2sing ] , @xmath1079 * as @xmath401 , @xmath1080e^{\frac{2}{3}{\mathrm{i}}\z^{\frac{3}{2}}\sigma_3}. \end{aligned}\ ] ] for the parametrix itself , we put @xmath1081 with @xmath1082^{\frac{2}{3}}\sim\left(4t\sqrt{2m(m^2-m^2)}\right)^{\frac{2}{3}}(\lambda+m),\ \ |\lambda+m|<r\ ] ] and the locally analytic multiplier @xmath1083 @xmath1084n(\lambda)\ ] ] so that as @xmath1019 such that @xmath256 $ ] , @xmath1085 uniformly for @xmath460 .
this completes the construction of the relevant local model functions . for @xmath84 close to @xmath1086
, we can simply work with the same functions @xmath472 and @xmath489 as in and .
we work with the ratio function @xmath496 defined in but with @xmath388 from , @xmath436 from , @xmath456 from and @xmath352 as in . by construction , @xmath496 is characterized by the following properties .
* @xmath496 is analytic for @xmath498 where the contour @xmath497 is shown in figure [ figure13 ] .
this follows directly from our construction of @xmath352 and @xmath388 as both functions precisely match the jump behavior of @xmath346 in the gap @xmath334 , compare . *
the jumps of @xmath496 are only slightly different from the ones listed in section [ ratiosec1 ] .
we choose not to list all expressions explicitly , instead only collect the important estimation @xmath1087 as @xmath1065 such that @xmath256 $ ] for @xmath254 fixed , provided we stay away from the singular points @xmath1088 . * as @xmath94 , we have that @xmath1089 . with the help of estimation the rhp for @xmath496 can be solved asymptotically in @xmath540 and we have for its unique solution @xmath1090 just as in the regular case , we have the exact identity @xmath1091_{12}=2\sqrt{-x}\,e^{2t\ell}\left[\big(n_1\big)_{12}+\frac{1}{2}\tilde{\mathcal{e}}(\varkappa ) \right]\ ] ] where @xmath1092 notice that with and identities of theta functions , @xmath1093 and hence we obtain @xmath1094.\ ] ] now using proposition [ helpful ] , combining it with and the identities @xmath1095 one deduces @xmath1096 and for any given @xmath177 there exist positive constants @xmath564 and @xmath565 such that @xmath1097 provided @xmath182 are uniformly bounded away from the exceptional set @xmath1014 introduced in .
the relaxation of the lower bound for @xmath154 is handled via the same approach as in section [ relax ] , i.e. for @xmath741 and @xmath1098 sufficiently large such that @xmath1099 we obtain with the help of contracting radii the following analogue of @xmath1100 uniformly as @xmath190 such that @xmath593 for any @xmath594 away from the exceptional set .
after that we would use the analogue of , respectively remark [ bettererror ] , for the singular analysis and this combined with improved @xmath604 estimations gives @xmath1101 we summarize , [ singimp1 ] for any given @xmath609 there exist positive constants @xmath610 and @xmath611 such that @xmath1102 provided @xmath182 are bounded away from the exceptional set . in order to obtain estimation in corollary [ singmat ] , we would choose @xmath576 such that @xmath144 and again @xmath182 stay away from .
then , by corollary [ cor1 ] we have @xmath1103 next , for @xmath144 with , ( compare ) , @xmath1104 and from stirling s formula @xmath1105 as @xmath617 . together @xmath1106
where ( compare theorem [ known:3 ] ) , @xmath1107 we substitute into , use the addition theorem for cosine as well as the identity @xmath1108 to obtain , as @xmath3 , @xmath1109 uniformly for @xmath144 and away from the zeros of the trigonometric function appearing in the last denominator .
hence , the jacobi elliptic function leading term reproduces also here the known singular oscillatory leading order of for @xmath194 such that @xmath576 and @xmath144 .
next we will also obtain control over the error @xmath1110 in the same region .
we go back to section [ rloweresti ] and assume throughout that both @xmath580 and @xmath1098 are sufficiently large such that @xmath1111 the difference between regular and singular transition asymptotics can be read off from , @xmath1112 hence we can still use most of the nonlinear steepest analysis worked out in section [ flower ] , however a crucial differences occurs in the matching relations and , compare @xcite where the same approach was first used in the derivation of for fixed @xmath6 .
we shall factorize , @xmath1113\psi^d(\lambda)\end{aligned}\ ] ] with @xmath1114 this factorization is necessary since with @xmath1115 i.e. the multiplier @xmath1116 is not close to the unit matrix .
similarly , @xmath1117\psi^d(\lambda)\end{aligned}\ ] ] with @xmath1118 this means that the ratio problem for @xmath683 can not be solved iteratively at this point and we have to use further transformations .
first , compare @xcite , equation @xmath1119 , we let @xmath1120 where we use ( this is in fact the only difference to @xcite ) a shrinking radius @xmath1121 the riemann - hilbert problem for @xmath1122 was analyzed in @xcite , its jump contour is identically @xmath1123 as shown in figure [ ratiolowerfig ] but with additional first order pole singularities at @xmath238 . we choose not to reproduce the analysis here , compare equations @xmath1124 in @xcite . subsequently the singular structure was resolved by another transformation @xmath1125 where the @xmath84 independent matrix @xmath368 is determined algebraically as in equation @xmath1126 in @xcite .
it is well defined for all values @xmath182 bounded away from the discrete set @xmath1127 compare again @xcite , shortly before remark @xmath1128 .
if we agree to stay away from the latter exceptional points , the jumps for @xmath1129 have the following structure @xmath1130 and @xmath1131 hence we have @xmath1132 so that ( just as before in ) , @xmath1133 for @xmath697 we estimate @xmath1134 with @xmath1135 thus , using again bounds for the error term @xmath1136 from @xcite , we obtain that with the contracting radius @xmath1137 combining estimations and we can thus solve the @xmath1138-rhp asymptotically subject to the scale as long as we stay away from the discrete set of points @xmath182 defined via .
the derivation of the leading order follows now as in @xcite , we only have to adjust the error terms according to : from equation @xmath1139 in @xcite , @xmath1140 where we used and the a - priori estimation for the unique solution of the @xmath1138-rhp .
the matrix entry @xmath1141 is then computed explicitly in @xcite and we only have to adjust the error terms . hence,@xmath1142 as @xmath3 subject to the scale for @xmath1143
provided we stay away from the points @xmath182 defined implicitly via .
the latter estimation , with @xmath717 , appears in corollary [ singmat ] , expansion .
moreover , with , we have now shown that @xmath1144 and [ singimp2 ] there exist positive constants @xmath720 and @xmath721 such that @xmath1145 provided @xmath182 are uniformly bounded away from the set .
this concludes the proof of estimation in theorem [ bet:2 ] , we just have to combine propositions [ singimp1 ] ( with @xmath723 ) and [ singimp2 ] .
we choose @xmath576 and @xmath1098 sufficiently large such that @xmath1146 this implies with , @xmath1147 in other words , we can always guarantee that for sufficiently large @xmath580 and @xmath1148 subject to the exceptional set @xmath1014 is empty , and thus @xmath107 is asymptotically pole - free .
this is important as we can now go back to section [ upextension ] and simply repeat our line of reasoning , modulo minor adjustments .
first , from , @xmath1149 valid for any @xmath1041 $ ] and where @xmath1150 but @xmath1151,\ ] ] and thus with and , subject to , @xmath1152 summarizing , proposition [ ou : match ] also applies to the singular transition @xmath190 subject to . now following the logic of section [ upextension ] , we would employ the sequence of transformations @xmath752 but not open lens , compare .
instead , just as in section [ upextension ] , a new parametrix @xmath1153 is introduced on the contracting segment @xmath753 with @xmath578 as in .
the difference between @xmath1153 and @xmath756 as considered in rhp [ nmodel ] is minor , we would simply enforce @xmath1154 this difference amounts to an additional sign in the @xmath1155-entry of which contains the cauchy transform .
however , subsequently all steps are identical to section [ upextension ] and we obtain @xmath1156 with for any given @xmath800 there exist positive constants @xmath801 and @xmath802 such that @xmath1157 this completes the proof of theorem [ bet:2 ] and expansion in corollary [ singmat ] is derived as in the regular transition case .
in order to derive theorem [ res:9 ] we would choose identical steps as those in section [ near ] : we work with the same outer parametrix @xmath748 as in and the same endpoint parametrices @xmath876 in and @xmath1158 in .
the only ( minor ) difference occurs in the definition of @xmath843 , i.e. in the definition of the origin parametrix . for the singular transition we require the jump behavior @xmath1159 instead of
this difference is resolved by an additional sign in the @xmath1155-entry containing the cauchy transform in .
after that we have to replace the matching through @xmath1160 and then simply copy all subsequent steps carried out in section [ near ] .
this leads us to @xmath1161 with for any given @xmath149 there exist positive @xmath150 and @xmath151 such that @xmath1162 and thus completes the proof of theorem [ res:9 ] .
we frequently use the elliptic integrals @xmath1163 and @xmath1164 defined as @xmath1165 for @xmath272 or @xmath273 , the expansions of @xmath1166 are well known , compare for instance @xcite : @xmath1167 with @xmath1168 and @xmath1169 denoting the digamma function .
furthermore , @xmath1170
the jacobi theta functions @xmath1171 used in the text are defined by the formul ( see e.g. @xcite ) where @xmath1172 , @xmath1173 the zeros of the theta functions are located at @xmath1174 up to shifts by vectors of the lattice @xmath1175 ; all zeros are simple . now set @xmath131 and denote @xmath1176 . with @xmath1177,\ \ \
\k=\left(\frac{\theta_2(0,q)}{\theta_3(0,q)}\right)^2\in(0,1),\ ] ] the jacobi elliptic functions @xmath1178 and @xmath1179 are given by the formul@xmath1180 for all values @xmath1181 away from the zeros of @xmath1182 , resp .
@xmath1183 appearing in the denominators .
finally we note that by an application of a landen transformation , @xmath1184 f. dyson , the coulomb fluid and the fifth painleve transendent , appearing in _
chen ning yang : a great physicist of the twentieth century _
, eds . c. s. liu and s .- t .
yau , international press , cambridge , ma , ( 1995 ) , 131 - 146 a. its , a. kapaev , the method of isomonodromy deformations and connection formulas for the second painlev transcendent , _ izv . aka .
ssr ser . mat .
_ * 51 * , pp .
78 ( 1987 ) ( russian ) ; _ math .
ussr - izv .
_ * 31 * , pp .
193 - 207 ( 1988 ) ( english ) a. kitaev , the justification of asymptotic formulas that can be obtained by the method of isomonodromic deformations , _ zap . nation .
lomi _ , * 179 * , pp .
101 - 109 ( 1989 ) ( russian ) ; _ j. soviet math .
_ * 57 * no .
3131 - 3135 ( 1991 ) ( english ) b. suleimanov , the connection between the asymptotics at the different infinities for the solutions of the second painlev equation , _ differentsialnya uravneniya _ * 23 * , pp .
834 - 842 ( 1987 ) ( russian ) ; _ differential equations _ * 23 * , pp .
569 - 576 ( 1987 ) ( english ) | we consider real - valued solutions @xmath0 of the second painlev equation @xmath1 which are parametrized in terms of the monodromy data @xmath2 of the associated flaschka - newell system of rational differential equations .
our analysis describes the transition , as @xmath3 , between the oscillatory power - like decay asymptotics for @xmath4 ( ablowitz - segur ) to the power - like growth behavior for @xmath5 ( hastings - mcleod ) and from the latter to the singular oscillatory power - like growth for @xmath6 ( kapaev ) .
it is shown that the transition asymptotics are of boutroux type , i.e. they are expressed in terms of jacobi elliptic functions . as applications of our results we obtain asymptotics for the airy kernel determinant @xmath7 in a double scaling limit @xmath8 as well as asymptotics for the spectrum of @xmath9 . |
s. c. chapman and b. hnat acknowledge support from the pparc and g. rowlands from the leverhulme trust .
we thank the ace swepam instrument team and the ace science center for providing the ace data .
37 m. k. verma , j. geophys . res . *
101 * , 27543 - 27548 ( 1996 ) .
j .cho and a. lazarian , phys . rev .
lett . * 88 * , 245001 ( 2002 ) .
f. cattaneo , t. emonet , and n. weiss , apj .
* 588 * , 11831198 ( 2003 ) .
n. e. l. haugen and a. brandenburg , phys .
e * 70 * , 026405 ( 2004 ) .
mller , d. biskamp , phys .
lett . * 84*(3 ) , 475 ( 2000 ) .
j. paret and p. tabeling , phys .
lett . * 79 * , 4162 ( 1997 ) .
m. a. rutgers , phys .
* 81 * , 2244 - 2247 ( 1998 ) .
m. l. goldstein and d. a. roberts , phys .
plasmas * 6 * , 41544160 ( 1999 ) .
tu and e. marsch , space sci . rev . * 73 * , 1210 ( 1995 ) .
matthaeus and m. l. goldstein , j. geophys .
res * 87 * , 6011 ( 1982 ) .
l. f. burlaga , j. geophys .
106 * , 15,91715,927 ( 2001 ) .
e. marsch and c .- y .
geophysicae * 12 * , 1127 ( 1994 ) .
l. sorriso - valvo , v. carbone , p. giuliani , p. veltri , r. bruno , v. antoni and e. martines , planet .
space sci . * 49 * , 11931200 ( 2001 ) .
b. hnat , s. c. chapman , g. rowlands , phys .
e * 67 * , 056404 ( 2003 ) .
kraichnan and d. montgomery , rep .
phys . * 43 * , 547 ( 1980 ) .
v. carbone , p. veltri , and r. bruno , phys .
lett . * 75 * , 31103113 ( 1995 ) .
p. veltri , plasma phys .
fusion * 41 * , a787a795 ( 1999 ) .
r. bruno , v. carbone , l. sorriso - valvo , b. bavassano , j. geophys . res . * 108 * , 1130 , 10.1029/2002ja009615 ( 2003 ) .
tu , e. marsch , h. rosenbauer , geophys .
* 17 * , 283 ( 1990 ) .
kraichnan , phys .
fluids * 8 * , 1385 - 7 ( 1965 ) .
m. dobrowolny , a. mangeney , and p. l. veltri , phys .
* 45 * , 144147 ( 1980 ) .
p. goldreich and h. sridhar , astrophys .
j. * 438 * , 763 ( 1995 ) .
m. velli , r. grappin , and a. mangeney , phys .
lett . * 63 * , 18071810 ( 1989 ) .
d. montgomery , m. r. brown , w .h .
matthaeus , j. geophys .
res * 92 * , 282284 ( 1987 ) .
s. r. spangler and l. g. spitler , phys . of plasmas
* 11 * , 1969 - 1977 ( 2004 ) .
g. p. zank , w .h .
matthaeus , phys .
rev . lett . * 64 * , 1243 ( 1990 ) .
f. malara , p. veltri , l. primavera , phys .
e * 56 * , 3508 - 3514 ( 1997 ) .
a. bershadskii and k. r. sreenivasan , phys .
lett . * 93 * , 064501 ( 2004 ) .
a. la porta , g. a. voth , a. m. crawford , j. alexander , e. bodenschatz , nature * 409 * , 10171019 ( 2001 ) .
e. c. stone _
_ , space sci .
rev . * 86 * , 1 ( 1998 ) .
g. stolovitzky , k. r. sreenivasan , and a. juneja , phys .
e * 48 * , r3217-r3220 ( 1993 ) .
u. frisch , _ turbulence .
the legacy of a.n .
( cambridge university press , cambridge , 1995 ) .
r. benzi , s. ciliberto , r. tripiccione , c. baudet , f. massaioli , and s. succi , phys .
e * 48 * , r29-r32 ( 1993 ) .
t. watanabe and t. gotoh , new journal of physics * 6 * , 40 ( 2004 ) . g. ruiz - chavarria , c. baudet , s. ciliberto , physica d * 99 * , 369 - 380 ( 1996 ) .
kraichnan , phys .
. lett . * 72 * , 1016 - 1019 ( 1994 ) .
m. j. aschwanden and c. e. parnell , apj .
* 572 * , 1048 - 1071 ( 2002 ) . | incompressible magnetohydrodynamics is often assumed to describe solar wind turbulence .
we use extended self similarity to reveal scaling in structure functions of density fluctuations in the solar wind .
obtained scaling is then compared with that found in the inertial range of quantities identified as passive scalars in other turbulent systems .
we find that these are not coincident .
this implies that either solar wind turbulence is compressible , or that straightforward comparison of structure functions does not adequately capture its inertial range properties .
the supersonic and super - alfvnic flow of the solar wind offers a unique opportunity to investigate the properties of the magnetized and turbulent plasma .
the transition from a laminar to turbulent flow requires large reynolds number @xmath0 , and its magnetic counterpart @xmath1 , where @xmath2 is the energy injection scale length , @xmath3 is the velocity difference on scale @xmath2 , @xmath4 is the viscosity and @xmath5 is the magnetic diffusivity .
estimations of the hydrodynamic and magnetic reynolds numbers in the solar wind exceed @xmath6@xcite compared with @xmath7 achieved in direct numerical simulations ( dns)@xcite and just few hundreds in some magnetized liquid laboratory experiments @xcite .
the presence of turbulence in the solar wind is strongly suggested by numerous observations .
these include power law power spectra with @xmath8 kolmogorov - like slopes in the kinetic and magnetic energy densities ( eg . , @xcite ) and non - gaussian probability density functions ( pdfs ) ( e.g. , @xcite ) found for fluctuations in the velocity and the magnetic field .
these observations imply that solar wind turbulence shares many of its statistical properties with incompressible isotropic hydrodynamic turbulence @xcite . as a result ,
the turbulent dynamics of the solar wind is often modelled assuming that the plasma density is constant .
this assumption of incompressibility is particularly convenient in analytical and numerical studies of magnetohydrodynamic ( mhd ) turbulence .
the assumption of incompressibility also appears to be in good agreement with the results of compressive mhd simulations where generation of compressive modes from alfvnic turbulence was found to be suppressed@xcite . in the context of the solar wind
, incompressibility has been suggested to be a reasonable approximation for plasma in fast wind streams@xcite .
considerable progress has been made by treating the solar wind as dominated by alfvnic fluctuations ( e.g.,@xcite ) and with nearly incompressible magnetohydrodynamic theory@xcite .
however , statistical features associated with turbulence have also been identified in the density fluctuations derived from the solar wind observations .
a @xmath9 scaling in the omnidirectional wave number spectrum of electron and proton density has been reported as early as the @xmath10s ( see @xcite and the references therein )
. models of the solar wind can offer possible mechanisms for generating strong density fluctuations close to the heliospheric current sheets@xcite .
recent studies suggest that , at least in the slow solar wind , the coupling of alfvnic and magnetosonic mhd modes could not fully describe the nature of these fluctuations .
for example , the ratio @xmath11 was found to be nearly constant and independent from the amplitude of the magnetic field fluctuations @xmath12 in the slow solar wind@xcite .
more recently the density fluctuations were also found to exhibit non trivial , approximately self - affine scaling@xcite similar to that found in other solar wind bulk plasma parameters . in this letter
we examine the scaling properties of the proton density @xmath13 in fast and in slow solar wind .
we focus on a comparison of the scaling properties of fluctuations in density with that of various passive scalars identified in experiments , and direct numerical simulations , of fluid turbulence .
previously , these passive scalars have been shown to exhibit scaling that corresponds closely to that of the magnitude of the magnetic field @xmath14 in the solar wind , although the data interval was not ordered by solar wind speed @xcite . by assuming incompressibility it can be argued @xcite that @xmath14 should act as a passive scalar .
we will for completeness also examine the scaling properties of magnetic field magnitude in fast and slow solar wind .
it is instructive to recall that the dynamics of a passive scalar @xmath15 in a velocity field @xmath16 is given by the advection equation @xmath17 where @xmath18 is the diffusivity .
dynamical and statistical properties of a passive scalar are more tractable analytically , as compared to active fields like velocity , since the equation ( [ adv ] ) is linear in @xmath19 . in the case of hydrodynamics ,
theoretical predictions have also been verified , to some extent , experimentally using tracer particles that do not disturb the flow@xcite .
let us consider the compressible mhd equations for the evolution of the magnetic field and the density : @xmath20 given the assumption of incompressibility ( @xmath21 ) , it immediately follows that @xmath13 should behave as a passive scalar in the turbulent solar wind flow : @xmath22 to investigate the scaling properties of the density @xmath13 and the magnetic field magnitude @xmath14 we use @xmath23 seconds averaged data from the ace spacecraft @xcite set spanning from @xmath24 to @xmath25 .
this interval includes dates previously considered in ref.@xcite .
the slow and fast solar wind are known to exhibit distinct phenomenology ( e.g. , @xcite ) .
we thus split the data into slow and fast solar wind sets using @xmath26km / s wind speed as a separation criteria .
the resulting data sets consist of @xmath27 samples for the slow wind and @xmath28 for the fast wind .
we apply structure function@xcite analysis to fluctuations in density for slow and fast wind separately .
generalized structure functions @xmath29 of fluctuations in say , the density @xmath30 , on timescale @xmath31 are defined as moments @xmath32 through @xmath33 where the ensemble @xmath34 is taken in the time domain @xcite . if scaling is present in the time series we expect these to show a power law dependence on the temporal scale @xmath31 , i.e. , @xmath35 .
finite , experimental data sets include a small number of extreme events ( outliers ) that , due to poor statistics , may obscure the correct scaling of the high order moments . here , we will exclude these events by the use of _ conditioning _ @xcite .
this approach puts a limit on the range of fluctuations used in computing structure functions .
this limit is varied with the temporal scale @xmath31 to account for the growth of range with temporal scale in the signal . in our case
we defined this threshold as @xmath36 , where @xmath37 is a standard deviation of fluctuations on temporal scale @xmath31 .
we stress that conditioning improves the scaling where it already exists but does not enforce it on the investigated data , if the applied threshold is sufficiently large . in practice , for the limit chosen here , we eliminate less then @xmath38 of the data points .
figure [ fig1 ] shows the structure functions @xmath29 plotted versus @xmath31 on logarithmic axes for orders @xmath39 for fluctuations in density in the slow solar wind .
the plot shows a scaling region extending from @xmath40 minutes to @xmath41 hours ( @xmath42 decades on the logarithmic scale ) .
the quality of the scaling deteriorates when fast wind streams are considered .
the scaling regions can be still identified but they extend only @xmath43 decade from @xmath44 minutes to @xmath45 hour .
it has been empirically shown that extended self similarity ( ess ) approach can considerably extend the region of scaling in structure functions@xcite .
the method is based on the assumption that @xmath46 .
this suggests that scaling should emerge when the quantity @xmath29 is plotted as a function of @xmath47 .
the scaling exponent @xmath48 can then be obtained straightforwardly from the relation @xmath49 . for the density fluctuations in the solar wind , we find that @xmath50 is close to unity and we plot @xmath29 versus @xmath50 on logarithmic axes for fluctuations in the density in slow and fast solar wind , in figures [ fig2 ] and [ fig3 ] respectively .
these figures demonstrate that we can extend scaling in the density in both the slow and the fast wind streams to over @xmath51 decades when ess is applied .
ess was previously applied to the magnitude of magnetic field in an undifferentiated interval of solar wind@xcite and for completeness we give the analysis in slow and fast wind here .
figures [ fig4 ] and [ fig5 ] show @xmath29 versus @xmath50 on logarithmic axes for fluctuations in the magnitude of magnetic field .
previously , the scaling exponents were obtained by considering @xmath29 versus @xmath52@xcite and we have verified that these closely correspond to the values found here .
the local slopes of the third order structure functions @xmath50 then yield an estimate of the exponent @xmath53 and these are detailed in table [ tab1 ] .
we can now directly compare the scaling found for the density @xmath13 with that identified for quantities acting as passive scalars and that of the magnetic field magnitude @xmath14 .
the resulting functional form of the scaling exponents @xmath54 for the density and magnetic field magnitude in slow and fast solar wind are shown in figures [ fig6 ] and [ fig7 ] respectively .
for comparison , the scaling exponents obtained for passive scalars from the dns @xcite and the the wind tunnel experiment@xcite are also shown .
we immediately see that whereas the exponents for magnetic field magnitude and the passive scalars fall close to each other on these plots ( as also reported in @xcite ) they are distinct from those obtained for the solar wind density in both slow and fast solar wind .
the fluctuations in @xmath55 , for slow and fast wind , and that of the passive scalars exhibit multi - fractal scaling .
intriguingly , the density fluctuations are nearly self - affine with scaling exponent @xmath56 that differs in slow and fast wind .
the values of these exponents are : @xmath57 and @xmath58 .
this suggests one of two possible conclusions .
the first of these is that the turbulent solar wind is compressible and equation ( [ mhdrincomp ] ) does not hold , so that the density is an active quantity .
this is rather surprising and calls into question the significant body of work on mhd turbulence in the solar wind that relies on the assumption of incompressibility ( for example , @xcite ) .
this also invalidates the arguments in ref.@xcite required to cast the advection equation for @xmath14 in the form of a passive scalar which require incompressibility , yielding : @xmath59 the second possibility is that commonality of scaling ( as measured by structure functions ) does not imply shared phenomenology , and as a corollary , an absence of such a commonality does not imply distinct phenomenology .
the fact that several quantities share the same structure functions is then coincidental rather than expressing some universality of fluid turbulence .
this implies that the structure functions of a single quantity do not fully capture the phenomenology of a given turbulent system .
this has implications for analysis of turbulence which has been achieved through comparison between ( multi - fractal ) models of turbulence and the data , via the structure functions ( see , for example , @xcite ) .
.exponents @xmath53 obtained from figures [ fig3 ] . [ cols="^,^,^,^",options="header " , ] [ tab1 ] a possible resolution may be found in the form of equations ( [ mhdrincomp ] ) and ( [ mhdbincomp ] ) .
it is known that the molecular diffusivity can change the inertial range scaling properties of a passive scalar@xcite .
the observed differences could then be due to the presence of the diffusive term in the magnetic field equation ( [ mhdbincomp ] ) and its absence in the density equation ( [ mhdrincomp ] ) .
this may account for the different scaling found in these quantities .
intriguingly , equation ( [ mhdrincomp ] ) , when written for the moments of fluctuations in @xmath13 , has no explicit dependence on the order of the moment @xcite @xmath60 ( \delta \rho)^n = 0 , \label{mhdstrf}\ ] ] where @xmath61^n$ ] . in this case fluctuations in density should be simply those imposed on the initial condition @xmath62 .
these may be mediated via large scale coherent structures ( shocks and coronal mass ejections ) .
this may suggest a solar origin of these fluctuations in the density
. it may thus be informative to attempt to relate the scaling found in the density fluctuations in the solar wind , with that of the solar corona .
the power law scaling of x - ray flux from solar flares@xcite is intriguing in this regard . |
mean field approaches provide a reliable foundation for an approximate solution of the nuclear many - body problem . nowadays
, a lot of efforts is applied to move beyond the mean field approximation and account for missing correlations .
special attention is paid to the restoration of symmetries broken in mean field approaches @xcite , for example using projection techniques .
an alternative way is to develop a theory in which the trial wave functions preserve certain symmetries . in particular , this can be achieved by multiconfiguration methods widely used in various applications , including atomic , molecular , and condensed matter physics .
when the interaction is known , this kind of approach provides a very accurate description of a system . in a previous work @xcite , we proposed in the nuclear physics context a variational approach based on multiparticle - multihole ( @xmath5 ) proton and neutron configuration mixings that uses the two - body finite - range density - dependent gogny interaction in a fully microscopic way .
as a first step , the method was applied in a particular case : pairing correlations ( proton - proton and neutron - neutron ) were investigated for the ground states of several even - even tin isotopes . a pioneering work using the skyrme siii interaction for the mean field part and a schematic contact interaction for the residual part has been directed to the description of @xmath6-isomers in the @xmath7hf mass region @xcite .
the main objective of the present study is twofold : first , to test the ability of @xmath5 multiconfiguration approach to describe low - lying nuclear states and second , to discuss the statistical features of highly excited configurations in nuclei . with this aim in view ,
a few silicon isotopes have been chosen and their excited states calculated with the @xmath5 approach . concerning highly excited configurations ,
the leading idea is to investigate whether exponential convergence behavior , revealed in the standard shell model ( sm ) approach @xcite , also emerges from variational ( @xmath5 ) configuration mixing methods @xcite in which single particle excitations are not restricted to a single major shell .
as practical calculations inevitably require some truncation of the orbital space and order of excitation , the @xmath5 method proposes a promising scheme that allows one to predict the energies of low - lying states in a very accurate way .
the possible use of statistical properties of highly excited states , which display generic signatures of quantum chaos close to random matrix theory , drastically reduces the sizes of the explicit diagonalizations . in the literature ,
one finds other approaches proposing different algorithms as , for example , the density matrix renormalization group @xcite or monte carlo techniques @xcite , selecting the most relevant configurations for the description of many - body states .
the paper is organized as follows .
the main characteristics of the @xmath5 multiconfiguration approach applied in this study are presented in section [ section1 ] . in section [ section2 ] ,
the results obtained with the @xmath5 approach for low - lying states in @xmath0si are presented and compared with those derived from a five - dimensional approximate generator coordinate method ( gcm ) approach ( subsections [ section2a ] and [ lls ] ) .
differences between theoretical and experimental results are discussed . in particular , the systematic energy shift found in @xmath4si is analyzed in terms of proton - neutron matrix elements .
subsection [ pn ] highlights the crucial role played by the residual interaction between protons and neutrons .
section [ section3 ] is devoted to the analysis of the statistical properties of highly excited configurations , taking @xmath2si as an example .
conclusions and perspectives are given in section [ section4 ] .
in this part , we discuss a few characteristics of the @xmath5 configuration mixing approach . a general derivation of the corresponding formalism in nuclear physics , with two - body density - dependent interactions , has been introduced in ref .
it is worth to recall here the basics of the method not only to fix notations but also to provide an alternative formulation of equations in terms of a core + valence space " description . in the @xmath5 configuration mixing method ,
the effective hamiltonian is defined as a functional @xmath8 of the single - particle density matrix @xmath9 . in eq .
( [ eq1 ] ) , the hamiltonian contains a kinetic term @xmath10 ( which includes the one - body center - of - mass corrections ) and a two - body density - dependent potential term @xmath11 ( which includes the coulomb potential for protons as well as the two - body center - of - mass corrections ) . in our study ,
the d1s gogny interaction @xcite is adopted .
the trial wave functions @xmath12 that describe nuclear stationary states are expressed as linear combinations @xmath13 of direct products @xmath14 of proton and neutron slater determinants , @xmath15 and @xmath16 respectively , containing _ a priori _ any multiple _ p - h _ excitations that respect conserved quantum numbers .
( [ eq2 ] ) contains two sets of unknown parameters , the mixing coefficients @xmath17 and the single - particle orbitals .
both are determined by applying a variational principle to a functional @xmath18 related to the total energy of the system , @xmath19 where @xmath20 and @xmath21 are lagrange multipliers and @xmath22 possible additional constraints that we will leave out in the following .
one assumes that the one - body density @xmath9 entering the effective hamiltonian @xmath23 is the correlated one : @xmath24 .
the minimization of @xmath25 with respect to the @xmath17 leads to a non - linear secular equation that is equivalent to a diagonalization problem in the multiconfigurational space of a hamiltonian matrix @xmath26 , @xmath27 in eq .
( [ eq4 ] ) , the matrix @xmath28 contains contributions of the hamiltonian @xmath29 and of rearrangement terms that come from the density dependence of the interaction , @xmath30 where the summation over @xmath31 specifies the proton and neutron contributions to the generalized rearrangement terms with coefficients @xmath32 .
it is the presence of the rearrangement terms that transforms eq .
( [ eq4 ] ) into a non - linear eigenvalue problem . as can be seen from eq .
( [ e889 ] ) , the rearrangement terms contain contributions associated with the one - body density @xmath9 and the two - body correlation matrix @xmath33 defined by @xmath34 in appendix [ a1 ] , we express eqs .
( [ eq4])-([eq5 ] ) in a `` core + valence space '' scheme , a form which is explicitly used in the present study .
the minimization of @xmath25 with respect to the single - particle orbitals leads to inhomogeneous hartree - fock ( hf ) equations which depend on the amount of correlations contained in @xmath12 @xcite , @xmath35 = g(\sigma ) , \label{eqad1}\ ] ] where @xmath36 in eq.([eqad1 ] ) , @xmath37 is the one - body mean - field hamiltonian built with the one - body density @xmath9 and the two - body correlation matrix @xmath33 : @xmath38 explicit expressions for the fields @xmath39 , @xmath40 and @xmath41 are given in ref .
the basis that diagonalizes @xmath37 provides proton and neutron single - particle orbitals with energies @xmath42 . the diagonal part of eq .
( [ eq5 ] ) , that is the one obtained by taking @xmath43 , can be easily written in terms of the @xmath42 . in the limit where the mixing reduces to the sole hf configuration ,
one recovers the standard hf expression for @xmath42 , @xmath44 in eq .
( [ eqqq6 ] ) , summations over @xmath45 and @xmath46 run over the hole states . as discussed in ref.@xcite , a fully self - consistent solution of the multiconfiguration approach consists of solving simultaneously both equations ( [ eq4 ] ) and ( [ eqad1 ] ) .
it is important to notice that , in our approach , the mean field and beyond mean - field descriptions are obtained in a consistent way since they both follow from variations of the same functional .
a standard way to solve both equations is to use an iterative procedure . starting from a hf solution ( characterized by @xmath47 and @xmath48 ) that provides an initial set of single particle orbitals , one builds the multiconfiguration basis and solves the configuration mixing eq .
( [ eq4 ] ) .
then , from the mixing coefficients obtained from eq .
( [ eq4 ] ) , one calculates @xmath9 and @xmath33 and solves eq .
( [ eqad1 ] ) . with the new set of single particle orbitals , one redoes the procedure until convergence of @xmath9 and @xmath33 . in this way ,
the single - particle orbitals contain effects coming not only from the mean - field built with the one - body density matrix @xmath9 but also from the correlations contained in the matrix @xmath33 . in general , the actual solution of eq .
( [ eqad1 ] ) is a very difficult task . in a previous study dedicated to the particular case of pairing - type correlations in the ground states of even - even tin isotopes @xcite ,
one has solved eq .
( [ eqad1 ] ) in an approximate way , by neglecting @xmath33 . in the present study of even - even silicon isotopes , as a first step , we have solved only the configuration interaction ( ci ) part ( eq . ( [ eq4 ] ) ) of the multiconfiguration approach . as already pointed out in introduction , multiconfiguration methods are able to preserve certain symmetries . concerning particle numbers , the correlated wave function ( [ eq2 ] ) is a superposition of slater determinants that conserves exactly the numbers of protons and neutrons . as to angular momentum
, the multiconfiguration equations have been solved in the present study only for spherical nuclear configurations , with valence spaces comprising complete spherical subshells . as a consequence ,
nuclear states have a good total angular momentum @xmath49 .
actually , since our formalism has been developed in axial symmetry in order to be able to introduce quadrupole deformation explicitly , the only conserved quantum numbers we have are the projection @xmath6 of the total angular momentum and the parity .
the conserved quantum number @xmath49 is then obtained in the usual manner by performing calculations for successive values @xmath50 and identifying degenerate multiplets .
let us note that the finite axial harmonic oscillator bases used for expanding single - particle states ( see below ) are defined with a spherical truncation .
this section is devoted to the description of the low - lying spectroscopy of @xmath0si using the @xmath5 configuration mixing approach . in the first part of this analysis
, we investigate the mean - field properties of the ground states provided by the hartree - fock - bogoliubov ( hfb ) approach using the same d1s gogny interaction and compare them with the results of a five - dimensional approximate gcm approach .
technically , the single - particle states introduced in the different approaches ( hf , hfb , @xmath5 ) are expanded onto the harmonic oscillator ( ho ) bases . in the present work , @xmath51=11 major spherical shells
have been taken .
this basis size has been found sufficient to ensure the convergence of all the results obtained in this work in the three approaches mentioned above .
for instance , the convergence of low - lying state energies in the @xmath5 configuration mixing approach has been achieved within an accuracy better than 0.1 kev . in order to investigate the properties of the ground states of @xmath0si , we start with triaxial hfb calculations constrained according to the dimensionless deformation parameters @xmath52 and @xmath53 , @xmath54 in eq .
( [ eqad4 ] ) , @xmath55 , @xmath56 and @xmath57=1.2 fm .
[ fig1 ] displays the triaxial hfb potential energy surfaces ( pes ) of @xmath0si in the @xmath52 and @xmath53 degrees of freedom .
isolines associated with total energy are indicated with a numbering corresponding to the height of the potential ( in mev ) relative to the minimum of the hfb potential ( dark blue online ) for each nucleus .
the @xmath58 and @xmath59 values for the hfb solution of lowest energy are indicated in the second and the third columns of table [ tab6d ] . .
@xmath52 and @xmath53 deformation properties of the @xmath0si ground states from the hfb and 5dch approaches and the d1s gogny interaction .
[ cols="^,^,^,^,^",options="header " , ]
one of the main issues raised by multiconfiguration approaches is the number of relevant configurations for describing low - lying states . because of the proton - neutron excitations
, this number can rapidly explode .
in realistic calculations , one has to think about truncations based on physical arguments .
when going beyond the mean - field with the @xmath5 approach , it is assumed that short - range correlations have already been taken into account through the effectiveness of the nucleon - nucleon interaction used ( the d1s gogny interaction in our case ) .
our aim is to treat explicitly the long - range correlations corresponding to the attractive part of the nucleon - nucleon interaction .
two standard types of truncation can be proposed , independently of the choice ( or not ) of a valence space : a truncation on the order of the excitations and/or a truncation on the configuration excitation energies .
both types of truncations seem to be reasonable in the present context . in order to define appropriate truncations based on relevant physics argument
, we discuss below the behavior of highly excited configurations .
we have followed the direction used in refs . @xcite where the analysis was based on the properties of the strength function associated with the slater determinants . using second quantization and the standard wick s theorem decomposition , the hamiltonian @xmath60 , eq .
( [ eq5 ] ) , is the sum of an independent particle part @xmath61 ( one - body ) and a residual part @xmath62 ( one - body and two - body ) , @xmath63 the eigenfunctions @xmath64 of the unperturbed hamiltonian @xmath65 , @xmath66 describe noninteracting fermions . in the basis @xmath67
, the residual interaction @xmath68 has both diagonal , @xmath69 , and off - diagonal , @xmath70 , matrix elements : @xmath71 .
full diagonalization leads to the stationary states @xmath72 and their energies @xmath73 , @xmath74 the eigenfunctions @xmath75 may have a complicated structure in the original basis @xmath64 , @xmath76 where @xmath77 is a unitary matrix : @xmath78 a completely delocalized wave function @xmath72 would have a number of relevant components close to the total dimension @xmath79 of the multiconfiguration space ( for given exact quantum numbers ) . in this limit
the typical magnitude of each component is @xmath80 . in general , a number @xmath81 of principal components @xmath67 characterizes the delocalization of a state @xmath72 in the given basis ( [ eq8 ] ) .
indeed , a two - body interaction can not couple configurations differing by more than two particle states which implies a band - like hamiltonian matrix , favoring the localization of eigenfunctions in the hilbert space .
conversely , the fragmentation of simple basis states over the energy spectrum can be provided by the strength function defined by the quantity @xmath82 si . the x - axis corresponds to the number of centroids ordered by increasing energies.,height=264 ] the strength function contains rich information but requires the knowledge of the full nuclear spectrum .
fortunately , one can study the main characteristics of the system from the first and the second moment of the strength function which does not require the actual diagonalization of @xmath26 .
these two moments are the centroid , @xmath83 , and dispersion , @xmath84 , of the state distribution , given by @xmath85 and @xmath86 respectively .
the centroid @xmath87 coincides with the unperturbed energy @xmath88 , whereas the dispersion depends only on the off - diagonal matrix elements @xmath89 .
+ below we discuss the example of the @xmath2si isotope .
similar calculations have been done for other nuclei with similar conclusions . at this point
, it is important to recall that the @xmath5 configuration mixing formalism presented in section [ section1 ] has been developed , in practice , using an axially - deformed harmonic oscillator basis . in order to preserve the spherical symmetry ,
all the @xmath5 calculations displayed in this study have been done for @xmath90 . in the following , our analysis is done in terms of different projections @xmath6 of a given angular momentum @xmath49 .
all configurations of the @xmath91-shell ( up to @xmath92 ) have been introduced in the wave functions ; the maximum size of the hamiltonian matrices that have been fully diagonalized was @xmath9390 000 @xmath94 90 000 .
[ fig10 ] displays the values of the centroid energies @xmath83 , where centroids are labeled by n@xmath95 and ordered by increasing energy .
one observes a characteristic behavior for all values of @xmath6 from 0 to 4 .
the lowest centroid energy , the one associated with the @xmath96 configuration , is -239.203 mev .
only few configurations have small excitation energy .
the level density increases rapidly with excitation energy , and most configurations are located in the range @xmath97 $ ] mev . the final increase , beyond -160 mev ,
is an artifact of the finite valence space that reduces the number of possible highly excited states .
si strength functions associated with @xmath98 and 4@xmath99 components in linear scale ( upper panel ) and logarithmic scale ( lower panel ) .
logarithmic representation is shown in lower panel .
calculations have been performed with a 0.1 mev bin.,title="fig:",height=264 ] si strength functions associated with @xmath98 and 4@xmath99 components in linear scale ( upper panel ) and logarithmic scale ( lower panel ) .
logarithmic representation is shown in lower panel .
calculations have been performed with a 0.1 mev bin.,title="fig:",height=264 ] fig .
[ fig11 ] shows the dispersions @xmath84 of eq .
( [ eq13 ] ) for all configurations characterized by @xmath100 and 3 and their excitation energies @xmath101 ( in mev ) .
the color code indicates the number of configurations in a given excitation energy and @xmath84 bin . as expected , the structure of the configuration distribution is different for different values of @xmath6 but most configurations are concentrated along a central line , and characteristic structures appear in the most dense areas .
such areas are more pronounced in the most dense zone in the @xmath102 case for all values of @xmath101 .
a large majority of configurations display a dispersion @xmath84 that is rather constant along the spectrum ; its mean value can be evaluated to be @xmath939 mev .
this important result which implies that strength functions can be characterized by essentially the same width is the first indication of expected exponential convergence .
si slater determinant distributions @xmath103 ( upper panel ) and the @xmath2si locally normalized strength functions @xmath104 ( lower panel ) for @xmath98 and 4@xmath99 components .
calculations have been performed with a 0.1 mev bin.,title="fig:",height=264 ] si slater determinant distributions @xmath103 ( upper panel ) and the @xmath2si locally normalized strength functions @xmath104 ( lower panel ) for @xmath98 and 4@xmath99 components .
calculations have been performed with a 0.1 mev bin.,title="fig:",height=264 ] the strength function of @xmath2si is presented in fig .
[ fig7a ] for different values of @xmath6 as a function of the energy counted from the corresponding centroid , @xmath105 .
the value of the bin is set to 0.1 mev .
an average over all configurations has been done in order to reveal possible generic behavior .
the upper panel displays the strength function in linear scale and the lower panel in logarithmic scale .
the central part of the distribution is intermediate between a gaussian and a lorentzian one , whereas the behavior in the wings is found to be a decreasing exponential .
this is clearly visible in the logarithmic scale . in agreement with @xcite ,
the coupling of highly excited configurations with low - energy eigenfunctions therefore exhibits an exponential regime and the 3 @xmath33 " rule characterizing the start of the exponential convergence , which is important in practical calculations of the level density by methods of statistical spectroscopy @xcite , seems to hold here also . in the case of the @xmath102 projection ,
one observes two unusual bumps at energies @xmath106 40 mev and 60 mev .
actually , the strength function ( [ st1 ] ) combines two types of information : the density of configurations and the mixing coefficients whose values are determined by the nucleon - nucleon interaction . in order to disentangle these effects , we calculate the density of configurations @xmath107 and the locally normalized strength function @xmath108 where @xmath109 .
these quantities are shown in fig .
[ fig7c ] ( the density of configurations , upper panel ( a ) , and the locally normalized strength function in logarithmic scale , lower panel ( b ) ) .
the configuration density has a regular shape expected for a finite hilbert space whereas the bumps of the strength function in fig .
[ fig7a ] appear magnified on the normalized strength function plot ( lower panel ) . one also notices that the coupling of matrix elements is fully exponential ( linear on logarithmic scale ) , if one forgets about the bumps .
as the density of configuration is quasi - linear in logarithmic scale at large energy , one understands the exponential convergence behavior of the strength function observed on fig .
[ fig7a ] .
the two bumps come from the specific interaction rather than from the statistics .
the matrix elements that are responsible for the bumps correspond to the same proton - neutron me discussed in section [ lls ] for @xmath4si , which appear to be too large at the level of the approximate resolution of the @xmath5 configuration mixing method presented in this study . reducing by hand the values of these proton - neutron me makes the bumps disappear .
hence , the study of the chaotic behaviour of highly excited slater determinants can be considered as an additional way of highlighting not well - calibrated coupling me . in other words
, statistical properties may serve as an additional criterion in the validation process of phenomenological effective interactions .
this result is consistent with and confirms our previous discussion of section [ lls ] .
this analysis is similar in spirit to the statistical search of interaction matrix elements responsible for the equilibrium prolate deformation @xcite .
the exponential convergence observed in fig . [ fig7c ] is an interesting feature that might be evaluated analytically in the particular case of the finite - range gogny interaction .
for example , taking the single - particle states as plane waves , the two - body matrix elements of the brink - boecker part of the gogny force read : @xmath110 where we have used the conservation law @xmath111 .
( [ w1 ] ) reduces to the fourier transform of a gaussian , @xmath112 with @xmath113 and @xmath114 .
thus , the matrix element behaves , in the case of the gogny interaction , as a decreasing exponential with respect to excitation energy .
the value of the two ranges introduced in the gogny interaction may serve as a guide to decide an upper limit where the exponential convergence regime settles .
however , this behavior exists for any physically reasonable interaction as revealed in shell model calculations @xcite . below we give three examples of exponential convergence behavior with increasing excitation energy of configurations , namely for the total energy , for the components of the correlated wave functions and for the occupation probabilities .
[ fig20 ] displays the evolution of the total energies of the ground and excited states in @xmath2si , according to the number of configurations in the correlated wave function , ordered by increasing centroid energies .
the four plots correspond to the different values of the total momentum projection @xmath6 . in all cases , the total energy changes rapidly when only a few slater determinants with lowest energy are included .
then , for a larger number of configurations a smooth regime settles , a behavior independent of the value of @xmath6 . in fig .
[ figsiwf ] , the evolution of the global components , @xmath115 , eq .
( [ eqqq1 ] ) , of the wave functions of the @xmath116 and @xmath117 states in @xmath2si is displayed as a function of the number @xmath118 of centroids ordered by centroid energy . for @xmath1199 ,
the @xmath115 are smaller than 10@xmath120 and they are not shown .
the evolution of the occupation probabilities for @xmath121 , @xmath122 and @xmath123 orbitals according to @xmath118 is displayed on fig .
[ figocc ] .
as can be seen from the two figures , both the @xmath115 and the occupation probabilities display the exponential convergence . with all exponents being close
, one can loosely interpret this behavior as a signature of thermalization in a self - bound mesoscopic system . , @xmath122 and @xmath123 orbitals as function of @xmath124.,height=264 ]
in this work , we have investigated the application of multiconfiguration methods for the description of low - energy nuclear spectroscopy .
a few even - even silicon isotopes have been studied , using the d1s gogny interaction . at this stage , only the configuration mixing part of the method has been put into place ; the renormalization of single - particle orbitals due to correlations has not been discussed in this paper .
the results for the positive parity states in low - lying spectroscopy of @xmath0si have been found in rather good agreement with experiment , taking into account the fact that the d1s gogny interaction has not been _ a priori _ fitted to be employed in such a kind of approach . in particular , from the study of @xmath4si , it has been found that , at the level of approximate resolution of the @xmath5 configuration mixing method ( no renormalisation of orbitals ) a few residual proton - neutron matrix elements of pairing type , that are not constrained in the fitting of the gogny interaction , might disturb the reproduction of excitation energies . in relation to this , the importance and magnitude of the proton - neutron residual interaction has been discussed . the question of the pertinent configurations that have to be introduced in the mixing has also been addressed . statistical generic behavior of highly excited configurations has been put forward . in particular ,
the exponential convergence already revealed in shell model studies has been confirmed in our approach .
this is an encouraging feature that may help to handle the very large number of configurations that appear in multiconfiguration methods , in particular in nuclear physics where two kinds of particles exist .
the present study proposes interesting and challenging issues at different levels .
the renormalization of orbitals under the influence of correlations is an important question that will be analyzed in further studies . in atomic physics , this renormalization has been proved to play a key role in strongly correlated systems . even though the associated orbital equation , see eq .
( [ eqad1 ] ) , looks simple , its exact solution is far from being trivial . at present , in the most advanced applications to atomic physics , it is solved approximately as the correlation term that depends on the two - body density matrix can be very complicated .
a second issue concerns the improvement of the gogny interaction in order to be able to use it not only in hfb , rpa and gcm - type methods but also in multiconfiguration approaches .
work is in progress in this direction .
the third issue deals with the generic behavior of highly excited configurations .
the exponential convergence and corresponding extrapolations can be of considerable help for controlling in a safe way possible new truncation schemes introducing explicitly only pertinent configurations .
to this aim , a formalism of the feshbach type projection might be quite useful .
it is worth mentioning that the analysis of statistical regularities allowed us to identify specific matrix elements of the hamiltonian responsible for spectroscopic inadequacy ( discussion on @xmath4si ) . for the specific goal of nuclear spectroscopy
, the investigation of transition probabilities and negative parity states in silicon isotopes would be of great interest and may provide essential information on the residual interaction and properties of single - particle orbitals , including nuclei far from stability .
n.p . would like to thank the national superconducting cyclotron laboratory at michigan state university for hospitality and b.a .
brown for providing the shell model matrix elements and results that were of great help for the analysis presented in this work .
is also grateful to m. girod and j. libert for providing with 5dch results and a. zuker and m. dufour for interesting discussions .
v.z . acknowledges the support from the nsf grants phy-0758099 and phy-1068217 .
part of the calculations were carried out on ccrt supercomputers of cea - dam ile de france .
eqs . ( [ eq2 ] ) and ( [ eq4 ] ) are quite general .
as mentioned in the introduction , the hilbert space has to be truncated in any realistic calculation .
several truncation schemes , supported by physics arguments , can be utilized , for example a limitation on the excitation order of @xmath5 configurations , a limitation in the number of single - particle states used for the configuration mixing , etc .
this appendix deals with a truncation that corresponds to the description in terms of core + valence space " . in this approach
, the system , that comprises @xmath125 nucleons of each isospin , is separated into : + - an even - even core where the @xmath126 lower individual states are fully occupied for each isospin ; + - a set of active orbitals containing @xmath127 particles for each isospin ; + - a set of unoccupied higher - energy individual states for each isospin .
in addition , the hamiltonian @xmath130 , eq . ( [ eq1 ] )
, can be written by separating isospin contributions as @xmath131 then , from eq .
( [ eq5 ] ) , @xmath132 + \hat{v}^{\pi \nu } ( \rho ) .
\label{eqq3b}\ ] ] where the generalized rearrangement coefficients @xmath133 are given by @xmath134 with @xmath135 the single - particle wave functions .
the difficulties in the calculation of @xmath136 come from the evaluation of the rearrangement field @xmath137 from the correlated wave function ( [ eq2 ] ) : @xmath138 the expressions ( [ eqq2 ] ) , ( [ eqq3b ] ) and ( [ e889 ] ) are specific to the d1s gogny interaction as only the proton - neutron terms are generated by the density - dependent part of the interaction and the density - dependence is a contact interaction . in the limit of no configuration mixing , one recovers the standard hf expression for the rearrangement term . from eq .
( [ eqq3b ] ) , one derives explicit expressions for @xmath139 in terms of the core + valence space " formulation . in the following evaluation of @xmath140 ,
only the terms carrying the core contributions are given .
the proton and neutron diagonal contributions are obtained for @xmath141 and @xmath142 as @xmath143 which yields @xmath144 in eq .
( [ e81 ] ) , @xmath145 stands for the occupied single - particle state in the @xmath146 slater determinant .
core + valence space " separationthen leads to : @xmath147 in eq .
( [ eqq4 ] ) , the first and third terms are pure core contributions .
the second and forth terms are the contributions from the valence space .
the mixed fifth term includes single - particle orbitals of both the core and the valence space .
the proton - neutron diagonal contribution is obtained for @xmath148 and @xmath149 , as @xmath150 expansion of eq .
( [ e84 ] ) yields @xmath151 making the `` core + valence space '' separation leads to @xmath152 in eq .
( [ e86 ] ) , the first term comes from the core and the second one from the valence space .
the last two terms express the coupling between the core and the valence particles .
the proton or neutron one - body contributions arise when @xmath153 and @xmath154 differ in one particle state and @xmath142 or when @xmath155 and @xmath156 differ from one particle state and @xmath157 , respectively .
then , the contributions to @xmath26 are @xmath158 and @xmath159 in eqs .
( [ eqq5])-([eqq5b ] ) , the indices @xmath160 and @xmath161 refer to the single - particle state different between the configurations @xmath162 and @xmath163 . the first term in eqs .
( [ eqq5])-([eqq5b ] ) corresponds to the core contribution and the second one to the contribution of the valence space . as for the proton and neutron one - body contributions ,
the proton - neutron one - body contribution arises when @xmath153 and @xmath154 differ by one particle state and @xmath142 or when @xmath155 and @xmath156 differ by one particle state and @xmath157 .
similarly to eqs .
( [ eqq5])-([eqq5b ] ) , the proton - neutron non - diagonal one - body contribution is then given by @xmath164 in eq .
( [ eqq6 ] ) , the indices @xmath160 and @xmath161 refer to the single - particle states that differ between the configurations @xmath162 and @xmath163 .
still , the first term describes the core contribution and the second one the contribution of the valence space .
50 p. ring and p. schuck , _ the nuclear many - body problem _ ( springer - verlag , new - york , 1980 ) .
m. anguiano , j.l .
egido and l.m .
robledo , nucl . phys . *
a696 * , 467 ( 2001 ) .
m. anguiano , j.l .
egido and l.m .
robledo , phys . lett . *
b545 * , 62 ( 2002 ) .
m. bender , g. f. bertsch , and p .- h .
heenen , phys .
c * 73 * , 034322 ( 2006 ) .
m. bender and p .- h .
heenen , phys .
c * 78 * , 024309 ( 2008 ) .
rodriguez , j.l .
egido , l.m .
robledo , and r. rodriguez - guzman , phys .
c * 71 * , 044313 ( 2005 ) .
r. rodriguez - guzman , l. egido , and l.m .
robledo , nucl . phys . * a709 * , 201 ( 2002 ) . t.r . rodriguez and j.l .
egido , phys . rev .
c * 81 * , 064323 ( 2010 ) . j .-
delaroche , m. girod , j. libert , h. goutte , s. hilaire , s. pru , n. pillet , and g.f .
bertsch , phys .
c * 81 * , 014303 ( 2010 ) .
j. libert , m. girod and j .- p .
delaroche , phys .
c * 60 * , 054301 ( 1999 ) .
n.pillet , j .- f .
berger , and e. caurier , phys .
c * 78 * , 024305 ( 2008 ) .
n. pillet , p. quentin , and j. libert , nucl . phys . *
a697 * , 141 ( 2002 ) .
m. horoi , b.a .
brown , and v. zelevinsky , phys .
c * 50 * , r2274 ( 1994 ) .
v. zelevinsky , b.a .
brown , n. frazier , and m. horoi , phys . rep .
* 276 * , 85 ( 1996 ) .
m. horoi , a. volya , and v. zelevinsky , phys .
82 * , 2064 ( 1999 ) .
m. horoi , b.a .
brown , and v. zelevinsky , phys . rev .
c * 67 * , 034303 ( 2003 ) .
m. horoi , j. kaiser , and v. zelevinsky , phys .
c * 67 * , 054309 ( 2003 ) . j. decharg and d. gogny , phys
c * 21 * , 1568 ( 1980 ) ; j .- f .
berger , m. girod , and d. gogny , comp .
. comm . * 63 * , 365 ( 1991 ) .
s. pittel and b. thakur , rev .
de fisica * 55 * suppl .
, 108 ( 2009 ) .
s. pittel and n. sandulescu , phys .
c * 73 * , 014301 ( 2006 ) .
j. rotureau , n. michel , w. nazarewicz , m. ploszajczak , and j. dukelsky , phys .
lett . * 97 * , 110603 ( 2006 ) .
t. papenbrock and d.j .
dean , j. phys .
g * 31 * , s1377 ( 2005 ) .
j. dukelsky , s. pittel , s. dimitrova and m. stoitsov , phys .
c * 65 * , 054319 ( 2002 ) .
t. otsuka , m. honma , t. mizusaki , n. shimizu and y. utsuno , prog . part .
* 47 * , 319 ( 2001 ) .
a.obertelli , s. pru , j .-
delaroche , a. gilibert , m. girod , and h. goutte , phys .
c * 71 * , 024304 ( 2005 ) . t. otsuka , r. fujimoto , y. utsuno , b.a .
brown , m. honma , and t. mizusaki , phys .
lett . * 87 * , 082502 ( 2001 ) .
t. otsuka , y. utsuno , r. fujimoto , b.a .
brown , m. honma , and t. mizusaki , eur .
j. a * 13 * , 69 ( 2002 ) .
smirnova , a. de maesschalck , a. van dyck , and k. heyde , phys .
c * 69 * , 044306 ( 2004 ) .
m. anguiano , private communication .
b. h. wildenthal , prog .
* 11 * , 5 ( 1984 ) .
senkov , m. horoi , and v. zelevinsky , phys .
b * 702 * , 413 ( 2011 ) .
m. horoi and v. zelevinsky .
c * 81 * , 034306 ( 2010 ) . | a multiconfiguration microscopic method has been applied with the gogny effective interaction to the calculation of low - lying positive - parity states in even - even @xmath0si isotopes .
the aim of the study is to compare the results of this approach with those of a standard method of gcm type and to get insight into the predictive power of multiconfiguration methods employed with effective nucleon - nucleon force taylored to mean - field calculations .
it is found that the multiconfiguration approach leads to an excellent description of the low - lying spectroscopy of @xmath1si , @xmath2si and @xmath3si , but gives a systematic energy shift in @xmath4si .
a careful analysis of this phenomenon shows that this discrepancy originates from too large proton - neutron matrix elements supplied by the gogny interaction at the level of the approximate resolution of the multiparticle - multihole configuration mixing method done in the present study .
these proton - neutron matrix elements enter in the definition of both single - particle orbital energies and coupling matrix elements .
finally , a statistical analysis of highly excited configurations in @xmath2si is performed , revealing exponential convergence in agreement with previous work in the context of the shell model approach .
this latter result provides strong arguments towards an implicit treatment of highly excited configurations . |
a space @xmath4 is said to be a _ codimension one manifold factor _ provided that @xmath5 is a manifold .
it is a long standing unsolved problem as to whether all resolvable generalized manifolds are codimension one manifold factors @xcite .
this is the so - called product with a line problem and it is the essence of the famous generalized r. l. moore problem @xcite .
the product with a line problem speaks directly to one of the most fundamental questions in geometric topology , which is how to recognize manifolds @xcite . because manifolds have a rich structure which is useful to exploit in many areas of mathematics and its applications , it is important to recognize when one is dealing with a space that is a manifold .
one notable example is the relevance of the product with a line problem to the famous busemann conjecture in metric geometry @xcite .
one might wonder even if a decomposition of @xmath1 into convex sets could give rise to a decomposition space topologically distinct from @xmath1 .
this problem was investigated for several years beginning with bing in the 1950 s @xcite . in 1970 ,
armentrout @xcite produced the first example of a decomposition of @xmath6 into convex sets that yields a non - manifold .
then in 1975 , eaton @xcite demonstrated that a certain decomposition of @xmath6 into points and straight line segments , originally proposed by bing @xcite , is indeed topologically distinct from @xmath6 .
hence , this type of complexity is significant .
it should also be noted that there are no known examples of a non - manifold resulting from a decomposition of @xmath7 into convex sets . in this paper
we shall investigate how the type of complexity represented by decompositions of @xmath1 into convex sets can affect the classification of a decomposition space as a codimension one manifold factor .
we shall demonstrate that decompositions of @xmath1 , @xmath2 , into convex sets are always codimension one manifold factors .
in particular , we shall show that such spaces have a particularly strong general position property , the disjoint arc - disk property .
we briefly review some basic definitions and notations . recall that a map @xmath8 is said to be _ proper _ if whenever @xmath9 is a compact subset of @xmath10 , then @xmath11 is compact .
there are various equivalent definitions of upper semicontinuous decompositions @xcite , but the following will be the most useful for our purposes : a decomposition @xmath0 of @xmath12 into compact sets is said to be _ upper semicontinuous _ ( usc ) if and only if the associated decomposition map @xmath13 is a proper map . a compact subset @xmath9 of a space @xmath4
is said to be _ cell - like _ if for each neighborhood @xmath14 of @xmath9 in @xmath4 , @xmath9 can be contracted to a point inside @xmath14 @xcite .
a usc decomposition @xmath0 of @xmath12 is said to be _ cell - like _ if each element @xmath15 is cell - like .
a map @xmath16 is said to be _ cell - like _ if for each @xmath17 , @xmath18 is cell - like .
resolvable generalized n - manifold _ is an @xmath19-dimensional space @xmath4 that is the image of a cell - like map @xmath20 where @xmath12 is an @xmath19-manifold .
convex sets are contractible , and hence they are cell - like .
thus , a usc decomposition @xmath0 of @xmath1 into convex sets is a cell - like decomposition and the associated decomposition map @xmath21 is a cell - like map .
the fact that @xmath3 is finite - dimensional follows from a result of zemke ( see ( * ? ? ? * theorem 5.2 ) ) .
therefore , in this setting , @xmath3 is a resolvable generalized @xmath19-manifold . for resolvable generalized manifolds
, we have the following very useful approximate lifting theorem , which follows from ( * ? ? ?
* ; * ? ? ?
* theorem 17.1 and corollary 16.12b ) : [ alt ] suppose that @xmath0 is a cell - like decomposition of a manifold @xmath12 , with decomposition map @xmath22 , and that the quotient space @xmath23 is finite - dimensional .
then for any map @xmath24 of a finite - dimensional compact polyhedron @xmath25 , and any @xmath26 , there exists a map @xmath27 such that @xmath28 is an @xmath29-approximation of @xmath30 .
general position properties are very useful in detecting codimension one manifold factors @xcite . for our results
, we shall only need to employ the following : a space @xmath4 is said to have the _ disjoint arc - disk property _
( dadp ) provided that any two maps @xmath31 and @xmath32 can be approximated by maps with disjoint images , where @xmath33 denotes the unit interval and @xmath34 denotes a disk .
the following theorem was demonstrated in ( * ? ? ?
* ; * ? ? ?
* proposition 2.10 ) : [ dadp thm ] a resolvable generalized manifold having dadp is a codimension one manifold factor .
useful in discussions of the dadp is the local @xmath35-co - connectedness property .
a set @xmath36 is said to have the _ local @xmath35-co - connectedness property _
( @xmath35-lcc ) in @xmath4 if for every @xmath37 , each neighborhood @xmath14 of @xmath38 contains another neighborhood @xmath39 of @xmath38 so that any two points in @xmath40 are path connected in @xmath41 .
note that if @xmath25 is nowhere dense in @xmath4 , then @xmath42 .
the following theorem can be found in ( * ? ? ?
* ; * ? ? ?
* corollary 26.2a ) : [ 0-lcc ] each @xmath43-dimensional closed subset of a generalized @xmath19-manifold , where @xmath44 , is @xmath35-lcc . since a @xmath43-dimensional closed subset of a generalized @xmath19-manifold @xmath4 , where @xmath45 , is nowhere dense in @xmath4 , we have the following corollary : [ path ] if @xmath25 is a @xmath43-dimensional closed subset of a generalized @xmath19-manifold @xmath4 , where @xmath44 , then any path @xmath46 can be approximated by a path @xmath47 .
the main result of this paper is the following theorem : let @xmath0 be an upper semicontinuous decomposition of @xmath1 into convex sets , where @xmath2 .
then @xmath3 is a codimension one manifold factor .
this theorem will follow immediately as a corollary of theorem [ dadp thm ] and the following theorem : [ dadp-4 thm ] let @xmath0 be an upper semicontinuous decomposition of @xmath1 into convex sets , where @xmath2
. then @xmath3 has the dadp .
let @xmath48 and @xmath49 .
it follows from corollary [ path ] that it suffices to show that there is an @xmath29-approximation @xmath50 of @xmath30 such that @xmath51 is @xmath52-dimensional .
let @xmath53 be a piecewise linear map , that is an @xmath54-approximate lift of @xmath30 .
we shall show that @xmath55 is then the desired map .
let @xmath56 denote a triangulation of @xmath57 .
we claim that if @xmath58 is a @xmath52-simplex of @xmath56 , then @xmath59 is @xmath52-dimensional . to see this ,
let @xmath60 be the decomposition induced over @xmath61 , i.e. @xmath60 is the decomposition of @xmath1 having as the only nontrivial elements , the nontrivial elements of @xmath0 that meet @xmath58 .
let @xmath62 be the associated decomposition map .
note that @xmath63 is necessarily a proper map , being a decomposition induced over a closed set in the decomposition space of a usc decomposition .
let @xmath64 be the @xmath52-dimensional plane in @xmath65 that contains @xmath58 .
let @xmath66 denote the restriction of @xmath63 to @xmath64 .
then @xmath66 is also a proper map .
thus @xmath66 determines a usc decomposition of the plane into convex sets , elements that do not separate the plane .
it now follows from a classical result of moore @xcite , that @xmath66 is a near - homeomorphism onto its image .
thus @xmath67 is at most @xmath52-dimensional .
but @xmath67 is homeomorphic to @xmath68 , which in turn is homeomorphic to @xmath61 .
thus @xmath61 is at most @xmath52-dimensional subset of @xmath3 .
hence @xmath69 is a @xmath52-dimensional subset of the generalized @xmath19-manifold @xmath3 @xcite .
as we have seen , the complexity represented by decompositions into convex sets does not inhibit a decomposition space from being a codimension one manifold factor .
the fact that such spaces satisfy the dadp is a pleasant result .
it is well known that not all codimension one manifold factors satisfy the dadp , and hence the dadp is not a general position property that provides a characterization of codimension one manifold factors .
in fact , the dadp condition is a relatively weak tool for detecting codimension one manifold factors , compared to other general position properties such as : * the disjoint homotopies property @xcite ; * the plentiful @xmath52-manifolds property @xcite ; * the @xmath35-stitched disks property @xcite ; * the method of @xmath70-fractured maps @xcite ; and * the disjoint topographies ( or disjoint concordance ) property @xcite .
r. j. daverman and d. m. halverson , _ path concordances as detectors of codimension - one manifold factors _ , exotic homology manifolds , oberwolfach 2003 , geom .
* 9 * , geom . topol .
coventry , 2006 , pp . 715 .
d. m. halverson and d. repov , _ a survey on the generalized r. l. moore problem _ , proc .
`` computational and geometric topology '' ( bertinoro , italy , june 17 - 19 , 2010 ) , atti semin . mat .
modena reggio emilia * 58 * ( 2011 ) , 175191 .
d. repov , _ detection of higher dimensional topological manifolds among topological spaces _ , giornate di topologia e geometria delle variet , bologna 1990 , m. ferri , ed .
degli studi di bologna 1992 , pp . | we show that if @xmath0 is an upper semicontinuous decomposition of @xmath1 , @xmath2 , into convex sets , then the quotient space @xmath3 is a codimension one manifold factor .
in particular , we show that @xmath3 has the disjoint arc - disk property . |
let @xmath0 be an inner product over the real or complex number field @xmath1 .
the following inequality is known in the literature as _ schwarz s
inequality_:@xmath2where @xmath3 @xmath4 the equality occurs in ( [ 1.1 ] ) if and only if @xmath5 and @xmath6 are linearly dependent . in @xcite ,
the following _ reverse _ of schwarz s inequality has been obtained:@xmath7provided @xmath8 and @xmath9 are so that either@xmath10or , equivalently,@xmath11holds .
the constant @xmath12 is best possible in ( [ 1.2 ] ) in the sense that it can not be replaced by a smaller constant .
if @xmath13 satisfy either ( [ 1.3 ] ) or ( [ 1.4 ] ) , then the following reverse of schwarz s inequality also holds @xcite@xmath14 } { \left [ \func{re}\left ( \overline{a}% a\right ) \right ] ^{\frac{1}{2 } } } \label{1.5 } \\
& \leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1% } { 2}}}\left\vert \left\langle x , y\right\rangle \right\vert , \notag\end{aligned}\]]provided that , the complex numbers @xmath15 and @xmath16 satisfy the condition @xmath17 in both inequalities in ( [ 1.5 ] ) , the constant @xmath18 is best possible . an additive version of ( [ 1.5 ] ) may be stated as well ( see also ssd3)@xmath19 } { % \func{re}\left ( \overline{a}a\right ) } \left\vert \left\langle x , y\right\rangle \right\vert ^{2}. \label{1.6}\]]in this inequality , @xmath12 is the best possible constant .
it has been proven in @xcite , that@xmath20provided , either @xmath21or , equivalently,@xmath22where @xmath23 @xmath24 the constant @xmath12 in 1.7 is also best possible .
if we choose @xmath25 @xmath26 @xmath27 @xmath28 @xmath29 , then by ( [ 1.8 ] ) , ( [ 1.9 ] ) we have,@xmath30or , equivalently,@xmath31imply the following reverse of schwarz s inequality:@xmath32the constant @xmath12 in ( [ 1.12 ] ) is sharp .
note that this inequality is an improvement of ( [ 1.2 ] ) , but it might not be very convenient for applications .
now , let @xmath33 be a finite or infinite family of orthornormal vectors in the inner product space @xmath34 i.e. , we recall that @xmath35 in @xcite , we proved that if @xmath33 is as above , @xmath36 is a finite part of @xmath37 such that either@xmath38or , equivalently,@xmath39holds , where @xmath40 @xmath41 are real or complex numbers , then we have the following reverse of _ _ bessel s inequality:__@xmath42the constant @xmath12 in both inequalities is sharp .
this result improves an earlier result by n. ujevi obtained only for real spaces @xcite . in @xcite , by the use of a different technique ,
another reverse of bessel s inequality has been proven , namely:@xmath43provided that @xmath44 @xmath45 @xmath46 @xmath5 and @xmath47 are as above . here
the constant @xmath12 is sharp in both inequalities .
it has also been shown that the bounds provided by ( [ 1.15 ] ) and ( 1.16 ) for the bessel s difference @xmath48 can not be compared in general , meaning that there are examples for which one is smaller than the other @xcite .
finally , we recall another type of reverse for bessel inequality that has been obtained in @xcite:@xmath49provided @xmath40 @xmath50 satisfy ( [ 1.13 ] ) ( or , equivalently ( [ 1.14 ] ) ) and @xmath51 here the constant @xmath12 is also best possible .
an additive version of ( [ 1.17 ] ) is @xmath52 \right\ } } { \sum_{i\in f}\func{re}% \left ( \phi _ { i}\overline{\varphi _ { i}}\right ) } .
\notag\end{aligned}\]]the constant @xmath12 is best possible .
it is the main aim of the present paper to point out new reverse inequalities to schwarz s , triangle and bessel s inequalities .
some results related to grss inequality in inner product spaces are also pointed out .
natural applications for integrals are also provided .
the following result holds .
[ t2.1]let @xmath0 be an inner product space over the real or complex number field @xmath1 @xmath53 and @xmath54 @xmath55 are such that@xmath56 1 .
if @xmath57 then we have the inequality@xmath58 ^{2}\leq r^{2}\left\vert x\right\vert ^{2}. \label{2.2}\]]the constant @xmath59 in front of @xmath60 is best possible in the sense that it can not be replaced by a smaller one .
if @xmath61 then@xmath62the constant @xmath63 is best possible in both inequalities .
3 . if @xmath64 then@xmath65here the constant @xmath63 is also best possible . since @xmath66 then obviously @xmath67 which is equivalent to @xmath68 1 . if @xmath57 then we may divide ( [ 2.5 ] ) by @xmath69 getting @xmath70using the elementary inequality@xmath71we may state that@xmath72making use of ( [ 2.6 ] ) and ( [ 2.7 ] ) , we deduce@xmath73taking the square in ( [ 2.8 ] ) and re - arranging the terms , we deduce the third inequality in ( [ 2.2 ] ) .
the others are obvious . + to prove the sharpness of the constant , assume , under the hypothesis of the theorem , that , there exists a constant @xmath74 such that@xmath75 ^{2}\leq cr^{2}\left\vert x\right\vert ^{2 } , \label{2.9}\]]provided @xmath76 and @xmath77 + let @xmath78 @xmath79 @xmath80 with @xmath81 and @xmath82 put @xmath83 then obviously @xmath84 @xmath85 and @xmath86 , @xmath87 and thus @xmath88 ^{2}=\varepsilon .$ ] using ( [ 2.9 ] ) , we may write that@xmath89giving @xmath90letting @xmath91 we get from ( [ 2.10 ] ) that @xmath92 and the sharpness of the constant is proved .
the inequality ( [ 2.3 ] ) is obvious by ( [ 2.5 ] ) since @xmath93 the best constant follows in a similar way to the above .
3 . the inequality ( [ 2.3 ] )
is obvious .
the best constant may be proved in a similar way to the above .
we omit the details . the following reverse of schwarz s inequality holds .
[ t2.2]let @xmath0 be an inner product space over @xmath1 and @xmath94 @xmath95 such that either@xmath96or , equivalently,@xmath97holds . 1 .
if @xmath98 then we have the inequalities@xmath99 \right\ } ^{2}}{\func{re% } \left ( \gamma \overline{\gamma } \right ) } \label{2.13 } \\ & \leq \frac{1}{4}\cdot \frac{\left\vert \gamma + \gamma \right\vert ^{2}}{% \func{re}\left ( \gamma \overline{\gamma } \right ) } \left\vert \left\langle x , y\right\rangle \right\vert ^{2}. \notag\end{aligned}\]]the constant @xmath12 is best possible in both inequalities .
2 . if @xmath100 then @xmath101 \leq \left\vert \gamma + \gamma \right\vert \left\vert \left\langle x , y\right\rangle \right\vert . \label{2.14}\ ] ] 3 . if @xmath102 then @xmath103 \label{2.15 } \\ & \leq -\func{re}\left ( \gamma \overline{\gamma }
\right ) \left\vert y\right\vert ^{2}+\left\vert \gamma + \gamma \right\vert \left\vert \left\langle x , y\right\rangle \right\vert .
\notag\end{aligned}\ ] ] the proof of the equivalence between the inequalities ( [ 2.11 ] ) and ( 2.12 ) follows by the fact that in an inner product space @xmath104 for @xmath105 is equivalent with @xmath106 ( see for example @xcite ) .
consider , for @xmath107 @xmath108 and @xmath109 then@xmath110 1 .
if @xmath98 then the hypothesis of ( i ) in theorem [ t2.1 ] is satisfied , and by the second inequality in ( [ 2.2 ] ) we have@xmath111 \right\ } ^{2}\leq \frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}\left\vert x\right\vert ^{2}\left\vert y\right\vert ^{2}\]]from where we derive@xmath112 \right\ } ^{2},\]]giving the first inequality in ( [ 2.13 ] ) .
+ the second inequality is obvious . + to prove the sharpness of the constant @xmath113 assume that the first inequality in ( [ 2.13 ] ) holds with a constant @xmath114 i.e. , @xmath115 \right\ } ^{2}}{\func{re}\left ( \gamma \overline{\gamma } \right ) } , \label{2.16}\]]provided @xmath116 and either ( [ 2.11 ] ) or ( [ 2.12 ] ) holds .
+ assume that @xmath117 and let @xmath118 then ( [ 2.11 ] ) holds and by ( [ 2.16 ] ) we deduce@xmath119giving@xmath120let @xmath121 and choose in ( [ 2.17 ] ) , @xmath122 @xmath123 to get @xmath124 for any @xmath125 letting @xmath126 we deduce @xmath127 and the sharpness of the constant is proved .
+ \(ii ) and ( iii ) are obvious and we omit the details .
we observe that the second bound in ( [ 2.13 ] ) for @xmath128 is better than the second bound provided by ( [ 1.5 ] ) .
the following corollary provides a reverse inequality for the additive version of schwarz s inequality .
[ c2.3]with the assumptions of theorem [ t2.2 ] and if @xmath129 then we have the inequality:@xmath130the constant @xmath12 is best possible in ( [ 2.18 ] ) .
the proof is obvious from ( [ 2.13 ] ) on subtracting in both sides the same quantity @xmath131 the sharpness of the constant may be proven in a similar manner to the one incorporated in the proof of ( i ) , theorem [ t2.2 ] .
we omit the details .
it is obvious that the inequality ( [ 2.18 ] ) is better than ( [ 1.6 ] ) obtained in @xcite .
for some recent results in connection to schwarz s inequality , see @xcite , @xcite and @xcite .
the following reverse of the triangle inequality holds .
[ p2.4]let @xmath0 be an inner product space over the real or complex number field @xmath1 @xmath132 and @xmath133 @xmath55 are such that @xmath134then we have the inequality@xmath135 using the inequality ( [ 2.8 ] ) , we may write that@xmath136giving@xmath137since@xmath138then by ( [ 2.21 ] ) , we have@xmath139giving the desired inequality ( [ 2.20 ] ) .
the following proposition providing a simpler reverse for the triangle inequality also holds .
[ p2.5]let @xmath0 be an inner product space over @xmath1 and @xmath94 @xmath140 such that either@xmath141or , equivalently,@xmath142holds .
then we have the inequality@xmath143{mm}}\sqrt{\func{re}% \left\langle x , y\right\rangle } .
\label{2.24}\ ] ] choosing in ( [ 2.8 ] ) , @xmath144 @xmath145 we get@xmath146giving @xmath147following the same arguments as in the proof of proposition [ p2.4 ] , we deduce the desired inequality ( [ 2.24 ] ) . for some results related to triangle inequality in inner product spaces ,
see @xcite , @xcite , @xcite and @xcite .
we may state the following result .
[ t4.1]let @xmath0 be an inner product space over the real or complex number field @xmath1 @xmath148 and @xmath149 with @xmath24 if @xmath150 and @xmath151then we have the inequality@xmath152the inequality ( [ 4.2 ] ) is sharp in the sense that the constant @xmath153 in front of @xmath154 can not be replaced by a smaller constant .
apply schwarz s inequality in @xmath155 for the vectors @xmath156 @xmath157 to get ( see also @xcite)@xmath158using theorem [ t2.1 ] for @xmath159 we may state that@xmath160utilizing ( [ 4.3 ] ) and ( [ 4.4 ] ) , we deduce@xmath161which is clearly equivalent to the desired inequality ( [ 4.2 ] ) .
the sharpness of the constant follows by the fact that for @xmath162 @xmath163 we get from ( [ 4.2])@xmath164provided @xmath165 and @xmath166 the inequality ( [ 4.6 ] ) is sharp , as shown in theorem [ t2.1 ] , and the theorem is thus proved .
another companion of the grss inequality may be stated as well .
[ t4.2]let @xmath0 be an inner product space over @xmath1 and @xmath149 with @xmath167 suppose also that @xmath168 @xmath169 such that @xmath170 @xmath171 if either@xmath172or , equivalently,@xmath173holds , then we have the inequality@xmath174the constant @xmath12 is best possible . we know , by ( [ 4.3 ] )
, that@xmath175if we use corollary [ c2.3 ] , then we may state that@xmath176and@xmath177utilizing ( [ 4.10 ] ) ( [ 4.12 ] ) , we deduce@xmath178which is clearly equivalent to the desired inequality ( [ 4.9 ] ) .
the sharpness of the constant follows from corollary [ c2.3 ] , and we omit the details . with the assumptions of theorem [ t4.2 ] and
if @xmath179 ( that is actually the interesting case ) , one has the inequality@xmath180the constant @xmath12 is best possible .
the inequality ( [ 4.9 ] ) provides a better bound for the quantity @xmath181than ( 2.3 ) of @xcite .
for some recent results on grss type inequalities in inner product spaces , see @xcite , @xcite and @xcite .
let @xmath0 be a real or complex infinite dimensional hilbert space and @xmath182 an orthornormal family in @xmath183 , i.e. , we recall that @xmath184 if @xmath185 , @xmath186 and @xmath187 for @xmath188 .
it is well known that , if @xmath189 then the sum @xmath190 is convergent and the following inequality , called _ _ bessel s inequality__@xmath191holds . if @xmath192 where @xmath193 or @xmath194 , is the hilbert space of all complex or real sequences that are @xmath63-summable and @xmath195 then the sum @xmath196 is convergent in @xmath183 and if @xmath197 then @xmath198 we may state the following result .
[ t5.1]let @xmath0 be an infinite dimensional hilbert space over the real or complex number field @xmath1 , @xmath199 an orthornormal family in @xmath200 @xmath201 and @xmath55 with the property that@xmath202if @xmath203 is such that@xmath204then we have the inequality@xmath205 \right ) ^{2}}{\sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}-r^{2 } } \label{5.4 } \\ & \leq \frac{\left\vert \sum_{i=1}^{\infty } \overline{\lambda _ { i}}% \left\langle x , e_{i}\right\rangle \right\vert ^{2}}{\sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}-r^{2 } } \notag \\ & \leq \frac{\sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}}{% \sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}-r^{2}}% \sum_{i=1}^{\infty } \left\vert \left\langle x , e_{i}\right\rangle \right\vert ^{2 } ; \notag\end{aligned}\]]and@xmath206 applying the third inequality in ( [ 2.2 ] ) for @xmath207 we have@xmath208 ^{2}\leq r^{2}\left\vert x\right\vert ^{2 } \label{5.6}\]]and since@xmath209 , \end{aligned}\]]then by ( [ 5.6 ] ) we deduce@xmath210 ^{2}\leq r^{2}\left\vert x\right\vert ^{2},\]]giving the first inequality in ( [ 5.4 ] ) .
the second inequality is obvious by the modulus property .
the last inequality follows by the cauchy - bunyakovsky - schwarz inequality@xmath211the inequality ( [ 5.5 ] ) follows by the last inequality in ( [ 5.4 ] ) on subtracting in both sides the quantity @xmath212 the following result provides a generalization for the reverse of bessel s inequality obtained in @xcite .
[ t5.2]let @xmath0 and @xmath199 be as in theorem [ t5.1 ] .
suppose that @xmath213 @xmath214 are sequences of real or complex numbers such that@xmath215if @xmath203 is such that either@xmath216or , equivalently,@xmath217holds , then we have the inequalities@xmath218 \right ) ^{2}}{% \sum_{i=1}^{\infty } \func{re}\left ( \gamma _ { i}\overline{\gamma _ { i}}\right ) } \label{5.10 } \\ & \leq \frac{1}{4}\cdot \frac{\left\vert \sum_{i=1}^{\infty } \left ( \overline{\gamma _ { i}}+\overline{\gamma _ { i}}\right ) \left\langle x , e_{i}\right\rangle \right\vert ^{2}}{\sum_{i=1}^{\infty } \func{re}\left ( \gamma _ { i}\overline{\gamma _ { i}}\right ) } \notag \\ & \leq \frac{1}{4}\cdot \frac{\sum_{i=1}^{\infty } \left\vert \gamma _ { i}+\gamma _ { i}\right\vert ^{2}}{\sum_{i=1}^{\infty } \func{re}\left ( \gamma _ { i}\overline{\gamma _ { i}}\right ) } \sum_{i=1}^{\infty } \left\vert \left\langle x , e_{i}\right\rangle \right\vert ^{2}.
\notag\end{aligned}\]]the constant @xmath12 is best possible in all inequalities in ( 5.10 ) .
we also have the inequalities:@xmath219here the constant @xmath12 is also best possible . since @xmath220 , @xmath221
then also @xmath222 showing that the series@xmath223are convergent . also , the series @xmath224 are convergent in the hilbert space @xmath225 the equivalence of the conditions ( [ 5.8 ] ) and ( [ 5.9 ] ) follows by the fact that in an inner product space we have , for @xmath226 @xmath104 is equivalent to @xmath227 and we omit the details .
now , we observe that the inequalities ( [ 5.10 ] ) and ( [ 5.11 ] ) follow from theorem [ t5.1 ] on choosing @xmath228 @xmath188 and @xmath229 the fact that @xmath12 is the best constant in both ( [ 5.10 ] ) and ( [ 5.11 ] ) follows from theorem [ t2.2 ] and corollary [ c2.3 ] , and we omit the details . note that ( [ 5.10 ] ) improves ( [ 1.17 ] ) and ( [ 5.11 ] ) improves ( 1.18 ) , that have been obtained in @xcite . for some recent results related to bessel inequality , see @xcite , ssdjs , @xcite and @xcite .
the following result related to grss inequality in inner product spaces , holds .
[ t6.1]let @xmath0 be an infinite dimensional hilbert space over the real or complex number field @xmath1 , and @xmath199 an orthornormal family in @xmath225 assume that @xmath230 and @xmath231 with the properties that@xmath232if @xmath8 are such that@xmath233then we have the inequalities@xmath234 applying schwarz s inequality for the vectors @xmath235 @xmath236 we have@xmath237since@xmath238and@xmath239then by ( [ 5.5 ] ) applied for @xmath5 and @xmath240 and from ( [ 6.4 ] ) , we deduce the first part of ( [ 6.3 ] ) .
the second part follows by bessel s inequality .
the following grss type inequality may be stated as well .
[ t6.2]let @xmath0 be an infinite dimensional hilbert space and @xmath182 an orthornormal family in @xmath225 suppose that @xmath241 @xmath242 @xmath243 @xmath244 are sequences of real and complex numbers such that@xmath245if @xmath8 are such that either@xmath246or , equivalently,@xmath247holds , then we have the inequality@xmath248 ^{\frac{1}{2}}\left [ \sum_{i=1}^{\infty } \func{re}\left ( \phi _ { i}\overline{\phi _ { i}}\right ) % \right ] ^{\frac{1}{2}}}\left\vert x\right\vert \left\vert y\right\vert \notag\end{aligned}\]]the constant @xmath12 is best possible in the first inequality
. follows by ( [ 5.11 ] ) and ( [ 6.4 ] ) .
the best constant follows from theorem [ t4.2 ] , and we omit the details .
we note that the inequality ( [ 6.8 ] ) is better than the inequality ( 3.3 ) in @xcite .
we omit the details .
let @xmath249 be a measurable space consisting of a set @xmath250 a @xmath251algebra of parts @xmath252 and a countably additive and positive measure @xmath253 on @xmath252 with values @xmath254 let @xmath255 be a @xmath256measurable function on @xmath257 with @xmath258 denote by @xmath259 the hilbert space of all real or complex valued functions defined on @xmath257 and @xmath260integrable on @xmath250 i.e.,@xmath261it is obvious that the following inner product@xmath262generates the norm @xmath263 of @xmath264 and all the above results may be stated for integrals .
it is important to observe that , if @xmath265 \geq 0% \text { \ for } \mu -\text{a.e . }
s\in \omega , \label{7.3}\]]then , obviously,@xmath266
\label{7.4 } \\ & = \int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) \geq 0 .
\notag\end{aligned}\]]the reverse is evidently not true in general .
[ p7.1]let @xmath269 and @xmath270 with the properties that@xmath271and @xmath272 then we have the inequalities@xmath273 ^{2 } \notag \\ & \leq r^{2}\int_{\omega } \rho \left ( s\right ) \left\vert g\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) .
\notag\end{aligned}\]]the constant @xmath153 in front of @xmath60 is best possible .
[ p7.2]let @xmath269 and @xmath274 such that @xmath275 and@xmath276 \geq 0\text { \ for } \mu -\text{a.e . }
s\in \omega .
\label{7.8}\]]then we have the inequalities@xmath277 \right\ } ^{2}}{\func{re}\left ( \gamma \overline{\gamma } \right ) } \notag \\ & \leq \frac{1}{4}\cdot \frac{\left\vert \gamma + \gamma \right\vert ^{2}}{% \func{re}\left ( \gamma \overline{\gamma } \right ) } \left\vert \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right\vert ^{2}. \notag\end{aligned}\]]the constant @xmath12 is best possible in both inequalities .
if the space is real and we assume , for @xmath279 that@xmath280then , by ( [ 7.9 ] ) and ( [ 7.10 ] ) , we deduce the inequalities@xmath281 ^{2}d\mu \left ( s\right ) \int_{\omega } \rho \left ( s\right ) \left [ g\left ( s\right ) % \right ] ^{2}d\mu \left ( s\right ) \label{7.12 } \\ \leq \frac{1}{4}\cdot \frac{\left ( m+m\right ) ^{2}}{mm}\left [ \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) g\left ( s\right ) d\mu \left ( s\right ) % \right ] ^{2}.\end{gathered}\]]and @xmath282 ^{2}d\mu \left ( s\right ) \int_{\omega } \rho \left ( s\right ) \left [ g\left ( s\right ) \right ] ^{2}d\mu \left ( s\right ) \label{7.13 } \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ -\left [ \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) g\left ( s\right ) d\mu \left ( s\right ) \right ] ^{2 } \notag \\ & \leq \frac{1}{4}\cdot \frac{\left ( m - m\right ) ^{2}}{mm}\left [ \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) g\left ( s\right ) d\mu \left ( s\right ) % \right ] ^{2}. \notag\end{aligned}\]]the inequality ( [ 7.12 ] ) is known in the literature as cassel s inequality .
[ p.7.3]let @xmath283 with @xmath284 and @xmath168 such that @xmath285 and@xmath286 & \geq & 0 , \\ \func{re}\left [ \left ( ah\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{g\left ( s\right ) } -\overline{b}\overline{h\left ( s\right ) } \right ) % \right ] & \geq & 0\text{,}\end{aligned}\]]for @xmath287a.e .
@xmath288 then we have the inequalities@xmath289the constant @xmath12 is best possible .
dragomir , a generalisation of the cassels and grueb - reinboldt inequalities in inner product spaces , preprint , _ _ mathematics ar__@xmath290_iv _ , math.ca/0307130 , ` [ on line http://front.math.ucdavis.edu/ ] ` s.s .
dragomir , on bessel and grss inequalities for orthornormal families in inner product spaces , _ rgmia res .
rep . coll .
_ * 6*(2003 ) , _ supplement _ , article 12 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] ` s.s .
dragomir , a counterpart of bessel s inequality in inner product spaces and some grss type related results , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 10 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] ` s.s .
dragomir , some new results related to bessel and grss inequalities for orthornormal families in inner product spaces , _ rgmia res . rep .
_ * 6*(2003 ) , _ supplement _ , article 13 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] ` s. s. dragomir and b. mond , on the superadditivity and monotonicity of schwarz s inequality in inner product spaces . _
oddel . mat .-
* 15 * ( 1994 ) , no .
2 , 522 ( 1996 ) . | reverses of schwarz , triangle and bessel inequalities in inner product spaces that improve some earlier results are pointed out .
they are applied to obtain new grss type inequalities in inner product spaces .
some natural applications for integral inequalities are also pointed out . |
figure 1 : : : portion of the synchrotron x - ray diffraction patterns for nd@xmath8ca@xmath17mno@xmath5 at 100k with zero bias , in the region of the @xmath27 ( a)@xmath28 @xmath29 ( 122 ) ( b ) and @xmath30 ( c ) superlattice peaks .
figure 2 : : : @xmath31 versus bias current at 100k for nd@xmath32ca@xmath17mno@xmath5 ( a ) and pr@xmath6ca@xmath7mno@xmath5 ( b ) .
the arrow denotes the estimated treshold current .
figure 3 : : : cell volume versus temperature for zero bias ( filled symbol , lower scale ) and cell volume versus bias current at 100k ( upper scale ) figure 4 : : : integrated intensity of the @xmath33erlattice peaks as a function of the bias current at 100k .
a ) nd@xmath8ca@xmath17mno@xmath5 ; b ) pr@xmath34ca@xmath7mno@xmath25 the arrows denote the occurence of the non - linear regime .
figure 5 : : portion of the x - ray diffraction patterns for nd@xmath8ca@xmath35mno@xmath5 at 100k with bias 0 ma , 1 ma and 5 ma in the region of the @xmath36 superlattice peak . | high - resolution x - ray scattering , in the presence of an applied current , has been used for studying the stability of the charge ordered phase of manganese oxides upon current biasing .
we find that the charge ordered structure is unchanged when a current is flowing in the sample .
such a result indicates that the non - linear conduction observed in charge ordered manganites can not be ascribed to a current - induced destabilisation of charge ordering .
rare earth manganites with general chemical formula @xmath0 ( where @xmath1 and @xmath2 being trivalent rare - earth and divalent ions , respectively ) show a wide variety of electronic and magnetic properties such as colossal magnetoresistance and charge ( co ) or orbital ordering ( oo).@xcite the physics of the detabilization of the co by different types of perturbation has recently been under very active investigation.@xcite for instance , it is now well established that the co is stabilized by lattice distortions ; thus , a perturbation of this distortion by changing the average - a - site cationic radius can weaken the co.@xcite besides , it has been reported that irradiation by x - rays@xcite or light and application of pressure@xcite can also induce a melting of the co , leading to a transition from a charge - localized state ( cl ) to a charge - delocalized state ( cd ) .
application of an electric field is also known to induce a cl - cd transformation ; this latter feature is always associated with a strong non - linearity of the tension - current characteristics ( v - i).@xcite up to now , the optical and irradiation induced transition are argued to be the result of classical percolation transport in a non - homogeneous medium ; however , it is not clear whether the underlying mechanism is the same for all perturbations , in particular for the electric field induced transition . in the latter case
, this feature is interpreted in terms of current induced destabilization of the co - cl state leading to the creation of low resistivity conducting path along the current flow.@xcite recent experimental results do not agree with such a current - induced co melting scenario.@xcite indeed , electrical measurements , performed on the non - charge ordered compounds pr@xmath3ca@xmath4mno@xmath5 , also exhibit a current induced cl - cd transition associated with a strong non linearity of the v - i characteristics .
in such a case , the destruction of the co can not be invoked , suggesting that the mechanism is more complex than the one initially proposed . in this paper , we present synchrotron x - ray diffraction data for the charge - ordered manganites pr@xmath6ca@xmath7mno@xmath5 and nd@xmath8ca@xmath9mno@xmath5 polycrystalline samples .
details for synthesis of the sample are reported elsewhere.@xcite measurements have been performed as a function of temperature _ and current _ on the diffractometer id31 at esrf ( grenoble , france ) using an incident wavelength of 0.413 ( e=30 kev ) .
the typical sizes of the samples allow a homogeneous sample cooling together with suitable beam transmission .
two contacts of in were soldered onto the section ( 1x2 mm@xmath10 ) of the sample , leaving an uncovered section l= 3 mm .
v - i data were taken with current biasing ( keithley 236 ) .
our x - rays diffraction patterns , analyzed with the rietveld method , lead to an orthorhombic cell with @xmath11 @xmath12 , @xmath13 in a @xmath14 space group at room temperature .
previous neutron - diffraction measurements@xcite showed that the co structure is characterized by a doubling of the a - axis and the appearance of additional peaks of the type ( h/2 k l ) .
a careful search was made for such superlattice peaks associated with charge ordering of @xmath15and @xmath16 in pr@xmath6ca@xmath7mno@xmath5 and nd@xmath8ca@xmath17mno@xmath5 below @xmath18 = 235k and 200k , respectively .
figure 1 shows narrow regions covering the strongest of these peaks for nd@xmath8ca@xmath17mno@xmath5 at a temperature below @xmath18 ( @xmath19 , solid line ) together with the results of similar scans for t=295k @xmath18 ( symbols ) . in the latter case ,
no superlattice peaks can be observed whereas , for t= 100 k the superlattice peaks are well defined .
high resolution x - ray scattering measurements in the presence of an applied direct current would permit to determine unambiguously whether the current induced cl - cd transition associated with the non - linear v - i characteristics can be linked to a co destruction .
transport properties were simultaneously measured in situ during the experiments . a strong non - linearity i. e a deviation from the ohm s law ,
is observed when the bias current attains a threshold value for pr@xmath6ca@xmath7mno@xmath5 and nd@xmath8ca@xmath17mno@xmath5 powder samples .
this non - linearity is even more obvious when @xmath20 @xmath21 @xmath22 curves are plotted ( see figure 2a and 2b ) .
as expected the resistance is independent of the bias current in the ohmic regime and is strongly decreased for a critical value of the current.@xcite one can observe that the current value where the non - linearity sets in and the width of the transition are strongly temperature dependent ( both increase as the temperature increases ) . because of the rounding of the variation of @xmath20 @xmath21 @xmath22 , it is very difficult to define a value for a critical current .
however , a rough estimate gives @xmath23 = 4 ma and 1 ma for pr@xmath24ca@xmath7mno@xmath5 and nd@xmath8ca@xmath17mno@xmath5 , respectively .
the temperature dependence of the lattice parameters in these compounds is rather complex .
it is well known that their temperature behavior is characterized by the presence of changes of slopes corresponding to the characteristic structural and magnetic ordering temperatures.@xcite for t@xmath18 , a comparison of the current _ and _ temperature dependence of the cell volume determined from the synchrotron x - ray diffraction data allows us to discard definitely any joule heating effect .
indeed , no evolution of the cell volume is observed as a function of the applied current ( imax = 10 ma , well above the value of the current for which the non - linearity sets in ) whereas , in the range of temperature above 100k , a steep increase of the cell volume is observed with zero biased current ( see figure 3 for nd@xmath8ca@xmath17mno@xmath5 ) .
such a result is found for both composition pr@xmath6ca@xmath7mno@xmath5 and nd@xmath8ca@xmath17mno@xmath25 figure 4a and 4b show for both compounds the integrated intensity of the superlattice peaks corresponding to the co as a function of the applied current . in figure 5 ,
a characteristic portion of the x - ray diffraction patterns is shown at 100k with i = 0 , 1 and 5 ma .
the shapes and the intensities of the superlattice peaks are not modified whatever the value of the applied current lying in the ohmic or non - ohmic region
. such a result does not give a direct interpretation of the process involved in the non - linear conduction but , at least , allows us to discard any co destabilization as an origin for this feature .
other interpretations , based on a preserved superstructure when a current is flowing , must now be proposed . in charge
ordered compounds , observation of non - linear conduction setting in along with a large broad band noise suggests that the charge ordered state gets depinned at the onset of the non - linear regime.@xcite besides , the possibility of a charge density wave ( cdw ) condensate in this charge ordered regime of the manganese oxides is now well established.@xcite this might be a track to explain the occurrence of the non - linear conduction in such systems .
recently , we have proposed an interpretation in terms of collective excitation of cdws.@xcite in such a framework , the non - linear conduction arises from the motion of the cdw domains which carries a net electrical current . e. o. wollan and w. c. koehler , phys . rev . *
100 * , 545 ( 1955 ) ; j. b. goodenough , _ ibid .
_ * 100 * , 564 ( 1955 ) . for a review , see _ colossal magnetoresistance , charge ordering and related properties of manganese oxides , _ edited by c. n. r.rao and b. raveau ( world scientific , singapore , 1998 ) and _ colossal magnetoresistive oxides _ , edited by y. tokura ( gordon and breach science , new york ) .
s. jin , t. h. tiefel , m. mccormack , r. a. fastnacht , r. ramesh , and l. h. chen , science * 264 * , 413 ( 1994 ) ; r. von helmolt , j. wecker , b. holzapfel , l. schultz , and k. samwer , phys .
rev .
lett .
* 71 * , 2331 ( 1993 ) .
m. r. lees , j. barratt , g. balakrishnan , d. mck .
paul , and m. yethiraj , phys .
rev .
b. * 52 * , r14 303 ( 1995 ) .
h. yoshizawa , h. kawano , y. tomioka , and y. tokura , phys .
rev .
b. * 52 * , r13 145 ( 1995 ) .
y. tomioka , a. asamitsu , y. moritomo , and y. tokura , j. phys .
. jpn . * 64 * , 3626 ( 1995 ) .
v. kiryukhin , d. casa , j. p. hill , b. keimer , a. vigliante , y. tomioka , and y. tokura , nature ( london ) * 386 * , 813 ( 1997 ) .
m. fiebig , k. miyano , y. tomioka and y. tokura , science * 280 * , 1925 ( 1998 ) .
k. ogawa , w. wei , k. miyano , y. tomioka and y. tokura , phys .
rev .
b * 57 * , r15033 ( 1998 ) .. y. moritomo , h. kuwahara , y. tomioka and y. tokura , phys . rev .
b * 55 * , 7549 ( 1997 ) .
a. guha , a. raychaudhuri , a. raju and c. n. r. rao , phys .
rev .
b * 62*,5320 ( 2000 ) . c. martin , a. maignan , m. hervieu , and b. raveau , phys .
rev .
b * 60 * , 12 191 ( 1999 ) .
a. asamitsu et @xmath26 , nature * 388 * , 50 ( 1997 ) .
a. guha , n. khare , a. raychaudhuri and c. n. r. rao , phys .
rev .
b * 62*,r11941 ( 2000 ) c. n. r. rao , a. raju , v. ponnambalam , s. parashar and n. kumar , phys . rev .
b * 61 * , 594 ( 2000 ) r. c. budhani , n. k. pandey , p. padhan , s. srivastava and r. p. s. m. lobo , phys .
rev .
b * 65 * , 014429 ( 2001 ) s. mercone , a. wahl , ch .
simon and c. martin , phys .
rev .
b * 65 * , 214428 ( 2002 ) .
a. maignan , c. martin , f. damay , and b. raveau , z. phys .
b * 104 * , 21 ( 1997 ) .
z. jirk , s. krupika , z. ima , m. dlouh , and s. vratislav , j. magn .
magn . mater .
* 53 * , 153 ( 1985 ) .
d. e. cox , p. g. radaelli , m. marezio and s-w.cheong , phys .
rev .
b * 57 * 3305 ( 1998 ) .
a. wahl , s. mercone , a. pautrat , m. pollet , ch .
simon , submitted to phys .
rev .
b a. guha , a . gosh , a. raychaudhuri , s .
parashar , a. raju and c. n. r. rao , app .
phys .
lett . * 75 * 3381 ( 1999 ) n. kida and m. tonouchi , phys .
rev .
b * 66 * , 024401 ( 2002 ) .
y. d chuang , a. d. gromko , d. s. dessau , t. kimura and y. tokura , science * 292 * , 1509 ( 2001 ) t. asaka , s. yamada , s. tsutsumi , c. tsuruta , k.kimoto , t. arima and y. matsui , phys .
rev .
lett .
* 88 * , 097201 ( 2002 ) g. grner , rev .
mod .
phys . * 60 * , 1129 ( 1988 ) and references therein . |
hot , x - ray emitting gas is an important component of the gaseous content of the universe .
it usually dominates the interstellar medium of elliptical galaxies ( @xcite ) .
it also accounts for most of the baryonic mass of rich clusters of galaxies and up to 30% of their total gravitating mass ( @xcite ) .
such observations led suto et al .
( 1996 ; hereafter smio ) to suggest that the local group of galaxies might have a substantial x - ray halo of its own .
this hypothesized hot intragroup medium ( igm ) could contribute much of the excess in the soft ( @xmath11 kev ) x - ray background .
recent measurements have shown that this excess is well - fit by a thermal model with solar metallicity and a temperature of 0.16 kev ( @xcite ) .
determining the contributors to the x - ray background is crucial to understanding it .
furthermore , smio point out that such a local group x - ray halo would produce a significant quadrupole anisotropy in the microwave background through the sunyaev - zeldovich ( sz ) effect ( @xcite ) .
this quadrupole signature , @xmath2 , could in principle account for the entirety of the quadrupole signal observed by cobe : @xmath3k ( e.g. , @xcite ) .
the cobe measurement provides a critical constraint on cosmological models .
the power spectrum of primordial density fluctuations is usually parameterized as @xmath4 .
the spectral index @xmath5 is a primary component of any cosmological model , with the scale invariant value @xmath6 being the preferred `` natural '' value of inflationary models .
the cobe observations provide both the normalization @xmath7 and powerful constraints on @xmath5 .
the cobe normalization on large angular scales is far too high relative to the power on smaller ( galaxy to cluster ) scales for the predictions of the `` standard '' @xmath6 cold dark matter model ( e.g. , @xcite ) .
the need for relatively more power on large scales as indicated by cobe is a primary motivation for considering contrived models with admixtures of hot as well as cold dark matter .
a significant local contribution to @xmath8 through the sz effect would alter the range of viable cosmological models significantly , perhaps even reviving standard cold dark matter .
it is therefore very important to place limits on the potential @xmath2 of the putative local group hot igm .
since we are embedded in the hot interstellar medium of the galaxy , it is exceedingly difficult to disentangle the possible contribution of a hot igm to the diffuse x - ray background from other , more local sources .
however , the local group can be placed in context of observations of other groups of galaxies .
while the suggestion of a significant local group igm was made as a parallel to the highly luminous intracluster media of rich clusters , groups provide a more enlightening comparison . with the advent of the @xmath0 pspc ,
many observations of galaxy groups of varying richness have been obtained , so such comparisons can be made .
while poor groups of galaxies tend to be somewhat more populated and more dense ( in projection ) than the local group , they provide a much better comparison for the x - ray properties of the local group than rich clusters .
the @xmath0 pspc has provided the necessary sensitivity and angular resolution to make a detailed study of the diffuse x - ray emission seen in some galaxy groups .
pspc observations reveal that , in general , x - ray
luminous groups contain more elliptical galaxies than spirals and the first - ranked galaxy in these groups is an elliptical ( pildis , bregman , & evrard 1995 , @xcite ) .
very few groups with a spiral majority ( the local group has no bright ellipticals ) have detected x - ray emission , perhaps due to their not being sufficiently compact to contain an igm or the temperature of the igm being too low for the pspc to detect .
lcccccc suto et al . & & 150 & @xmath9 & 1 & 2.3 & @xmath10 hcg 12 & e & 45 & @xmath11 & 0.72 & 7.8 & @xmath12 hcg 62 & e & 60 & @xmath13 & 1.1 & 5.8 & @xmath14 hcg 68 & e & 33 & @xmath15 & 0.98 & 10.6 & @xmath16 hcg 97 & e & 8 & @xmath17 & 0.97 & 43.8 & @xmath18 n2300 group & e / s & 27 & @xmath19 & 0.93 & 13.0 & @xmath20 hcg 2 & s & 33 & @xmath21 & 1 & 10.6 & @xmath22 hcg 10 & s & 33 & @xmath23 & 1 & 10.6 & @xmath24 hcg 44 & s & 33 & @xmath25 & 1 & 10.6 & @xmath26 hcg 79 & e & 33 & @xmath27 & 1 & 10.6 & @xmath28 hcg 93 & s & 33 & @xmath29 & 1 & 10.6 & @xmath30 since over two dozen galaxy groups have been detected with the pspc , their properties can be used to constrain models of the possible igm of the local group .
the relevant quantities are t , the temperature of the plasma , the column density , which one can parameterize as @xmath31 ( where @xmath32 is the central electron density and @xmath33 is the core radius ) , and the distance of the milky way from the center of the local group , @xmath34 .
smio assumed a local group igm with t=1 kev , @xmath32=@xmath35 @xmath36 , and @xmath33=150 kpc , and assumed @xmath34=350 kpc .
table 1 lists the values for these quantities for the groups analyzed by pildis et al .
( 1995 ) , as well as the values that smio assumed . for the purposes of this comparison
, we retain the value of @xmath34 used by smio , as if one were to replace the local group igm with that of the other groups .
three groups in the sample of pildis et al .
( 1995 ) are not included in table 1 because they are not spatially resolved in the pspc observations : hcg 4 , 92 , and 94 ( see @xcite for details ) .
the upper limits assume the typical core radius ( 33 kpc ) and temperature ( 1 kev ) observed in the x - ray luminous groups , and use a calculated central density determined from the 3@xmath37 upper limit on the number of counts within the group radius in their pspc observations ( @xcite ) .
mulchaey et al .
( 1996 ) do not list central densities , but find average temperatures of @xmath11 kev and core radii ranging from 9 kpc to over 300 kpc , with a median value of 43 kpc ( excluding the groups from @xcite ) .
note that the observed groups tend to have smaller values of @xmath33 and greater values of @xmath32 than smio assumed for the local group igm , while the values for t are similar .
figure 1 shows an expanded version of fig . 4 of smio , with the addition of the data presented in table 1 and smio s own model for the local group igm .
the theoretical curves follow from equation ( 9 ) of smio , @xmath38^{-1},\ ] ] where @xmath39 is the electron mass , @xmath40 the speed of light , @xmath41 the thomson scattering cross section , and @xmath42 is the boltzmann constant .
there is also a numerical fudge factor @xmath43 due to the spherical harmonic multipoles ( see equations 4 and 5 of smio ) which smio calculate numerically but do not explicitly state .
we have adopted @xmath44 to match their curve for @xmath45k .
one detected group , hcg 97 , does not appear on the graph due to its large value of @xmath34/@xmath33 .
note that the ngc 2300 group , which is dominated by neither spirals nor ellipticals , has the lowest value of @xmath31 t of all the detections .
even the brightest elliptical - dominated groups in the pildis et al .
( 1995 ) sample fall well short of producing the observed @xmath46k . also well below the curve ,
puzzlingly enough , are the igm parameters assumed by smio .
it is rather mysterious why they explicitly suggest that `` the lg x - ray halo can potentially have a significant effect on the quadrupole of the cmb anisotropies '' when their own optimistic estimate of the local group parameters give @xmath47 .
the maximum quadrupole anisotropy consistent with the spiral - dominated groups appears to be 0.1 @xmath48k , a factor of 60 less than @xmath8 .
so , unless the local group is extremely unusual , any hot igm that it may contain is simply not relevant to the cobe quadrupole anisotropy measurement .
while it is clear that many systematic effects come into play in determining the characteristics of an igm from an x - ray observation ( notably the assumed metallicity of the gas and the background level of the observation ; see @xcite for an enlightening discussion ) , these effects could change the core radius and central density fitted to these groups by no more than a factor of a few , and likely much less .
the temperature of galaxy groups are well - determined to be close to 1 kev on average . with the further observation that the local group is less populated and more diffuse than even the average _ undetected _ spiral - dominated group
, it appears extremely unlikely that a local group igm could produce a significant fraction of the cobe quadrupole anisotropy .
perhaps a more plausible source for the soft excess in the x - ray background is a combination of the probable x - ray halo of the milky way plus emission from the local superbubble .
previous studies of the 1/4 and 3/4 kev x - ray backgrounds point towards their origination in the galaxy rather than the local group ( @xcite ) .
recent observations of x - ray shadowing by high - latitude molecular clouds demonstrates that the 3/4 kev background is dominated by emission from the local superbubble , while the 1/4 kev background is likely to arise from a patchy hot halo surrounding the milky way ( @xcite ) .
a significant fraction of spiral galaxies considered to be similar to our own galaxy are also seen to have patchy soft x - ray haloes ( @xcite ) .
since the x - ray emission in spiral - dominated groups is due solely to the component galaxies , there is no reason to expect that a local group igm would be an important component of the local x - ray background .
in light of the possibly substantial effect a local group x - ray emitting halo could have on the local soft x - ray and microwave backgrounds , as suggested by smio , we have considered the x - ray properties of poor groups of galaxies .
sparse , spiral - dominated groups like the local group have yet to be detected in x - rays , and even elliptical - dominated groups have low luminosities relative to rich clusters . we therefore conclude that 1 . unless the local group is exceedingly unusual for groups of its type
, it does not have a significant 1 kev x - ray halo .
the contribution of any hot gas in the local group igm to the soft ( 1 kev ) x - ray background is negligible .
a more important contributor might be hot gas associated with the milky way itself .
the contribution of a local group x - ray halo to the cobe quadrupole anisotropy measurement through the sunyaev - zeldovich effect is completely negligible : @xmath49k @xmath50k . since spiral - dominated groups as yet have only upper limits rather than actual detections of their igm , this upper limit on @xmath2 is quite hard unless the local group igm is very exceptional ( e.g. , comparable to that of a rich cluster of galaxies ) .
thus , the quadrupole anisotropy measured by cobe must represent the true cosmic signature .
the level at which this places the normalization of the primordial power spectrum remains an uncomfortable fact which cosmological models must fit .
bennett , c.l .
1994 , , 436 , 423 bregman , j.n . , hogg , d.e . , & roberts , m.s .
1992 , , 387 , 484 bregman , j.n . & pildis , r.a .
1994 , , 420 , 570 burrows , d.n . &
mendenhall , j.a .
1991 , , 351 , 629 elbaz , d. , arnaud , m. , & bhringer , h. 1995 , , 293 , 337 fabbiano , g. 1989 , , 27 , 87 fisher , k.b . ,
davis , m. , strauss , m.a . , yahil , a. , & huchra , j.p .
1993 , , 402 , 42 gendreau , k.c . ,
1995 , , 47 , l5 grski , k.m .
1994 , , 430 , l89 mccammon , d. & sanders , w.t .
1990 , , 28 , 657 mulchaey , j.s . ,
davis , d.s . , mushotzky , r.f . , &
burstein d. 1996 , , 456 , 80 pildis , r.a . ,
bregman , j.n . , &
evrard , a.e .
1995 , , 443 , 514 snowden , s.l . ,
mebold , u. , hirth , w. , herbstmeier , u. , & schmitt , j.h.m.m .
1991 , science , 252 , 1529 snowden , s.l . , mccammon , d. , & verter , f. 1993 , , 409 , l21 suto , y. , makishima , k. , ishisaki , y. , & ogasaka , y. 1996 , , 461 , l33 [ smio ] vogler , a. , pietsch , w. , & kahabka , p. 1996
, , 305 , 74 white , s.d.m . , navarro , j.f . ,
evrard , a.e . , & frenk , c.s .
1993 , , 366 , 429 zeldovich , ya.b . & sunyaev , r.a .
1969 , , 4 , 301 | suto et al . ( 1996 , , 461 , l33 ) have suggested that an x - ray halo in the local group might explain both the observed low - energy excess in the x - ray background and the quadrupole anisotropy in the cosmic microwave background .
recent observations of poor groups of galaxies by the @xmath0 pspc set reasonable limits on how extensive and dense such a halo could be .
the poor groups most similar to the local group do not have a detectable halo , and the upper limits of these observations suggest that any local group halo would be nearly two orders of magnitude too tenuous to produce the effects suto et al .
( 1996 ) discuss .
in particular , the sunyaev - zeldovich effect can not contribute significantly to the quadrupole anisotropy measured by cobe . |
the relationship between the dust and gas in a molecular cloud is crucial in understanding the cloud properties and the star formation activities therein .
dust provides the most reliable and useful tracer of the total gas column density ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , while molecules like and are more useful as tracers of the local physical and chemical conditions ( e.g. , gas temperatures and velocities , co column densities , etc . ) a comparison between them allows us to probe the variation of molecular gas abundances , i.e. , [ ] and [ ] ( [ molecule ] @xmath12 n(molecule)/n(h ) where n is column density ) , caused by such factors as photodissociation and chemical depletion / desorption in cold / warm environments .
this helps to reveal the physical and chemical properties of the molecular cloud , which enables us to better understand the initial conditions of star formation and its potential dependence on such factors as cloud chemistry and structure .
such variations have been noted in previous observations of a variety of gmcs , including perseus @xcite , taurus @xcite , and orion @xcite , made at relatively high spatial resolutions ( 0.2 - 0.4 pc ) .
here we report the results of a combined co - dust extinction study of the california molecular cloud ( cmc , * ? ? ?
* ) at even higher spatial resolution ( @xmath0 0.04 pc ) in order to investigate in more detail the relation between co gas and dust in this interesting nearby gmc .
the cmc is a useful laboratory for such studies . in particular
, it contains a massive filamentary structure in its southeast region that consists of both cold molecular gas with little star formation activity and hot molecular gas associated with an active star forming region containing the massive star lkh@xmath13 101 and its accompanying embedded cluster .
this provides a physical environment with large temperature and density variations that are interesting to explore .
indeed , as reported in @xcite , the cmc overall shows an order of magnitude lower star formation rate ( sfr ) than the orion molecular cloud , even though these equally distant gmcs have similar mass , size , and morphology .
recently , also found the clumps in the cmc to be characterized by similar kinematic states as those in orion .
@xcite proposed that the difference in sfrs between the two clouds was a result of a difference in cloud structure , specifically the amount of dense gas in each cloud .
moreover , this close connection between dense gas mass and the sfr appears to be a general physical property of galactic gmcs @xcite .
it would be of interest to compare other properties of the cmc with similar ones in the orion cloud to look for additional factors , such as cloud chemistry for example , that could contribute to the difference in the sfr between the two clouds .
a combined co - extinction study can also be used to investigate the co x - factor ( @xmath12 n(h@xmath14)/w@xmath15 , where w is integrated intensity ) which is widely used to derive molecular cloud masses from the co(1 - 0 ) line flux , especially in external galaxies ( see e.g. * ? ? ?
* ; * ? ? ?
despite being very optically thick , this co rotational transition is still thought to be a good tracer for the total molecular gas mass , due to the effect of photon transport in gas with large velocity gradients .
however , variations of physical conditions can have an effect on the x - factor and , in many instances , render its application rather uncertain .
apparently , there is no universally valid x - factor ( see recent review from * ? ? ?
consequently , investigating the dependence of the x - factor on differing physical conditions is clearly of great interest in establishing practical guidelines for its application .
combined co - extinction studies of nearby clouds can therefore provide potentially important insights into this issue . in this paper
, we present observational results of molecular lines ( 2 - 1 ) , ( 2 - 1 ) , and ( 2 - 1 ) .
the total gas mass ( or n@xmath16 ) is traced by extinction derived from nir observations , independent from the molecular lines . the area covered in our observation is mainly dense gas with @xmath10 3 mag . with the nicest technique , we are able to trace extinction up to @xmath0 33 mag , which is much deeper than achieved in a similar study of orion by @xcite who used the nicer technique to trace extinction depths up to @xmath0 15 mag
. we also observed the optically thinner ( 2 - 1 ) line , which provides insights into co chemistry ( e.g. , * ? ? ?
* ; * ? ? ?
* ) that can not be readily derived using only and observations . in the following ,
we describe our observations and data reduction in section [ sec : obs ] .
we present the results and analysis in section [ sec : res ] .
discussions and conclusions will be presented in section [ sec : discus ] and section [ sec : con ] , respectively .
llccccl cn-01 & 2009 - 12 - 24 & 04:30:58.81 & + 34:51:57.5 & @xmath17 & 1.49 & 19.00 + cn-01 & 2009 - 12 - 24 & 04:30:58.81 & + 34:51:57.5 & @xmath18 & 1.52 & 19.75 + cn-01 & 2009 - 12 - 24 & 04:30:58.81 & + 34:51:57.5 & @xmath19 & 1.39 & 19.25 + cn-02 & 2009 - 12 - 08 & 04:30:38.55 & + 35:04:20.4 & @xmath17 & 1.36 & 19.50 + cn-02 & 2009 - 10 - 09 & 04:30:38.55 & + 35:04:20.4 & @xmath18 & 1.59 & 20.75 + cn-02 & 2009 - 10 - 09 & 04:30:38.55 & + 35:04:20.4 & @xmath19 & 1.47 & 19.75 + cn-03 & 2012 - 12 - 23 & 04:30:07.84 & + 35:16:06.6 & @xmath17 & 1.11 & 22.25 + cn-03 & 2012 - 12 - 23 & 04:30:07.84 & + 35:16:06.6 & @xmath18 & 1.06 & 20.25 + cn-03 & 2012 - 12 - 23 & 04:30:07.84 & + 35:16:06.6 & @xmath19 & 1.12 & 19.25 + cn-04 & 2009 - 10 - 07 & 04:30:44.04 & + 35:27:49.5 & @xmath17 & 1.05 & 21.25 + cn-04 & 2009 - 10 - 07 & 04:30:44.04 & + 35:27:49.5 & @xmath18 & 0.96 & 19.50 + cn-04 & 2009 - 10 - 09 & 04:30:44.04 & + 35:27:49.5 & @xmath19 & 1.05 & 21.50 + cn-05 & 2009 - 10 - 07 & 04:30:44.04 & + 35:41:24.1 & @xmath17 & 0.91 & 21.25 + cn-05 & 2009 - 10 - 07 & 04:30:44.04 & + 35:41:24.1 & @xmath18 & 0.93 & 22.00 + cn-05 & 2009 - 10 - 06 & 04:30:44.04 & + 35:41:24.1 & @xmath19 & 1.17 & 19.75 + cn-06 & 2009 - 10 - 06 & 04:30:44.73 & + 35:54:54.8 & @xmath17 & 1.29 & 22.00 + cn-06 & 2009 - 10 - 06 & 04:30:44.73 & + 35:54:54.8 & @xmath18 & 1.26 & 21.00 + cn-06 & 2009 - 10 - 06 & 04:30:44.73 & + 35:54:54.8 & @xmath19 & 1.30 & 20.00 + cn-07 & 2010 - 01 - 02 & 04:29:08.58 & + 36:29:18.7 & @xmath17 & 1.07 & 20.75 + cn-07 & 2010 - 01 - 02 & 04:29:08.58 & + 36:29:18.7 & @xmath18 & 1.13 & 19.00 + cn-07 & 2010 - 01 - 02 & 04:29:08.58 & + 36:29:18.7 & @xmath19 & 1.18 & 18.75 + cn-08 & 2010 - 01 - 03 & 04:28:16.37 & + 36:29:58.4 & @xmath17 & 1.37 & 20.25 + cn-08 & 2010 - 01 - 03 & 04:28:16.37 & + 36:29:58.4 & @xmath18 & 1.35 & 20.00 + cn-08 & 2010 - 01 - 03 & 04:28:16.37 & + 36:29:58.4 & @xmath19 & 1.02 & 19.25 near - ir observations for this study were obtained at the 3.5 m telescope of the centro astronmico hispano alemn observatory ( hereafter caha ) at calar alto in almera , spain .
our observations were made with the omega 2000 camera , which has a field of view of 15@xmath20 . in this paper
we report observations for eight fields that cover the l1482 region .
these fields are part of a more comprehensive survey of the cmc with a total of 24 fields .
fields cn01 to cn08 ( table [ tab : obs ] ) were observed in the three main near - ir broadbands , @xmath17 , @xmath18 and @xmath21 , between october 2009 and january 2010 , with acceptable weather and seeing , except for one field , cn03 , which was repeated on december , 2012 .
all observations consisted of 20 exposures of 60 seconds , coadded as 1200 second exposure mosaics .
the fields observed with the caha 3.5 m telescope were selected to cover the main filament or spine " of l1482 , including the region of the lkh@xmath13 101 cluster .
the pixel scale of omega 2000 at the 3.5 m telescope is 0.45 @xmath22 , with excellent uniformity and negligible geometric distortion across the fov .
a list of all fields observed and considered for final analysis can be consulted in table [ tab : obs ] .
the table lists the field identification , the center of field positions , observation date , filter , an estimate of the seeing based on the average fwhm of the stars detected in each field , and the peak values for the brightness distributions , which are a good measurement of the sensitivity limits achieved .
below , we describe the observations and the data reduction process , including the construction of the photometric catalogs used to make our dust extinction map .
the near - ir image data from omega 2000 was reduced with pipelines that made use of standard iraf procedures complemented with stand alone routines .
these pipelines are based on the flamingos near - ir reduction , photometry and astrometry pipelines @xcite .
the data process is essentially identical to the one described in @xcite and we refer the reader to that paper for details . both photometry and astrometry of omega 2000 data products were calibrated relative to two - micron all sky survey ( 2mass ) lists obtained from the all - sky point source catalog ( psc ) .
the final photometry catalogs were merged with a combination of topcat - stil @xcite and idl routines . in the case of adjacent fields ,
overlapping areas were treated with a routine designed to handle duplicate detections in such a way that we list preferentially a higher quality observation ( e.g. a smaller photometric uncertainty ) over a lower quality one .
the 2mass data were also used to replace observations for saturated stars in all our frames .
we used , in all cases , lists retrieved from the 2mass all - sky point source catalog ( psc ) at the infrared processing and analysis center ( ipac ) .
the final photometry catalogs , containing either j , h , and ks , or h and ks photometry were merged with a combination of topcat - stil @xcite and self - made idl routines .
a series of observations were carried out with the heinrich hertz sub - millimeter telescope ( smt ) on mount graham , arizona , from november 2011 to april 2012 .
the smt is located at an elevation of 3200 meters .
the receiver has prototype alma band 6 sideband separating mixers with two orthogonal polarizations .
typical single sideband system temperatures during the observing were around 200 k. the molecular lines ( 2 - 1 ) ( 230.538 ghz ) , ( 2 - 1 ) ( 220.399 ghz ) , and ( 2 - 1 ) ( 219.560 ghz ) were observed over a selection of 17 @xmath23 square regions ( hereafter tiles " ) along the south - eastern dense ridge of the cmc , roughly at ra=@xmath24 , dec=@xmath25 ( see figure [ fig : obs ] , also figure 1 of * ? ? ?
these tiles have @xmath10 3 mag , and cover the higher extinction regions of the cloud as well as the lkh@xmath13 101 cluster ( roughly centered in tile 12 ) . in each tile
, we performed the observations using the on - the - fly ( otf ) " mode .
the scanning rate was @xmath26 per second along the direction of right ascension ( ra ) , with each row being separated by @xmath26 in declination ( dec ) .
each sampling point has a 0.4-second integration , corresponding to 4@xmath27 spatial sampling in ra with a beam size around @xmath28 , therefore the source is fully sampled .
the total integration time for each tile is roughly 1 hour .
we utilized dual side band mode and had two distinct sets of observations : obs1 : ( 2 - 1 ) in lower side - band ( lsb ) and ( 2 - 1 ) in upper side - band ( usb ) ; obs2 : ( 2 - 1 ) in lsb and ( 2 - 1 ) in usb . in each set ,
two orthogonal linear polarizations were observed simultaneously .
the total bandwidth we cover with the spectrometer is 128 channels of 0.25 mhz each or 32 mhz total , for each line ( , , and ) .
the channel resolution is 0.25 mhz , set by the filter widths , and this corresponds to about 0.34 @xmath29 velocity resolution , depending on the line rest frequencies via the doppler formula .
we checked the pointing on the compact nearby co source crl618 ( ra=@xmath30 , dec=@xmath31 ) at the beginning of every tile .
five - second integrations were performed on each point of a 5-point cross pattern centered on that source .
the pointing error was typically @xmath1 5@xmath27 . to calibrate the intensity , we observed the strong co source w3oh ( ra=@xmath32 , dec=@xmath33 ) between every two tiles .
all data were first processed through the class software package .
we fitted linear baselines and subtracted them from the data .
the data image cubes were ported into miriad software format @xcite for further analysis and display .
otf data were interpolated onto regular grids with grid spacing of 10@xmath27 before being stored into the miriad data set .
next the data were calibrated using w3oh .
a scale factor is derived through the observation toward w3oh between every two tiles .
the factor is used to scale the antenna temperature @xmath34 to the main - beam brightness temperature ( see * ? ? ? * for details ) . for each molecular transition
, the two polarizations were averaged to increase the snr .
finally all 17 tiles were combined together to form a single map ( weighted by their rms noise at the small overlapped area ) . the original data cube , with @xmath0 0.34 km s@xmath35 velocity channels ( set by the 0.25 mhz filter bandwidth ) were re - sampled onto a 0.15 km s@xmath35 spacing , by 3rd - order polynomial interpolation .
this re - sampling puts all of the co isotopologues on exactly the same velocity grid , so we can make ratio maps , for example .
we convolved all maps with gaussian kernels ( fwhm = 18.1@xmath27 for map , fwhm = 13.6@xmath27 for and maps ) to have a final map resolution ( fwhm ) of 38@xmath27 with a grid spacing of 19@xmath27 in order to match the resolution of and facilitate direct comparison with our dust extinction maps .
we calculated the rms noise for all emission - free channels in each map to be about 0.1 k for all three lines .
we compared the integrated intensity maps of the two sets of maps by subtracting one from the other .
we found a small spatial difference ( typically @xmath1 10@xmath27 between each pair of corresponding tiles ) , and we shifted and re - gridded obs1 and maps to obs2 .
we averaged the two sets of maps weighted by their rms noise to lower the noise in the averaged map to 0.07 k ( rms ) .
figure [ fig : obs ] shows the deep dust extinction map of the southeast portion of the cmc derived from our observations . for this map calar alto data were supplemented with data from 2mass in some outer portions of the surveyed area where calar alto data were not available . also shown
are the boundaries of the 17 tiles of the multi - line co mapping survey .
the extinction map was constructed with the optimized version of the near infrared color excess revised ( nicer ) technique , known as nicest .
the nicer / nicest technique allows us to measure dust extinction from the infrared excess in the colors of background stars caused by dust absorption .
the excess is derived by assuming as intrinsic colors , the colors of stars detected in a nearby off - cloud control field with negligible extinction ( see table 1 ) . for the measurements of extinction , we avoid the use of sources with intrinsic color excess , such as candidate young stellar objects and dusty galaxies .
nicest is essentially identical to nicer , except that the estimator is modified to account for small - scale inhomogeneities , due to a bias introduced by the decrease in the number of observable stars as extinction increases toward the denser areas of a molecular cloud .
the bias is removed by modifying the estimator with an additional term that accounts for the expected decrement .
extinction measurements for individual sources are smoothed with gaussian filter with a fwhm of 38@xmath27 , and the maps are constructed with 19@xmath27 spatial sampling . at each ( line - of - sight )
position in the map , each star falling within the beam is given a weight calculated from a gaussian function , and the inverse of the variance squared , which is in turn also used for the bias - correction factor . a preliminary value of at the map position is calculated as the weighted median of all possible values within the beam .
then , we estimate the mean absolute deviation ( m.a.d . ) of these values to remove large deviates using m.a.d .- clipping .
the remaining values are used by the nicest estimator to calculate the final at the corresponding position . in the case of pixel positions where only 2mass sources are available ,
the number of sources per beam is significantly smaller , and there are cases where there are not enough stars to estimate ; in those cases , our code increases the size of the beam by a factor of 2 over the nominal value .
the nominal 38@xmath27 fwhm of our maps represents an increase of resolution by a factor of 2.1 compared to the earlier study of @xcite .
this resolution is comparable to the beam size of the co maps .
the resulting noise of extinction map is 0.05 mag for @xmath2 10 mag and 0.1 mag for @xmath36 10 mag . in order to provide an overview of the observations and the general quality of the data we first show in figure [ fig : spec ] the averaged co spectra derived for each of the @xmath23 tiles of our mapping grid . on this coarse scale ( 1.3 @xmath5 1.3 pc )
the profiles exhibit evidence for both self - absorption and multi - component structure .
the presence of self - absorption indicates that the emission is quite optically thick .
the profiles of the rarer isotopes appear to be less complex , mostly characterized by simple gaussian - like shapes .
maps of the emission from ( 2 - 1 ) , ( 2 - 1 ) , ( 2 - 1 ) integrated over velocity ranges , @xmath3712 to @xmath3810 km s@xmath35 , @xmath3710 to @xmath386 km s@xmath35 , @xmath374 to @xmath383 km s@xmath35 , respectively , are shown in figure [ fig : mom0 ] as grey scale maps with contours overlaid for clarity .
the maps display some common morphological features .
the strongest emission occurs in the mid - southern area ( tiles 11 - 14 , @xmath39 ) and is in close proximity to the lkh@xmath13 101 cluster ( which lies within tiles 10 - 12 , @xmath40 ) . on close inspection of the maps ,
a sharp drop off in co emission is observed along a warm filamentary structure southwest of the cluster .
this sharp drop in co emission together with the enhanced co emission within the warm filament likely indicates a physical interaction between the cluster stars and the filamentary ridge structure .
a similar structure with an accompanying drop off is also present in the extinction map of figure 1 . here
we can see that the warm filamentary structure is a small section of the more extended high extinction ridge that forms the backbone of the cloud in this region and which is the subject of this survey . in these more extended regions away from the cluster , the co emission is comparatively quiescent .
( a)(b)(c ) + the rms noise for the integrated intensity maps @xmath41 ( in k km s@xmath35 ) was first estimated using @xmath42 where @xmath43 ( in k ) is the rms noise per channel derived for emission - free velocities , @xmath44 is the velocity width over which the integrated intensity is calculated . n@xmath45 here is the number of real independent channels or filters ( of bandwidth 0.25 mhz , @xmath0 0.34 km s@xmath35 ) within @xmath44 .
the estimated @xmath41 for ( 2 - 1 ) , ( 2 - 1 ) , ( 2 - 1 ) integrated intensity maps are 0.16 , 0.22 , 0.16 @xmath46 , respectively .
we also calculated the rms for an integrated intensity map computed over the same number of emission free channels in the data cube .
the @xmath41 derived this way are 0.11 , 0.24 , 0.24 @xmath46 , respectively .
to be conservative we adopt the larger value for @xmath41 , i.e. , 0.16 , 0.24 , 0.24 @xmath46 , for ( 2 - 1 ) , ( 2 - 1 ) , ( 2 - 1 ) , respectively ( hereafter @xmath47 , @xmath48 , @xmath49 ) .
the and lines are very strong and in our subsequent analysis we only consider integrated intensities detected above 5@xmath41 . with , we only consider detections above 2@xmath41 .
we also note some striping apparent in tiles 5 , 16 and 17 of the map .
these are likely artifacts due to system temperature variations in some of the otf scans in those regions of the cloud .
prior to the subsequent analysis of these data reported below this map was cleaned " so that pixels containing the stripes were removed from the database in this paper we adopt the standard lte analysis to determine excitation temperatures and column densities . the main beam brightness temperature , , is given by : @xmath50 ( 1 - e^{-\tau_\nu})\ ] ] where @xmath51 is the rayleigh - jeans equivalent temperature of a blackbody of physical temperature t , the excitation temperature of the line under consideration
, the temperature of the background ( in this case the cmb ) , and @xmath52 the optical depth of the transition .
the excitation temperature of a given line can be derived from the peak temperature of the line , provided the line is optically thick .
typically emission is optically thick in gmcs . in the cmc a large opacity for the ( 2 - 1 ) line
is confirmed by the /integrated intensity ratio .
the probability density distribution for this ratio for all pixels is found to peak near a value of 3.5 , considerably smaller than the optically thin limit of @xmath53 40 - 70 ( see , e.g. , * ? ? ?
less than 10% of the pixels are characterized by ratios exceeding 10 .
this is consistent with the fact that the surveyed area was dominated by relatively high extinction ( dust column density ) . for @xmath54 , for the line is calculated from the following equation ( by solving eq.[eq : radt ] ) : @xmath55 where t@xmath56 is the peak brightness temperature in the line . at this frequency ,
@xmath57 = 0.19 k and is generally much less than the peak line temperature ( t@xmath56 ) .
the typical error in propagated from the 3 sigma uncertainty in t@xmath56 is @xmath0 0.22 k. the maximum percentage error in is below 5% .
note that in some regions , e.g. , tiles 12 and 14 , the line shows self - absorption , this would result in underestimation of t@xmath56 and thus . for t@xmath56 between 6 k and 50 k , a change of this quantity by 1 k produces a corresponding change of by 1 k. figure [ fig : texmap ] shows a map of the spatial distribution of the excitation temperatures in the surveyed region . over the vast majority of the cloud , is found to be in a narrow range around 10 k as expected for gas heated by cosmic rays and cooled by molecular lines .
higher temperatures are found in the southern part of the region .
the highest excitation temperatures approach 40 k and are found within the bright filamentary ridge southwest of the lkh@xmath13 101 cluster ( see figure ) suggesting that the cluster is providing additional heating of the gas in that area .
the optical depths of ( 2 - 1 ) ( @xmath58 ) and ( 2 - 1 ) ( @xmath59 ) were calculated assuming lte from : @xmath60.\ ] ] here for both transitions was taken to be that derived for ( 2 - 1 ) .
the probability density distribution of opacities for peaks around 0.5 with the majority ( 87% ) of pixels appearing to be relatively optically thin ( i.e. , @xmath61 @xmath2 1 ) . for the distribution peaks around 0.2 ,
significantly less than that for ( as expected ) , and with @xmath61 @xmath2 1 for essentially all pixels in the map .
we followed @xcite in calculating both the and column density using eq.[eq : ncol ] for j=2 - 1 transition .
the rotational constant b@xmath62 and the einstein coefficient a@xmath63 are taken from an online database .
the partition function q is estimated by eq.[eq : part ] , which is accurate to 10 percent for t @xmath3 5 k ( note that in our data , @xmath3 5 k everywhere , as shown in figure [ fig : texmap ] ) .
@xmath64{\rm d}v}\int t_{mb}{\rm d}v~(cm^{-2})\\ \end{split}\ ] ] @xmath65 we note that sub - thermal excitation for the optically thin ( 2 - 1 ) and ( 2 - 1 ) lines might be expected to be a concern in low extinction regions of the cloud where number densities may be lower than the critical density for thermal ( lte ) excitation ( @xmath53 3 @xmath5 10@xmath66 @xmath67 ) .
this could lead to underestimates of the co column densities in the outer regions .
however , since the studied area here is dominated by relatively high extinction ( @xmath68 3 - 4 magnitudes ) we expect that the effects of any sub - thermal excitation are minimized .
we can roughly estimate the magnitude of this effect using the radex , non - lte , radiative transfer code .
we find that with @xmath69 @xmath67 , t@xmath70 @xmath4 10 k , and w ( ) @xmath53 2.5 k km s@xmath35 , n@xmath71 @xmath53 2.0 n@xmath72 .
since our measurements of co column densities and dust extinction are set to the same angular resolution and spatial grid , we are in a position to directly compare the two data sets .
we note here that the angular scale of a pixel is 19@xmath27 , corresponding to a spatial scale of @xmath0 0.04 pc ( @xmath0 8000 au ) at a distance of 450 pc @xcite . assuming a constant gas - to - dust mass ratio , ( n(h)/=1.88@xmath510@xmath73 @xmath74 mag@xmath35 ) @xcite , the co abundance can be expressed as : [ co ] = n(co)/n(h ) = n(co)/[(@xmath75 ) ] . for optically thin emission lines , such as and
, we expect the gas column density to increase linearly with , provided the cloud is characterized by a constant co abundance ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in figure
[ fig : ncol ] we plot the molecular column density vs. extinction at each pixel in the mapped region for ( a ) and ( b ) , respectively .
contrary to our simple expectations , each plot exhibits considerable scatter with only weak correlations between gas and dust column densities .
if the gas in the cmc were characterized by a single , constant abundance , the relation between its column density and extinction would consist of a single straight line in each plot . at
fixed there would be only small scatter in n(co ) , which is clearly not the case .
for instance , at @xmath0 20 mag the ratio between the highest and lowest values of n ( ) and n ( ) is @xmath10 10 for each isotope . clearly
, the two plots suggest strong spatial variations in abundances of these two species across the cloud . on closer inspection of figure [ fig : ncol ]
one can almost make out two significant branches ( correlations ) of pixels in both plots .
first , a large group of pixels in each plot ( below the blue dashed line , i.e. , n ( ) @xmath1 1.0@xmath510@xmath76 @xmath74 , n ( ) @xmath1 1.0@xmath510@xmath77 @xmath74 ) display an almost constant , low column density across a wide range of
. this flat relation between the two quantities indicates a trend of _ decreasing _ molecular abundance ( [ ] and [ ] ) from low to high pixels .
second , another group ( above the blue dashed line ) of pixels is characterized by gas column densities that appear to rise more or less linearly with extinction to very high column density at @xmath0 20 mag ( 1.0@xmath510@xmath76 @xmath74 @xmath1 n ( ) @xmath1 6.0@xmath510@xmath76 @xmath74 , 1.0@xmath510@xmath77 @xmath74 @xmath1 n ( ) @xmath1 6.0@xmath510@xmath77 @xmath74 ) , indicating a more or less uniform molecular abundance with depth into the cloud for this group of pixels .
these branches likely mark two limiting extremes of spatially dependent variation in the chemical conditions within the cmc . spatial variations in abundances have been reported previously in a few other molecular clouds , although in these clouds the variations appear less pronounced than found here . for the perseus molecular cloud , @xcite found generally tighter correlations of column density with extinction than found here , although they explored the correlation at much lower extinctions ( i.e. , @xmath1 10 mag ) than in this study .
nonetheless , over this lower extinction range they did find measurable variations ( at the @xmath0 40% level ) in the abundances with position across the cloud .
for the orion a molecular cloud , @xcite measured abundances at both low and high extinction and found more significant positional variations in the abundance across the cloud than observed in perseus @xcite . at @xmath0 20 mag , n ( ) in orion a was found to vary by about a factor of 4 between various locations in the cloud .
further , @xcite were able to associate these positional variations in abundance with positional variations in the physical conditions within the cloud .
at the lowest extinctions ( @xmath1 3 mag ) , where there is not enough to self - shield against uv dissociation , they found extremely low abundances with significant ( factor of 8) spatial variations . at intermediate extinctions ( 3 @xmath1 @xmath1 10 mag ) , where self - shielding is considerably more effective , higher abundances were found with modest ( factor of 2 ) spatial variations . at the highest extinctions they observed the highest abundances but also the largest spatial variations in abundances across the cloud .
this latter regime included cold portions of the cloud where co depletion depressed the abundances as well as hot regions heated by young stars where co desorption produced enhanced co abundances .
the molecular abundance is regulated by the physical and chemical conditions in the gmc .
for instance , the gas - phase molecules can condense onto dust grains to form ice mantles , and the molecules on grain surfaces can be returned to the gas via thermal evaporation ( e.g. , * ? ? ?
* ; * ? ? ?
these processes are tightly linked to the local temperature ( especially the dust temperature ) , and are responsible for setting the gas - phase molecular abundance . at low extinctions where column densities of dust and gas are low , abundances are very sensitive to fuv radiation which can drive volatile chemistry via processes of fractionation and selective photodissociation . the large scatter in figure [ fig : ncol ]
could plausibly be caused by strong variation of such physical / chemical conditions in different sub regions of the surveyed area of the cloud .
if this is the case , we would expect to observe a tighter correlation between column density and if we restricted the plots to cover smaller spatial areas within the surveyed region , since the physical conditions in such smaller regions would be expected to be more uniform .
indeed , @xcite demonstrated the connection between such physical conditions and spatial co abundance variations in the orion cloud .
cccc 01 & 9.21&0.014&0.42 + 02 & 9.31&0.027&0.65 + 03 & 8.30&0.031&0.64 + 04 & 8.52&0.025&0.50 + 05 & 9.14&0.024&0.57 + 06 & 9.23&0.018&0.43 + 07 & 9.99&0.046&0.64 + 08 & 9.19&0.037&0.55 + 09 & 12.3&0.034&0.37 + 10 & 12.3&0.066&0.67 + 11 & 13.1&0.148&0.80 + 12 & 19.6&0.193&0.76 + 13 & 16.5&0.204&0.91 + 14 & 16.9&0.102&0.62 + 15 & 9.07&0.059&0.75 + 16 & 8.65&0.054&0.64 + 17 & 9.66&0.022&0.64 in order to examine the distributions of n ( ) and n ( ) versus on smaller spatial scales , we constructed these relations for each individually observed @xmath23 mapping tile ( c.f .
figures 1 and 2 ) . in figures [ fig : ncol13 ] and [ fig : ncol18 ] we plot the results for and , respectively . in the individual plots contained within these figures we indeed see a significant reduction in scatter compared to that characterizing figure [ fig : ncol ] .
more importantly , there is very strong spatial variation in the relation between extinction and co column density across the surveyed region of the cmc .
this is evident in figure 6 where we plot linear fits to the data .
the results of the fits are shown in table 2 where the slopes vary by nearly an order of magnitude across the cloud .
this is also readily apparent in figure [ fig : ncol18 ] where the axes on the plots have the same scaling and especially clear on close inspection of figure [ fig : ncol13 ] where the scale of the vertical axis ranges by as much as an order of magnitude between the various tiles . these spatial variations in and give rise to the large dispersion in these quantities in the merged relation constructed for the entire observed region as in figure [ fig : ncol ] . despite the variation in the range of column densities at different locations across the cloud
, the range in extinction spanned by the observations in all the plots is very nearly the same
. the large variation of column density of and with dust extinction confirms that the abundances of the two isotopologues exhibit large spatial variations .
consider figure [ fig : ncol13 ] .
the tiles can be placed into two groups , one containing tiles 1 - 10 and 15 - 17 , and the other consisting of tiles 11 - 14 . in the former group ,
rarely exceeds a 5 - 6 @xmath510@xmath77 @xmath74 , while in the latter group can reach values as high as 3 - 5 @xmath5 10@xmath76 @xmath74 over the same extinction range .
the plots in figure [ fig : ncol18 ] exhibit similar behavior . from comparison of figures [ fig : ncol18 ] and [ fig : ncol13 ] with figure [ fig : texmap ] we see that the tiles ( 11 - 14 ) with the largest values of co column density spatially coincide with the region of elevated excitation temperature . since the ( 2 - 1 ) line is generally so optically thick , and since our sampled region preferentially contains high column density ( @xmath10 3 - 5 mag ) gas , sub - thermal excitation of is unlikely
. therefore the excitation temperature map traces the gas kinetic temperature and the distribution of the molecular abundances [ ] and [ ] are tightly related to local physical conditions , in particular the gas kinetic temperature . as a further test of the connection between column density and temperature we have color - coded the points in figures [ fig : ncol13 ] and [ fig : ncol18 ] according to the gas temperature in each individual pixel .
inspection of the figures shows that the pixels with the highest gas temperatures ( @xmath10 15 k ) are also the pixels with the largest co column densities .
pixels with low gas temperatures ( @xmath1 15 k ) rarely reach high co column densities , independent of extinction .
these correlations imply that the abundances of and are physically related to the gas temperature distribution . to examine more directly the relation between the molecular abundances and the gas temperature
, we plot the log ( [ ] ) - log ( ) relation and log ( [ ] ) - log ( ) relation in figure [ fig : abuntex ] .
both molecular abundances show dependence on , although [ ] is apparently sensitive to over a larger range of temperature than [ ] .
this confirms that [ ] and [ ] are both regulated by gas temperature , but over different ranges of temperature in the extinction range ( 3 mag @xmath1 @xmath1 25 mag ) we observed . to quantify the relations
, we empirically fit a power - law functions to the unbinned data for both correlations .
is derived from the line , which has very good snr .
since the noise of extinction map is relatively small , i.e. , 0.05 mag for av@xmath210 mag and 0.1 mag for av@xmath3610 mag , uncertainties in the abundances are dominated by the noise in the molecular line data .
the fits were performed over different ranges of for the two lines : @xmath3 10 k for and @xmath3 17 k for .
the low temperature cutoffs were empirically chosen , by inspection , to correspond to the ranges where the relations appeared to be linear .
however , it did not escape our attention that the low temperature cutoff for corresponds to the co sublimation temperature of @xmath0 17 k @xcite , and as discussed later this may not be coincidental .
the linear regression gave the following results : [ ] @xmath8 t@xmath78 with correlation coefficient = 0.66 ; [ ] @xmath8 t@xmath79 with correlation coefficient = 0.34 .
we performed a @xmath80 test on the power - law model for [ ] - and [ ] - relations .
the @xmath80 is calculated using : @xmath81 where o and e stand for observed and modeled [ co ] ( calculated from the fitted power - law ) , respectively , and @xmath82 the local standard deviation derived in a 2 k bin enclosing the data . at a significance level of 0.05 , the fitted power - law models for [ ] - and [ ]
- relations are acceptable , and the @xmath80 per degree of freedom are 1.02 and 1.06 for [ ] and [ ] , respectively , meaning the fit is reasonable .
the implications of these results will be discussed in the next two sections of the paper .
we argue here that the dependence of the and abundances on gas temperature shown in figure [ fig : abuntex ] is a result of the combined effects of gas depletion and desorption in the cmc .
theoretical considerations suggest that sticking of co onto dust grains can remove the molecule from gas - phase very efficiently within cold ( t@xmath2 17 k ) and dense ( n(h@xmath14 ) @xmath3
10@xmath66 @xmath67 ) regions of molecular clouds ( see , e.g. , * ? ? ?
indeed , compelling observational evidence for reduction of [ ] has been reported for numerous cold ( t @xmath2 10 k ) and high extinction ( @xmath10 10 magnitudes ) regions of molecular clouds .
both the radial stratification of the clouds ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and the presence of emission from high dipole molecules such as nh@xmath83 from such regions indicate that they likely correspond to sufficiently high gas volume densities ( @xmath310@xmath66 @xmath67 ) to promote co condensation onto grain surfaces . in figure
[ fig : noverav ] we plot the relation between ( a ) and ( b ) abundance and extinction in the cmc .
although there is a large scatter in both relations , it is clear that the abundances of the two species decline with extinction in a manner consistent with similar abundance reductions seen in other molecular clouds and suggestive of co freeze out onto grain surfaces in the interior regions of the cloud .
if this decline in abundances at high extinctions is due to freeze out , we also expect the reduction of the gas - phase abundances to be occurring in the coldest regions of the cloud . in figure
[ fig : noverav ] the points are color - coded by the gas temperature and we see both a preponderance of cold gas in the cloud , and a prominent decrease in abundance with extinction for the cold pixels , while the warm pixels show no decrease .
this supports the hypothesis that the decline in abundances with extinction in cold regions is largely due to depletion .
of course , depletion requires low dust temperature and our analysis has been based on measurements of gas temperature .
gas temperature generally tracks dust temperature but direct observations of dust temperature would make for a more compelling case .
fortunately , published maps of dust temperature for the cmc based on herschel observations exist @xcite . comparison with our maps shows generally good agreement between the gas and dust temperature spatial distributions . in particular , in the cold regions away from lkh@xmath13 101 ( corresponding to our tiles 01 - 09 and 15 - 17 , respectively ) , the median @xmath0 14.5 k and the minimum @xmath0 10 k @xcite , far below pure co ice sublimation temperature ( @xmath0 17 k , * ? ? ?
* ) . under such conditions
, @xcite showed a relatively short co depletion timescale , @xmath2 10@xmath84 yr . so with @xmath0 10 k and @xmath85 = 10@xmath66 @xmath67 , co would unavoidably deplete .
this is consistent with the reduction of co abundance we observe in the cold , high extinction regions in the cmc . at the higher gas temperatures ( t @xmath10 17 k ) we would expect that thermal evaporation ( desorption ) of the gas from the grain surfaces releases any trapped co and increases its gas phase abundance relative to the cold regions of the cloud .
indeed , that is what is clearly shown in figure [ fig : abuntex ] where the observed abundances increase with increasing gas temperature .
the warmest co gas in our surveyed region is found in four contiguous tiles ( 11 - 14 ) that are in close proximity to the massive star lkh@xmath13 101 and its associated cluster ( shown as the blue circle in figure [ fig : texmap ] ) .
the cmc dust temperature map from @xcite ( see their figure 4 ) shows a vast region of high dust temperature ( @xmath0 28 k ) near lkh@xmath13 101 , which coincides with the positions of tiles 11 - 14 in our observation .
it is in this warm area that the co abundances appear to be at their highest .
co thermal evaporation is caused by high dust temperature .
for instance , @xcite reported that shortly after a
star turns on " , the dust temperature reaches = 25 k ( @xmath85 = 10@xmath66 @xmath67 , t@xmath70 = 20 k ) , and all co is in the gas phase within 100 yr .
the abundance enhancement near lkh@xmath13 101 is likely caused by hot dust heated by the cluster .
over most of the rest of the cloud @xcite find @xmath0 10 - 14 k. in these regions the gas temperatures are also found to be near 10 k and a significant amount of co should be locked in ice mantles on the dust grains at least at the highest extinctions ( also see * ? ? ?
these considerations suggest that the variation of [ ] and [ ] in the cmc is mostly regulated by the chemistry of depletion and desorption which in turn is driven by the ( dust ) temperature distribution .
recently , @xcite have reported similar results for the orion a molecular cloud .
they divided that cloud into 9 spatial partitions ( 1 - 9 ) and considered regions within those partitions with extinctions , @xmath10 5 magnitudes , similar to those studied here for the cmc .
they found that in regions with 5 @xmath1 @xmath1 10 magnitudes , was generally linearly correlated with .
however , in those partitions ( 2 , 3 & 4 ) that also contained higher extinction material and were characterized by mean gas excitation temperatures that were below @xmath0 22 k , the -relation became flat above @xmath53 10 magnitudes . on the other hand , in two partitions ( 5 & 6 ) with both high extinctions and
high mean gas excitation temperatures ( @xmath0 26 - 33 k ) , continued to rise above @xmath53 10 magnitudes .
@xcite posited that this behavior of with extinction was a result of depletion / desorption effects at @xmath10 10 magnitudes .
our results for the cmc are certainly consistent with those of @xcite for orion a. in this respect , the physical / chemical conditions in the orion and the cmc appear to be quite similar .
the spatial variations in abundances of the rarer isotopes of co are largely determined by depletion / desorption effects due to spatial variations in dust ( and gas ) temperatures resulting from localized star formation activity within the clouds .
in the outer , lower extinction regions of molecular clouds , fuv radiation is expected to play a role in cloud chemistry . in low extinction regions
( i.e. , @xmath1 3 mag ) fractionation of can increase its abundance relative to . at somewhat higher extinctions ( @xmath1 5 mag ) selective photodissociation of is likely to occur due to differences in self - shielding of the fuv intensity at the dissociation wavelengths for the various co isotopes .
the more abundant and isotopes are effectively self - shielded at relatively low extinctions ( @xmath1 1 magnitudes ) but because the dissociation wavelength of is slightly shifted compared to and , and its abundance is relatively low so it is not as effectively self - shielded as the main isotopes , fuv radiation can penetrate deeper into the cloud and dissociate the . since our surveys cover relatively high extinction regions ( @xmath10 5 mag ) we would not necessarily expect selective photodissociation to strongly effect the to abundance ratios . however , earlier studies of @xcite and @xcite suggest that the effects of selective photodissociation may be present at much higher extinctions . in figure [
fig : w1318 ] we plot the ratio of integrated intensities , w()/w ( ) , as a function of extinction .
this ratio is proportional to the abundance ratio for the two species for optically thin lines .
it has been shown to be sensitive to the effects of uv radiation on cloud chemistry and abundances ( e.g. * ? ? ?
* ; * ? ? ?
* ) . for a solar abundance and optically thin lines
, the ratio should be equal to about 5.5 and one would expect the points to scatter around this value , independent of extinction , provided the cloud is characterized by a constant relative abundance throughout and that the excitation of the two species is not too different .
although at lower extinctions ( @xmath1 15 magnitudes ) there indeed appears to be no obvious correlation between this ratio and extinction , the observed values scatter over an enormous range ( @xmath53 0 - 40 ) .
the highest values of the ratio exceed the solar value by nearly an order of magnitude level . to reduce the scatter in the plot due to non - detections of we set the integrated intensity equal to 2 @xmath49 for those pixels where w ( ) is below 2 @xmath49 .
we represent the resulting lower limits on the calculated ratio as triangles in the plot ( see figure ) . ] .
in contrast , at high extinctions ( @xmath10 15 magnitudes ) the relation is relatively flat with a relatively low dispersion , as might be expected for a constant abundance ratio , and is centered at a value ( approximately 4.5 ) near but somewhat less than the solar value .
this behavior does not appear to depend on gas temperature ( see figure ) and thus is not likely a result of depletion or desorption processes on grains .
it is more likely indicative of an active chemical processing of by fuv radiation in regions where @xmath1 15 magnitudes .
the scatter in figure [ fig : w1318 ] is so large at these extinctions because the relative abundance of and is highly unstable possibly due to the stochastic variations in fuv flux resulting , in turn , from such factors as the cloud structure and geometry and nature of the external radiation field . in figure
[ fig : w1318map ] we show a map of the distribution of the integrated intensity ratio over the southern portion of the cmc . examination of this map shows systematically increasing values of the integrated intensity ratio at the edges of the cloud with the material nearest the star lkh@xmath13 101 exhibiting the largest center to edge increases in the ratio .
this latter region appears to contain a well developed photon dominated region ( pdr ) while the dense cloud material to the north is much more quiescent in this regard . in their study of the orion cloud @xcite
reported averaged values of the to integrated intensity ratio ranging from about 10 to 17 in the seven pdrs they observed in that cloud .
moreover they also found the variations in this ratio to be linked to variations in the external fuv radiation field . in the inner , higher extinction regions of the cmc , where fuv photons can not penetrate ,
the abundances are more chemically stable and apparently somewhat below the solar value .
our observations do indicate , however , that fuv photons are present at relatively large ( projected ) depths ( @xmath0 15 magnitudes ) in the cloud confirming results from earlier studies ( e.g. * ? ? ?
* ; * ? ? ?
we now address the earlier observation that the relations between the co abundances and temperature ( figure [ fig : abuntex ] ) show positive correlations but over different temperature ranges .
one might have expected both and molecules to show abundances that increased with temperature over similar temperature ranges , since their binding energies are very similar .
however , the difference may result from the effects of selective uv dissociation of . in the cmc we also find to be correlated with in regions where warm gas is present .
figure [ fig : texavhot ] displays the relation between and for the warm tiles ( 11 - 13 ) in our map .
although the scatter is large at high , there is no doubt of a strong trend between the two quantities with @xmath8 .
interestingly in this region the temperature reaches the co sublimation temperature at roughly @xmath4 6 - 7 magnitudes .
therefore it is possible that , even though is being evaporated from grains at @xmath10 6 magnitudes , ( where @xmath10 17 k ) it is also being selectively photodissociated by deeply penetrating uv radiation until depths of @xmath0 15 magnitudes . for larger values of , dust absorption effectively removes the dissociating uv radiation and higher gas ( dust ) temperatures associated with this region liberate more from grain surfaces .
indeed , the abundances at these higher extinctions and temperatures may reach levels that also allow some self - shielding . consequently , does not reach the expected levels of increased abundance due to evaporation from grains until cloud depths are sufficient to shield the molecule from uv radiation . because of its higher abundance , self - shields at much lower cloud depths and the effects of its increased abundance , due to evaporation from grains , are observed at lower extinctions ( and corresponding lower temperatures ) than . even though h@xmath14 is the dominant constituent of molecular clouds , emission is the most accessible tracer of such gas in the ism .
it is often the only molecular tracer that can be readily detected in distant clouds and galaxies and consequently has been used to estimate the mass of the molecular component of the ism in these systems ( e.g. * ? ? ?
to make such estimations requires knowledge of the so - called x - factor , that is , the co conversion factor , @xmath86 the standard determination of this factor is via use of virial theorem techniques ( e.g. * ? ? ?
* ) . however , the most straightforward way to determine the x - factor is by comparing direct measurements of co and on sub - parsec scales in local clouds .
traditionally this is accomplished by measuring the slope of the correlation between the two quantities at low extinctions ( @xmath2 5 mag ) where the co appears to be effectively thin .
we adopt a slightly different approach and use all our observations to compute a global or average x - factor for the cmc cloud , that is , @xmath2@xmath87@xmath3 = @xmath2@xmath88@xmath3/@xmath2@xmath89@xmath3 @xmath4 2.53@xmath510@xmath6 @xmath74 ( k km s@xmath35)@xmath35 .
this value is 25% larger than the standard milky way value but within the range of values derived for individual molecular clouds in other studies ( e.g. * ? ? ?
another approach is to derive the on - the - spot , that is , we evaluate equation [ eq : xco ] at each pixel in our map and obtain @xmath2@xmath87@xmath90 = @xmath2@xmath91@xmath3 @xmath4 3.1@xmath510@xmath6 @xmath74 ( k km s@xmath35)@xmath35 .
the former @xmath2@xmath3 is equivalent to the latter weighted by the integrated intensity .
however , both recent observational studies @xcite and numerical simulations @xcite have shown that can be strongly dependent on local physical conditions . since the cmc displays a wide range of physical conditions ( e.g. temperature , star formation activity and spatially variable co abundances ) , it would be interesting to measure both and its internal variation on sub - cloud , sub - parsec scales . to do this we first performed the traditional pixel - by - pixel comparison between w@xmath92 and for the entire surveyed region , as shown in figure [ fig : wav12 ] . assuming a constant gas - to - dust ratio n(h@xmath14)/=9.4@xmath510@xmath6 @xmath74 mag@xmath35 @xcite and a constant line integrated intensity ratio w@xmath93/w@xmath15 = 0.7 , x@xmath15 is inversely proportional to w@xmath92/in the plot .
in the milky way disk , the typical value is x@xmath15 @xmath0 2@xmath510@xmath6 @xmath74 ( k km s@xmath35)@xmath35 ( e.g. , * ? ? ? * ) .
this implies a reference value /w@xmath93 @xmath53 0.30 ( k km s@xmath35)@xmath35 mag .
we plot a red line indicating this reference value in figure [ fig : wav12 ] , with a slope of w@xmath93/@xmath53 3.3 ( k km s@xmath35 ) mag@xmath35 . at fixed
, a point above this line indicates a smaller x@xmath93 , and vice versa .
we can see significant scatter in this plot indicating significant spatial variations in x@xmath93 on pixel or sub - parsec spatial scales . in order to investigate the spatial dependence of x@xmath93 , we determined the w@xmath92 relation in each individual tile of our co survey , following section [ subsec : tilewav ] .
these relations are shown in figure [ fig : twav12 ] .
there are several familiar features of these plots that are similar to what we have seen in figure [ fig : ncol13 ] , i.e. , less scatter in the relations within the individual tiles , systematic spatial variations in the w@xmath92 relation across the cloud , and good correlation between high w@xmath92 and high gas temperature ( @xmath10 15 k )
. the general similarity - ( figure [ fig : twav12 ] ) and n ( ) - ( figure [ fig : ncol13 ] ) relations .
the apparent relative differences between the the cold tiles in figure [ fig : ncol13 ] compared to figure [ fig : twav12 ] is a result of different y - axis scaling . ] in the behavior of w@xmath92 and n ( ) with extinction is perhaps surprising given that the emission is very optically thick while emission is considerably less thick and mostly optically thin . since is proportional to w@xmath94 , consider that in this situation @xmath95 @xmath96 \times \tau^{13}_\nu \delta v_{13}$ ] , while @xmath97 @xmath98 \times \delta v_{12}$ ] , where @xmath99 is the line width of the corresponding spectrum . unlike w@xmath94 ,
w@xmath92 has no dependence on @xmath61 or column density , but it is proportional to , the gas excitation temperature . particularly interesting in this regard is the behavior of w@xmath92 in tile 13 near where the embedded cluster is heating the cloud . here , there appears to be a clear linear relation between w@xmath92 and to the highest measured extinctions ( @xmath53 25 magnitudes ) despite the fact that the emission in this region of the cloud is so saturated that it is self - reversed ( see figure [ fig : spec ] ) . because the co emission is so optically thick , this behavior must be primarily the result of the gradient of with observed in this region of the cloud ( see figure [ fig : texavhot ] ) .
this is also evident in figure [ fig : wav12tile13 ] which shows an expanded view of the w@xmath92relation for tile 13 .
we can clearly see a temperature gradient in the distribution of color - coded points from low to high . both w(co ) and t increase by a factor of @xmath0 5 - 6 from low to high .
this indicates that the temperature gradient accounts for the bulk of the change in w(co ) with in this region .
of course , other factors could contribute to the increase in w(co ) such as an increase in velocity dispersion with temperature , or even increased [ co ] due to desorption as with and as discussed previously .
we do find some evidence for a marginal ( 35% ) increase in the velocity dispersion of the hot gas , but its magnitude is far from sufficient to account for the increased ( @xmath100 ) measures of w(co ) .
a desorption induced increase in co abundance most certainly has occurred in the hot gas , and although the line is very optically thick , we might expect that additional photons could leak out in the line wings and possibly increase the measured velocity dispersion .
but as we already remarked , this effect can not be significant since the change in peak brightness is observed to be comparable to the change in w(co ) and substantially larger than changes in the velocity dispersion .
consequently , in the set of tiles near the massive star lkh@xmath13 101 ( 11 - 14 ) we conclude that heating of co enhances w@xmath92 in a manner that induces a positive correlation with extinction despite the fact that the emission is saturated .
these observations suggest that heating by ob stars produces gradients in the excitation temperature of the co emitting gas such that in warm regions w ( ) is proportional to and thus to n(h@xmath14 ) over a wide range of extinctions .
this , in turn , indicates that a single , meaningful x - factor can be derived for these hotter regions using the observations . in figure
[ fig : twav12 ] , tile 13 shows the best correlation between w ( ) and .
we performed a linear regression for tile 13 , which is shown in figure [ fig : wav12tile13 ] .
our result implies @xmath101 in this tile over the range 3 mag @xmath2 @xmath2 25 mag . from the average of the individual , on - the - spot , values in tile 13 we find @xmath2@xmath3 @xmath4 @xmath102 , in excellent agreement with the value derived from the linear regression
this clearly supports the idea of a constant value in the hot gas . the derived value is less than both that found for the global average of the cmc and the value ( @xmath103 ) usually adopted for the milky way .
this result would suggest that in regions heated by ob stars ( e.g. , as hii regions , extragalactic starbursts , etc . ) , reasonably accurate gas masses could be derived using a standard x - factor analysis with a single value of the x - factor .
unfortunately it is not at all clear how one would predict the appropriate value to use in all situations .
however some clues may be found in the analysis of the regions away from ob stars , where there is only ambient heating of the gas from cosmic rays and the general galactic uv radiation field .
here we observe no correlation between w@xmath92 and at modest to high extinctions ( i.e. , @xmath10 3 - 5 mag ) .
this behavior is due to the fact the line is almost always saturated at high extinctions and the cold gas is very nearly isothermal .
consequently , these cold , quiescent regions , are not characterized by a single empirical x - factor , instead , the x - factor systematically increases with extinction and , in the cmc , is almost always larger than the milky way value .
this is shown in figure [ fig : xcovsav]a where we plot the relation between the on - the - spot x - factor and for the regions of the cloud not heated by the cluster ( i.e. , tiles 1 - 10 and 15 - 17 ) .
as expected there is a clear correlation and a linear least - squares fit to the unbinned data finds that @xmath8 n(h@xmath14)@xmath104 . for the cold tiles we find @xmath2@xmath3 = 3.4 @xmath5 10@xmath6 @xmath7 .
the ( constant ) x - factor in the hotter regions ( e.g. , tile 13 ) has a value considerably lower than those in the cold regions , suggesting that may also increase with decreasing gas temperature within the cloud .
such a trend is clearly evident in our observations and is shown in figure [ fig : xcovsav]b where we plot the relation between the on - the - spot and for the entire cloud .
a linear least - square fit to the ( binned ) data gives , @xmath4 2.0 @xmath5 10@xmath6(/10)@xmath11 @xmath7 .
in this instance we chose to fit the binned data to give greater weight to high temperature measurements .
a fit to the unbinned data gives a slightly steeper slope ( -0.9 vs. -0.7 ) .
the observed inverse correlation between and is potentially interesting .
it suggests that differences between globally averaged x - factors derived for different clouds could be a result of variations in the relative amounts of hot and cold gas within the clouds , provided their velocity dispersions are similar .
active star forming clouds with ob stars and hii regions , such as orion , might be expected to have average x - factors that are lower than more quiescent clouds , such as the cmc .
indeed , @xcite derive @xmath4 1.35 @xmath5 10@xmath6 @xmath7 for the entire orion complex from @xmath105 ray data , in agreement with recent determinations using co @xcite
. moreover , @xmath105 ray observations of orion with the fermi satellite by @xcite found that in the ( cold ) regions removed from the ob stars and hii regions , @xmath4 2.2 @xmath5 10@xmath6 @xmath7 , while in the ( hotter ) parts of the cloud complex associated with the hii regions , @xmath4 1.3 @xmath5 10@xmath6 @xmath7 , very similar to what we find in the california cloud . for the taurus cloud , which has no ob stars or hii regions , @xcite find = 2.1 @xmath5 10@xmath6 @xmath7 from co observations of that cloud . a relatively high value ( 2.54 @xmath5 10@xmath6 @xmath7 ) for was also derived from planck observations of high latitude diffuse and presumably cold , co gas .
we noted earlier that the global x - factor for the cmc was 25% higher than the milky way value and this is may be due to the relatively large tracts of cold regions included in the average for the cmc .
of course , variations in velocity dispersions between clouds could also contribute to the variation in their observed x - factors .
however , we would expect similar velocity dispersions to characterize gravitationally bound or virialized clouds of similar mass and size .
moreover , since the average column densities of molecular clouds are also constant , we would expect @xmath2@xmath3 @xmath0 @xmath2t@xmath106 for such clouds .
as recently pointed out by @xcite , galactic gmcs are characterized by very similar physical properties resulting in x - factors that span a relatively narrow range of values .
they suggest that this is a result of the competition between stellar feedback and gravity that is necessary to maintain the relative stability of these objects . here
we are suggesting that variation within that narrow range of observed may be due to variations in the relative fractions of hot gas in these clouds .
this hot gas , of course , is a direct product of the amount of stellar feedback the clouds are experiencing .
resolved maps of the internal x - factor distribution in additional clouds would be useful to determine the relative contribution of temperature to the derived values of within gmcs in general and could directly test this idea .
absent of such additional data , it is instructive to compare our observations to recent simulations of turbulent cloud evolution that resolve synthetic clouds and derive the internal spatial distribution of the x - factor @xcite .
these simulations are thus directly comparable to observations such as ours . the shetty et al .
simulations include chemical evolution and radiative transfer to enable predictions of co line emission
. however , the calculations do not include gas - dust chemical interactions and are thus not relevant to our earlier findings or discussion regarding [ ] , [ ] , and w()/w ( ) .
but since is so optically thick we safely assume that effects of depletion / desorption are not important for our x - factor determinations and thus comparisons with these models are appropriate . because the shetty et al .
simulations do not incorporate stellar feedback , the most appropriate comparison will be with our data in the regions of the cmc away from the embedded cluster ( all tiles excluding 11 - 14 ) .
the simulated cloud of @xcite whose physical conditions match most closely those of typical gmcs is characterized by a global lower than we observe in the cold tiles of the cmc map ( 2.2 vs. 3.4 @xmath5 10@xmath107 ) but closer to the milky way value . on sub - cloud scales
@xcite find a correlation between and ( see their figure 4 ) similar to but somewhat weaker than we observe in the cmc .
the simulations also predict that there should be a correlation between the x - factor and the velocity dispersion in the cloud , with @xmath8 @xmath108 . in figure
[ fig : xcovssigma ] we plot against the velocity dispersion , @xmath109 in the cold cloud regions is the second moment of the ( 2 - 1 ) profile @xmath110 where m1 is the first moment @xmath111 .
@xmath109 equals the velocity dispersion if line profile is gaussian . ] .
we find no correlation between these same quantities contrary to the predictions of the simulation .
although the simulations of individual clouds do not include any systematic temperature gradients or increases due to stellar heating , simulations were performed to investigate how increasing the overall cloud temperature can effect .
these simulations predict a trend of decreasing x with t , consistent with what we observe here .
the simulations find @xmath8 t@xmath112 for different temperature clouds which is only somewhat weaker than what is observed here , i.e. , @xmath8 t@xmath11 within the cmc .
overall , we conclude that the agreement between these simulations and the cmc is mixed and it is unclear whether any similarities between the simulations and the observations of the california cloud are more than coincidental .
in this study we presented the results of extinction and molecular - line mapping toward the most active star forming region within the nearby california gmc . the relationship between dust and gas properties in this active portion of the cloud
was investigated with the following results : \1 .
the lte molecular abundances of and in the cloud exhibit significant spatial variations .
these variations are correlated with the spatial variation in gas ( and presumably dust ) temperature in the cloud .
these temperature variations are likely caused by the heating from the adjacent , partially embedded , lkh@xmath13 101 cluster .
we argue that the spatial variations in the derived and gas abundances are largely due to temperature dependent gas depletion / desorption on / off dust grains .
the abundance ratio between and increases in lower extinction regions of the california cloud , suggesting selective photo - dissociation of in those regions .
this is likely caused by ambient uv radiation which apparently penetrates relatively deeply ( i.e. , @xmath1 15 mag ) into the cloud , particularly near the embedded cluster .
there is also evidence that the selective photo - dissociation of can suppress its abundance even in warm regions heated by the embedded cluster where evaporation off grain surfaces would otherwise lead to an increase in the abundance at these cloud depths .
dramatic spatial variations are also observed in the relationship between the integrated intensity , w@xmath92 , and , particularly at the intermediate and high extinctions ( @xmath10 3 - 5 mag ) primarily probed in this study .
however , unlike the case for its two rarer isotopologues , the variation in integrated intensity appears to be a direct result of a spatial gradient of the excitation temperature ( and gas kinetic temperature ) with extinction and not the result of any detectable variation in the abundance due to depletion / desorption effects .
we compute the x - factor ( @xmath4n(h@xmath14)/w(co ) ) for each individual pixel in our map and average the results to obtain @xmath2@xmath3 @xmath4 2.53 @xmath5 10@xmath6 @xmath7 , a value somewhat higher than the milky way average . \5 . on the sub - parsec scales in the cmc
there is no single empirical value of the x - factor that can characterize the molecular gas in cold ( @xmath1 15 k ) regions with @xmath10 3 magnitudes . for those regions we find that @xmath8 a@xmath113 with @xmath2@xmath3 @xmath4 3.4 @xmath5 10@xmath6 @xmath7 .
we do however find a clear correlation between w ( ) and in regions containing relatively hot ( @xmath10 25 k ) molecular gas at @xmath10 3 magnitudes suggesting that , unlike the cold gas , the warm material may be characterized by a single value of .
however in this warm gas we find a value for the x - factor , @xmath4 1.5 @xmath5 10@xmath6 @xmath7 , significantly lower than the averages for the cold gas , the overall cmc and the milky way .
overall we find an ( inverse ) correlation between and with @xmath8 @xmath11 .
such a correlation may potentially explain the observed variations in global x - factors between gmcs as being due in large part to variations in the relative amounts of warm gas heated by ob stars within the clouds .
we thank kelsey jorgenson for assistance with acquiring the co observations and acknowledge useful discussions with karin berg .
suggestions from an anonymous referee led to numerous improvements in the paper .
carlos roman acknowledges support from conacyt project cb2010 - 152160 , mexico .
this research is based in part on observations collected at the centro astronmico hispano alemn ( caha ) at calar alto , operated jointly by the max - planck - institut fr astronomie and the instituto de astrofsica de andaluca ( csic ) .
the heinrich hertz submillimeter telescope is operated by the arizona radio observatory , which is part of steward observatory at the university of arizona .
this work was supported in part by national science foundation grant ast-1140030 to the university of arizona .
ackermann , m. , ajello , m. , allafort , a. , et al .
2012 , , 756 , 4 alves , j. , lada , c. j. , lada , e. a. , kenyon , s. j. , & phelps , r. 1998 , , 506 , 292 alves , j. , lada , c. j. , & lada , e. a. 1999 , , 515 , 265 alves , j. f. , lada , c. j. , & lada , e. a. 2001 , , 409 , 159 bergin , e. a. , ciardi , d. r. , lada , c. j. , alves , j. , & lada , e. a. 2001 , , 557 , 209 bergin , e. a. , alves , j. , huard , t. , & lada , c. j. 2002 , , 570 , l101 bergin , e. a. , & tafalla , m. 2007 , , 45 , 339 bergin , e. a. , langer , w. d. , & goldsmith , p. f. 1995 , , 441 , 222 bieging , j. h. , peters , w. l. , & kang , m. 2010 , , 191 , 232 bolatto , a. d. , wolfire , m. , & leroy , a. k. 2013 , , 51 , 207 bohlin , r. c. , savage , b. d. , & drake , j. f. 1978 , , 224 , 132 caselli , p. , walmsley , c. m. , tafalla , m. , dore , l. , & myers , p. c. 1999 , , 523 , l165 dickman , r. l. , 1978 , , 37 , 407 digel , s. w. , aprile , e. , hunter , s. d. , mukherjee , r. , & xu , f. 1999 , , 520 , 196 forbrich , j. , berg , k. , lada , c. j. , et al .
2014 , , 568 , aa27 frerking , m. a. , langer , w. d. , wilson , r. w. , 1982 , , 262 , 590 goldsmith , p. f. , & langer , w. d. 1978 , , 222 , 881 goodman , a. a. , pineda , j. e. , & schnee , s. l. 2009 , , 692 , 91 harvey , p. m. , fallscheer , c. , ginsburg , a. , et al .
2013 , , 764 , 133 kramer , c. , alves , j. , lada , c. j. , et al . 1999 , , 342 , 257 lada , c. j. , lada , e. a. , clemens , d. p. , & bally , j. 1994 , , 429 , 694 lada , c. j. , lombardi , m. , & alves , j. f. , 2009 , , 703 , 52 lada , c. j. , lombardi , m. , & alves , j. f. 2010 , , 724 , 687 langer , w. d. , & penzias , a. a. 1993 , , 408 , 539 larson , r. b. 1981 , , 194 , 809 lee , m .- y . ,
stanimirovi , s. , wolfire , m. g. , et al .
2014 , , 784 , 80 , j. 2006 , phd thesis , university of florida li , d. l. , esimbek , j. , zhou , j. j. , et al .
2014 , , 567 , a10 lombardi , m. , & alves , j. 2001 , , 377 , 1023 lombardi , m. , alves , j. , & lada , c. j. 2006 , , 454 , 781 lombardi , m. 2009 , , 493 , 735 lombardi , m. , alves , j. , & lada , c. j. 2010 , , 519 , ll7 narayanan , d. , & hopkins , p. f. 2013 , , 433 , 1223 pineda , j. e. , caselli , p. , & goodman , a. a. 2008 , , 679 , 481 pineda , j. l. , goldsmith , p. f. , chapman , n. , et al .
2010 , , 721 , 686 planck collaboration , ade , p. a. r. , aghanim , n. , arnaud , m. , et al .
2011 , a&a , 536 , a19 rachford , b. l. , snow , t. p. , destree , j. d. , et al .
2009 , , 180 , 125 ripple , f. , heyer , m. h. , gutermuth , r. , snell , r. l. , brunt , c. m. , 2013 , , 431 , 1296 rllig , m. , & ossenkopf , v. 2013 , , 550 , a56 , c. g. 2006 , phd thesis , university of florida romn - ziga , c. g. , alves , j. f. , lada , c. j. , & lombardi , m. 2010 , , 725 , 2232 sandstrom , k. m. , leroy , a. k. , walter , f. , et al .
2013 , , 777 , 5 sault , r. j. , teuben , p. j. , & wright , m. c. h. 1995 , astronomical data analysis software and systems iv , 77 , 433 shetty , r. , glover , s. c. , dullemond , c. p. , et al . 2011a , , 415 , 3253 shetty , r. , glover , s. c. , dullemond , c. p. , & klessen , r. s. 2011b , , 412 , 1686 shimajiri , y. , kitamura , y. , saito , m. , et al .
2014 , , 564 , a68 solomon , p. m. , rivolo , a. r. , barrett , j. , & yahil , a. 1987 , , 319 , 730 taylor , m. b. 2005 , astronomical data analysis software and systems xiv , 347 , 29 van der tak , f. f. s. , black , j. h. , schier , f. l. , jansen , d. j. , & van dishoeck , e. f. 2007 , , 468 , 627 van dishoeck , e. f. , & black , j. h. 1988 , , 334 , 771 van dishoeck , e. f. , blake , g. a. , draine , b. t. , & lunine , j. i. 1993 , in protostars and planets iii , ed .
e. h. levy & j. i. lunine ( tucson , az : univ .
arizona press ) , 163 visser , r. , van dishoeck , e. f. , & black , j. h. 2009 , , 503 , 323 | we present results of an extinction co line survey of the southeastern part of the california molecular cloud ( cmc ) .
deep , wide - field , near - infrared images were used to construct a sensitive , relatively high resolution ( @xmath0 0.5 arc min ) ( nicest ) extinction map of the region . the same region was also surveyed in the ( 2 - 1 ) , ( 2 - 1 ) , ( 2 - 1 ) emission lines at the same angular resolution .
these data were used to investigate the relation between the molecular gas , traced by co emission lines , and the dust column density , traced by extinction , on spatial scales of 0.04 pc across the cloud .
we found strong spatial variations in the abundances of and that were correlated with variations in gas temperature , consistent with temperature dependent co depletion / desorption on dust grains .
the to abundance ratio was found to increase with decreasing extinction , suggesting selective photodissociation of by the ambient uv radiation field .
the effect is particularly pronounced in the vicinity of an embedded cluster where the uv radiation appears to have penetrated deeply ( i.e. , @xmath1 15 mag ) into the cloud .
we derived the cloud averaged x - factor to be @xmath2@xmath3 @xmath4 2.53 @xmath5 10@xmath6 @xmath7 , a value somewhat higher than the milky way average . on sub - parsec scales
we find there is no single empirical value of the x - factor that can characterize the molecular gas in cold ( @xmath1 15 k ) cloud regions , with @xmath8 a@xmath9 for @xmath10 3 magnitudes .
however in regions containing relatively hot ( @xmath10 25 k ) molecular gas we find a clear correlation between w ( ) and over a large ( 3 @xmath1 @xmath1 25 mag ) range of extinction .
this results in a constant @xmath4 1.5 @xmath5 10@xmath6 @xmath7 for the hot gas , a lower value than either the average for the cmc or the milky way .
overall we find an ( inverse ) correlation between and in the cloud with @xmath8 @xmath11 .
this correlation suggests that the global x - factor of a giant molecular cloud ( gmc ) may depend on the relative amounts of hot gas contained within the cloud . |
the small - world phenomenon is one of the most intriguing properties of human society .
this describes the fact that unrelated people in a society , who are a very large geographic distance apart from one another , tend to be connected by surprisingly short chains of acquaintances .
this phenomenon was hypothesized in 1929 by hungarian author frigyes karinthy @xcite and was first observed experimentally in the 1960 s with sociologist stanley milgram s seminal experiments @xcite wherein randomly chosen people were selected to mail a letter to an unknown target person , but were only allowed to send the letter to a friend , who would pass the letter along to another friend , etc .
, until the target was reached .
successful transmissions took surprisingly few intermediate people , lending credibility to the turn of phrase ` six degrees of separation , ' popularized by karinthy . understanding this phenomenon is an important sociological problem .
to study the underlying mechanism that led to milgram s results , computer scientist jon kleinberg modeled a society as follows @xcite .
begin with a large , regular square @xmath8 lattice .
each node is connected to its nearest lattice neighbors and to a single random node a large distance away .
the probability of nodes @xmath9 and @xmath10 being connected by such a long range contact is @xmath11 where @xmath12 is the euclidian distance between the two nodes and the sum runs over all nodes in the network except @xmath9 .
physically , the local lattice connections represent associations with immediate neighbors , fellow townspeople , etc . , while long - range contacts might model friends or relatives in another city or country .
we seek to pass the message from a random starting node @xmath13 to a random target @xmath14 .
of great importance is the fact that each node has no information beyond the locations of its contacts and the target node @xmath14 , so the operational algorithm must be _ local _ in character .
kleinberg has proved that no local algorithm can do better , functionally , than the _ greedy _ algorithm @xcite : each message holder passes the message along to whichever of its contacts is closest to @xmath14 , until the message reaches the target .
moreover , for @xmath15 the delivery time @xmath16 ( number of intermediate steps ) scales as a _ power _ of @xmath6 , while small world behavior and the weakest dependence on lattice size emerges for @xmath7 , where @xmath17 .
the distribution of nodes on a regular square lattice is too rigid , failing to mimic important features of actual distributions of populations ( or computer routers , etc . ) . in an effort to account for these
we have studied kleinberg navigation in fractals @xcite , showing that the optimal long - contact exponent is then @xmath18 , the fractal dimension of the lattice . in this letter
we study the effects of anisotropy another commonly encountered distortion of the ideal kleinberg lattice .
our results indicate that in the limit of lattice size @xmath4 the optimal contact exponent for two - dimensional lattices is still @xmath7 . for finite @xmath6
, we find an optimal exponent @xmath5 quite different from the infinite limit . the convergence to @xmath7 as @xmath4 shows interesting power - law dependence on the anisotropy strength .
we wish to study the isolated effect of anisotropy on kleinberg navigation .
to do this , we begin with a regular square lattice ( @xmath19 ) and introduce one of two forms of anisotropy .
* * lattice anisotropy : * the underlying lattice is stretched horizontally , along the @xmath20-axis , by a factor @xmath21 , such that the area of each cell goes from @xmath22 to @xmath23 . * * angular anisotropy : * long - range contacts are chosen preferably along the vertical direction , by a factor @xmath21 .
more precisely , the random angle @xmath24 of each long - range contact vector ( measured counter - clockwise , from the @xmath20-axis ) is modified to @xmath25 : @xmath26 both of these types of anisotropy tend to favor connections in the vertical @xmath27-direction , if @xmath28 , and along the @xmath20-direction , if @xmath29 .
to simulate kleinberg navigation efficiently , we use several tricks and approximations .
first , rather than testing a finite - size square @xmath8 lattice , we consider an infinite lattice and place the source and target at distance @xmath6 from one another .
since the message always progresses toward the target , by the greedy algorithm , the message holder remains within a disc of radius @xmath6 centered on the target node , so in practice only a finite number of sites would be explored anyways .
second , the computation of the normalizing sum in the denominator of eq .
( [ eqn : kleinbergnavigationprob ] ) is dependent ( in finite lattices ) upon the location of the node @xmath9 , and can be very time consuming .
the infinite lattice circumvents this problem , as the normalizing constant is the same for all nodes .
note , however , that @xmath30 does not converge for @xmath31 . in that case
, we imagine a lattice larger than the @xmath6-disc of activity , with periodic boundary conditions , such that the normalizing factor is still the same for all sites . because of the monotonic progression toward the target no site
is ever revisited in the process . moreover , as observed by kleinberg @xcite , one can think of the long - contact link out of node @xmath9 as being created at the very instant that the message arrives at @xmath9 .
thus , the full lattice is unnecessary , and we need keep in memory only the current location of the message holder ( and the location of the target ) .
when the message arrives at @xmath9 , we create a random long contact , compare the distances of all five neighbors of @xmath9 ( the four lattice neighbors and the long contact ) to the target , and move the message to the site closest to the target .
the long contact is created by choosing a random distance and angle , @xmath32 . in order to reproduce the correct @xmath33 the distance @xmath0
is taken from the distribution @xmath34 , to account for the linear growth of the area of the ring where the contact might fall .
the angle @xmath24 is distributed uniformly between 0 and @xmath35 . in the case of angular anisotropy ,
@xmath24 is replaced by @xmath25 , according to ( [ eqn : anisoangledef ] ) .
finally , a vector @xmath32 is drawn from site @xmath9 and the contact is placed on the site @xmath10 closest to the vector tip .
because of the anisotropy , the angular displacement from the source to the target makes a difference .
it is sufficient to test only the two extremes of @xmath36 and @xmath37 where the target is either parallel or perpendicular to the anisotropy direction .
we note , however , that anisotropy strength @xmath38 and a target at @xmath36 is equivalent to anisotropy @xmath39 and target at @xmath37 . for this reason
, we simply set the source and target at @xmath40 and @xmath41 , respectively , throughout , and let @xmath38 vary both below and above the isotropic divide of @xmath42 .
simulations were performed for various values of @xmath38 over a large range of @xmath2 and @xmath6 , each averaged 1000 times . for each @xmath38 and @xmath6 ,
the minimum @xmath2 was computed by first fitting a fifth - order polynomial to the averaged data , then using newton s method on the polynomial s derivative .
finally @xmath43 was plotted as a function of @xmath44 for each chosen value of @xmath38 .
these are shown in fig .
[ fig : anisodata ] and indicate that @xmath45 as @xmath46 , regardless of @xmath38 . in fig .
[ fig : accuracyextrap ] we show detailed results of the extrapolation to @xmath4 . to further clarify the behavior shown in fig .
[ fig : anisodata ] , the following procedure was performed .
first fit a cubic polynomial @xmath47 , using least squares , to each @xmath38 s curve .
then , subtract that polynomial from the isotropic case , @xmath48 .
this maps @xmath42 to the horizontal axis and gives the behavior of the @xmath49 curves `` relative '' to the isotropic curve .
these are shown in fig .
[ fig : renormedcubics ] .
the different behavior for each type of anisotropy is clear : for both types of anisotropy the results for @xmath28 show dramatic differences from the isotropic case of @xmath42 ( the differences for @xmath50 and large @xmath6 are negligible ) . for lattice anisotropy ,
the @xmath28 curves start above the @xmath42 curve and cross below until they eventually converge at a similar rate , as @xmath4 . for the angular anisotropy
, the @xmath28 curves approach @xmath51 at a different rate than the @xmath42 curve , resulting in distinctly different slopes in the plots of fig .
[ fig : angleanisodata ] .
the observed `` crossover '' behavior present in the lattice anisotropy is somewhat unexpected . the crossover point , @xmath52 ,
is explored by finding the zero of each @xmath48 .
these are plotted in fig .
[ fig : logloglatticecrossover]a , and seem to indicate a power law relationship , @xmath53 .
likewise , the different slopes for angular anisotropy , plotted in fig . [ fig : logloglatticecrossover]b , show power - law behavior and seem to increase roughly as @xmath54 . what is responsible for these phenomena remains an open question .
we have shown , by extensive numerical simulations , that kleinberg navigation in two - dimensional lattices with two types of anisotropy displays the same gross characteristics as navigation in isotropic lattices .
in particular , the optimal long - contact exponent in the limit of infinitely distant source and target remains @xmath7 , even in the presence of anisotropy .
it is worthwhile to note that the actual values for the optimal exponent @xmath5 for finite @xmath6 can differ considerably from the limit @xmath7 , even for reasonably large lattices .
thus , for practical applications the optimal exponent ought to be evaluated on a case by case basis .
the modes of convergence to the limit @xmath4 show intriguing power - law dependence upon the strength of the anisotropy .
a theoretical explanation for this behavior remains the subject of future work .
we gratefully acknowledge nsf award phy0555312 for partial funding ( dba ) , and the support of a nsf graduate research fellowship ( jpb ) .
jmc has been supported as a summer research undergraduate student by the mcnair program at carkson university . | we study the kleinberg problem of navigation in small world networks when the underlying lattice is stretched along a preferred direction .
extensive simulations confirm that maximally efficient navigation is attained when the length @xmath0 of long - range links is taken from the distribution @xmath1 , when the exponent @xmath2 is equal to @xmath3 , the dimension of the underlying lattice , regardless of the amount of anisotropy , but only in the limit of infinite lattice size , @xmath4 . for finite size lattices
we find an optimal @xmath5 that depends strongly on @xmath6 . the convergence to @xmath7 as @xmath4 shows interesting power - law dependence on the anisotropy strength . |
techniques that probe the momentum dependence of strongly correlated electrons provide valuable information that can be used as a test for models appropriate for the cuprates .
angle - resolved photoemission ( arpes ) provides important information about single - particle excitations of occupied states.@xcite electronic raman scattering reveals information about multiparticle excitations at long wavelengths .
@xcite resonant inelastic x - ray scattering ( rixs ) , a form of raman spectroscopy , recently has undergone vast improvements , allowing probes of momentum dependent excitations into unoccupied states .
@xcite improvements in resolution now offer the possibility of a direct comparison of arpes , raman , and rixs spectra in the several ev range with the hope of understanding the nature of strong correlations in the cuprates , and high temperature superconductivity , by studying electronic excitations at several energy scales . as a fundamental building block of low - energy effective theories of the cuprates , the zhang - rice singlet ( zrs )
@xcite is the starting point of the single - band hubbard and t - j models widely believed to capture the low energy physics of strongly correlated systems .
it is well - known that the zrs features prominently in rixs in the cuprates .
a particular test of models of the cuprates lies in the difference between rixs in edge - shared versus corner - shared copper oxide systems . in materials such as la@xmath0cuo@xmath4 and other corner sharing 2d members of the high t@xmath5 family ,
the zrs is stabilized by a strong gain in antiferromagnetic exchange energy between cu and o , and can propagate to neighboring cuo@xmath4 plaquettes , with effective hopping @xmath6 ev along the cu - o bond direction , and @xmath7 ev along the diagonal.@xcite in edge - shared cuprates , such as li@xmath0cuo@xmath0 and cugeo@xmath1 , the cu - o - cu bond approximately forms a @xmath8-angle,@xcite compared to @xmath9 in corner - shared cuprates , and therefore according to goodenough - kanamori - anderson,@xcite the nearest - neighbor copper - copper exchange tends to be slightly ferromagnetic .
however , higher order processes like cu - o - o - cu hopping also contribute comparable antiferromagnetic exchange terms .
for these reasons , small bond - angle variations among different edge - sharing compounds can make the exchange slightly anti - ferromagnetic or ferromagnetic . in either case , the zrs is less stable and mobile due to hybridization among different oxygen orbitals not aligned with cu @xmath10 orbitals .
@xcite recent cu @xmath11 edge rixs data on edge - sharing cuprate li@xmath3cuo@xmath3 @xcite has revealed a small , non - dispersive peak at 2.1 ev , attributed to interatomic @xmath12 excitations , a strong peak at 5.4 ev , and a weak peak at 7.6 ev , both attributed to charge transfer excitations .
the peak at 5.4 ev is doubly resonant for incident photon energies near 8986 and 9000 ev.@xcite in cugeo@xmath1 , clear excitations were found are 3.8 ev ( weak ) and 6 ev ( strong),@xcite and more recently 1.7 ev,@xcite which were attributed in ref . as a intra - atomic @xmath12 excitation ( 1.7 ev ) , a zrs excitation ( 3.8 ev ) , and a charge transfer excitation on a single copper oxide plaquette ( 6.4 ev ) .
similar features have also appeared in oxygen 1s rixs .
@xcite while a few momentum - dependent studies have been performed , the incident photon energy dependence , or resonance profile , is usually shown only for a few selected points in the brillouin zone . in this paper
we present a theory for rixs for the copper @xmath2-edge and present exact diagonalizations of hubbard clusters of edge - sharing copper oxide plaquettes shown in fig .
[ cluster ] .
we calculate both the rixs spectrum and the resonance profile for several cluster geometries , and determine the nature of excitations accessible via light scattering .
we pay specific attention to the intermediate core - hole and final excited - state wavefunctions of rixs accessible states .
we find good agreement with the results on edge - shared copper oxide systems , and confirm a number of previous excitation assignments .
the rixs process consists in scattering an incident photon of energy @xmath13 off a sample in the ground state @xmath14 having energy @xmath15 . for @xmath11edge rixs ,
an electron from the @xmath16 copper - core is photoexcited into the cu @xmath17-band , leaving behind a core hole .
the strong coulomb repulsion between the core hole and @xmath10 holes on the same cu atom reorganizes charge density , forming an intermediate state @xmath18 of eigenenergy @xmath19 .
the @xmath20 electron resides in a rather extended wave - function and therefore interacts weakly with @xmath21 and core states .
we neglect its interaction hence providing us with a rather simple description of rixs .
the @xmath20 electrons then enter into the problem as a spectator at energy @xmath22 and the intensity will be proportional to the projected @xmath20-band density of states ( dos ) at that energy .
the incident polarization determines the necessary orbital projection .
the @xmath20 electron then recombines with the @xmath16 core hole emitting a photon of energy @xmath23 , leaving the system in the final state @xmath24 of eigenenergy @xmath25 and a photon energy transfer @xmath26 .
this leads to a simplification of the general expression given in ref .. in this approximation the spectrum will be given by the following relation and with the @xmath13 dependence given by a convolution with the @xmath20 band dos : @xmath27 here @xmath28 represents the cu @xmath29 energy separation . for simplicity , in a first approach , we neglect specific @xmath20 orientation and photon polarizations .
the polarization dependence , as well as the related @xmath20 orientation and the @xmath20 projected dos , will be discussed in the last part of section [ section_res ] .
the parameter @xmath30 represents damping of the intermediate state due to fluorescence and auger @xmath16 decay , which we take as 1.0 ev . while this value determines the overall resonant enhancement , it does not affect the resolution of the energy - loss peaks in the spectrum .
we broaden the delta function with a width 0.1ev to represent current instrument resolution .
we note that in eq.([rixs - formula ] ) the states @xmath18 are identical to those in a core x - ray photoemission experiment .
the hamiltonian describing the open boundary cluster shown in fig .
( [ cluster ] ) , with rotated @xmath31 and @xmath32 local - directions along the @xmath33 orbitals , is given by @xmath34 $ ] , with @xmath35 here @xmath36 and @xmath37 creates and annihilates a @xmath38 , @xmath39 , planar @xmath33 or @xmath40 hole with spin @xmath41 at site @xmath42 , respectively , and the sum runs over nearest neighbors .
the hopping amplitude @xmath43 refers to the hopping between two nearest @xmath44 @xmath45 orbitals , and @xmath46 refers to the hopping between two nearest - neighbor @xmath47 oxygen orbitals , where @xmath48 .
the overlap phase factors are chosen such that @xmath49 for neighbors in the @xmath50 direction , and -1 for @xmath51 .
the oxygen phase factors are analogously defined as @xmath52 for oxygens at @xmath53 and @xmath54 , and -1 for the other combinations . here
@xmath55 ( resp .
@xmath56 ) creates ( annihilates ) a cu@xmath57 core - electron , and @xmath58 is the core - @xmath21 hole repulsion . for simplicity
we neglect interactions involving the @xmath20 photo - excited electron , such as the @xmath59 coulomb interactions .
these include off - diagonal couplings between @xmath21 states which may be important for singlet - triplet energy differences as well as providing pathways for symmetry - allowed @xmath12 excitations .
we also neglect exchange interactions involving the 1s core - hole , which are very small . in what follows ,
we use a standard set of parameters [ in ev ] @xcite : @xmath60 , and @xmath61 .
these parameters give a value for the copper magnetic exchange @xmath62 ev for corner - shared copper oxides , and 2.8 mev for edge - shared .
we first perform full exact - diagonalization of the hamiltonian without the core - hole , and then redo the calculation with the core - hole . since the hamiltonian conserves the total spin , we diagonalized in the s=1/2 sector . from the eigenvalues and eigenvectors of the two steps , the rixs spectrum for a given @xmath13
is computed from eq.([rixs - formula ] ) .
as our calculations involve determining the eigenvector of each state used to construct the matrix elements involved in rixs , we know the character of each excitation exactly and do not need to make a priori assumptions about which peak corresponds to which fundamental excitation .
this allows us to make direct correspondance between rixs and other core - level spectroscopic techniques , without any further assumptions .
this is quite different from the lanczos method , which truncates the hilbert space and focuses on the low - lying energy excitations . whereas this method allows investigation of slightly larger clusters , extensions to higher energies
is problematic particularly for rixs where eigenvectors are needed to evaluate the matrix elements between higher energy states in order to obtain both the resonance profile ( dependence on incident photon energies ) as well as the intensities of prominent peaks in the spectrum . since the core - hole interaction is essentially in nature ( on - site coulomb repulsion ) , leading to only a slight delocalization of the charge , we checked that the cluster - size does not play a crucial role in the qualitative shape of the spectra . in that sense ,
the calculations done in the present paper add more information and provide a more direct access to the relevant rixs excitations compared to other techniques used in previous studies in the context of standard cuprates.@xcite these studies treated the one - band hubbard model where the zhang - rice singlets are , by construction , already formed , and focused only on the momentum dependence of the mott gap excitation . by treating the problem in a multi - band model ,
the formation of zhang - rice excitations comes naturally in the problem and its signature in the rixs intensity , peak dispersion and resonance profile can be directly addressed .
in fact it is widely recognized that the gap is a charge transfer gap which involves transition from oxygen @xmath63 to copper @xmath64 states.@xcite
o@xmath65 cluster . bottom : xps spectrum as a function of photon energy transfer and incident photon energy , for corner and edge - sharing copper - oxide clusters.,width=188 ] in core level x - ray photoelectron spectroscopy ( xps ) , the final state corresponds to an intermediate state in the rixs process .
we first plot the xps spectrum for both edge - shared cluster cu@xmath66o@xmath65 and corner - shared cluster cu@xmath66o@xmath67 in fig .
( [ cluster ] ) . in both cases ,
the prominent features correspond largely to 1 ) a non - locally screened d - intersite zrs @xmath68 feature at 2.2 ev , where the core hole has pushed a @xmath69 hole from the on - site cu onto the oxygen ligands in the neighboring plaquette , 2 ) a local charge transfer ( ct ) excitation @xmath70 at 3.5 ev whereby the screening of the core hole comes from the hole transfer from copper to oxygen on the same plaquette,@xcite and 3 ) a 10 ev high - energy poorly screened @xmath69 cu hole on the core - hole site .
these results are quantitatively similar to those presented earlier , @xcite apart from a more realistic value of the exchange @xmath71 in the corner - shared geometry .
we note that x - ray absorption spectra in an extended system can be obtained from our xps spectrum via a convolution with the copper @xmath20 dos , giving a broad absorption edge followed by spectral intensity extending higher in energy governed by the @xmath20 bandwidth .
these excitations , present in both corner and edge - shared geometries , occur at roughly the same energy since they are governed by charge transfer energies and the exchange @xmath72 between copper and oxygen : the cu - cu exchange does not play a major role in setting these energy scales .
the main difference concerns the prominence of the non - locally screened d - intersite zrs excitation in the corner - shared compared with the edge - shared systems . while the zrs accounts for nearly 80 percent of the ground state wavefunction including the core hole for the corner - shared plaquettes , it is approximately 56 % for the edge - shared system , reflecting the number of oyxgen hybridization pathways available in the edge - shared configuration .
@xcite thus while the energy is comparable in both systems , the matrix elements are much weaker at the zrs scale in the edge - shared compared to corner - shared systems . in addition , the mobility of the zrs for each geometry determines the peak widths , reflecting the splitting between bonding and anti - bonding zrs on sites neighboring the core hole . in the case of the corner - shared systems
, this corresponds to @xmath73 , while due to the orientation of the orbitals in the edge - shared systems , the splitting is @xmath74 ev . as a result , the locally screened @xmath75 peak at 3.1 ev is much larger in edge - shared compared to corner - shared .
this has led the phenomenon of high - temperature superconductivity , occurring in corner - shared compounds and absent in edge - shared compounds , to be associated with the stability of the zrs.@xcite we note that the primary role of the oxygen orientation in the formation of the zrs is not captured in single - band hubbard model calculations .
the aim of this subsection is to discuss the overall shape of the rixs spectrum and resonant profile for a cu@xmath1o@xmath76 cluster in the inset of fig .
[ wave1]a .
we show how to attribute to each peak the corresponding excitation , as an example , we have chosen the same parameters as for xps in fig .
[ cluster ] , appropriate for li@xmath0cuo@xmath0 . from here on we set @xmath77 equal to 1 . here
the intermediate state core - hole is included on the edge of the cluster .
the calculations are also performed for different locations of the core hole on the cluster , which has one hole per copper - plaquette .
while the results differ slightly for different locations of the core hole , we note that the changes are only quantitative and give different broadening of the rixs peaks due to the different mobility of the d - intersite zhang - rice excitation on the end or middle of the cluster .
however , since the aim of this paper is to compare trends for different cluster geometries and charge transfer energies , we will not discuss further these relatively minor changes in the spectra . the rixs spectrum shown in the inset of fig .
[ wave1]a consists of two prominent excitations at approximately @xmath78 ev energy transfer , at incident photon energies of 3.1 and 10 ev .
this is the local ct excitation , resonant at both the locally screened @xmath75 ct and poorly screened @xmath69 intermediate states , as shown in fig .
[ cluster ] .
in addition a weaker d - intersite zrs - derived peak is present at @xmath79 ev , resonant for incident photon energies near 2.2 ev above the @xmath29 transition .
the xps spectra may be crudely viewed as a sum over all raman shifts of the rixs spectrum as a function of the incident photon energy @xmath13 .
o@xmath80 cluster as a function of photon energy transfer @xmath81 and incident photon energy @xmath82 , measured relative to the binding energy @xmath28 .
the inset shows a contour map over large frequencies .
matrix elements for cu@xmath83o@xmath80 are shown in b ) & c ) , for the main peaks in rixs at @xmath84 and @xmath85 , ev respectively .
here the left , right eigenvector is the core hole , final state , respectively , and the dot and the circle represent a singlet state formed by a hole on the copper site ( dot ) and a hole delocalized on the oxygens having the zr symmetry ( circle ) . ] in fig .
[ wave1]a we show a blow up of the low energy rixs spectrum . by examining the symmetries of the core - hole and final state wavefunctions , the nature of the rixs excitations , and the origin of their resonant profile , can be determined by inspection of matrix elements appearing in eq .
[ rixs - formula ] .
the ground state without the core hole @xmath86 is largely of @xmath69 character in each plaquette .
only those intermediate states which have a large overlap with the ground state are relevant to the rixs process .
to determine which final states @xmath87 are probed by rixs and at what resonant energy they occur , one inspects the overlap of the core hole states @xmath88 reachable via the ground state without the core hole @xmath86 , with other excited states @xmath89 in the same hilbert space . for example , since the ground state of the cluster @xmath86 is a singlet , it can not be coupled to triplet intermediate states having a core hole .
we immediately conclude that for our cluster all triplet states do not appear in the rixs process without spin - orbit coupling .
we sketch schematically in fig .
[ wave1]b and [ wave1]c the leading order contributions to the wave functions of the intermediate state with the core - hole and the final state with large matrix overlap , which correspond to the main excitations shown in the rixs spectra .
while the intermediate state is predominantly one configuration , the final states can either be bonding or anti - bonding combinations of excitations on the plaquettes , at energies approximately 2 ev above @xmath90 , split by zrs hopping . in fig .
[ wave1]b , it is clear that the resonance peak at @xmath91 and energy transfer @xmath92 comes from the combination of intermediate and final states corresponding to the creation of the zrs in the final state , having strong overlap with the non - local , well - screened @xmath93 intermediate state .
the two final states of bonding and anti - bonding combinations of zrs on the two plaquettes couple to the intermediate state , with a weak splitting in edge - shared systems . even with our broadening of @xmath94
the separate peaks are difficult to resolve .
the final states of the higher energy peak correspond to bonding and anti - bonding combinations of @xmath75 , having strong overlap with the intermediate core hole state with the ligand on the core hole site as shown in fig .
[ wave1]c .
this lies at higher energy @xmath95 , and is the most prominent final state in xps shown in fig .
[ cluster ] .
we note that we do not find any prominent @xmath12 excitations lying in the relevant energy range below 2 ev , even when including other @xmath21 orbitals.@xcite this follows from a symmetry analysis of the states , whereby @xmath93 and @xmath75 configurations can not hybridize other cu @xmath21 orbitals and combinations of oxygen ligands .
thus direct @xmath12 excitations are inaccessible in our model , in agreement with electron energy loss cluster calculations .
@xcite it is plausible that @xmath12 excitations may arise if symmetry - breaking interactions with the 4p are included .
this is different for oxygen 1s rixs , as the creation of the core hole breaks this symmetry , allowing direct @xmath12 excitations , as determined in recent cluster calculations.@xcite moreover , @xmath12 excitations may arise from multiple couplings in non - resonant scattering@xcite which are not treated in our calculation .
so far , our analysis did not include the copper-@xmath20 states .
however , in order to obtain results more closely connected to experiment and to include a photon polarization dependence for the cu @xmath2-edge process , the cu-@xmath20 dos must be included into the calculations . indeed , the cu-@xmath16 core - electron can only be excited into the @xmath20-levels if there is a finite dos for unoccupied states . due to crystal field effects ,
the 4p dos for different @xmath96 orbitals provide a polarization dependent shift of the resonance energies and secondary satellites governed by the peaks of each @xmath20 projected dos as well as the overall bandwidth .
the orientation of the incident and emitted photon polarizations determine which @xmath20 projected dos is accessed . since the @xmath20-states are rather extended orbitals their interaction with the hubbard subbands is quite weak .
however , depending on the incoming light - polarization , matrix elements are selected so that one can access different linear combinations of @xmath97 , @xmath98 and @xmath99 states .
since the dos distributions are differents , the rixs spectra reflect these characteristics . thus to deduce the influence of the @xmath20-states , we neglect minor real - space variations of the combined cluster-@xmath20 wavefunction and compute the convolution of our raw - rixs spectrum with the calculated @xmath20 projected dos @xmath100 : @xmath101 the projected @xmath20 dos is calculated using linearized augmented plane wave method ( lapw ) in a coordinate system consistent with experiment.@xcite in the following subsection we present results for different edge - shared cuprates li@xmath0cuo@xmath0 and cugeo@xmath83 . for li@xmath0cuo@xmath0 , the @xmath20 projected dos is shown in fig .
[ liconv](a ) .
the @xmath99 dos has two sharp peaks at roughly @xmath102ev , with subdominant peaks as shoulders to the main peak . the resulting rixs spectrum for @xmath103 photon polarizations shown in fig .
[ liconv](b ) thus has a shifted d - intersite zrs resonance at @xmath104ev instead of @xmath105ev in fig .
[ wave1 ] , yet remains sharply peaked even though the @xmath99 bandwidth is over 20 ev wide .
the @xmath75 ct and poorly screened excitation at @xmath106 are likewise shifted to higher resonance frequencies .
the @xmath107 dos in fig .
[ liconv](a ) has a double - peak structure , with a single sharp peak at 4 ev and a tightly structured set of peaks around 10 ev . the resulting rixs spectrum for photon polarizations along the @xmath31-axis , given in fig .
[ liconv](c ) , clearly shows two split resonances at @xmath108ev and @xmath109ev , a shift of the raw rixs spectrum ( fig .
[ wave1 ] ) by the @xmath110ev and 10@xmath111 @xmath107 dos peaks .
the difference in the relative intensities between fig .
[ liconv](b ) and fig .
[ liconv](c ) is due to the fact that the dos for the @xmath112 orbitals is more intense at its maximum than for the @xmath113 orbital
. partial density of states for li@xmath0cuo@xmath0 . * ( b ) * rixs spectrum for parameters corresponding to li@xmath0cuo@xmath0 , convoluted with the @xmath112-dos . * ( c )
* rixs spectrum for parameters corresponding to li@xmath0cuo@xmath0 , convoluted with the @xmath44-dos . for both ( b ) and ( c ) we took a core - hole lifetime of @xmath114 fig .
[ li_comp ] displays a side - by - side comparison of the present theory and recent experimental data by kim _
et al._@xcite on li@xmath0cuo@xmath0 .
the theory and experimental curves show striking similarities in both the energy transfer and the incoming photon energy range . the energy scale , prominence of the local ct excitation , and the double resonance profile shown in figs
. [ cluster],[wave1],[li_comp ] correspond well to the rixs experiments .
we remark that while ref . attributed the weak 2.1 ev peak to an interatomic @xmath12 excitation , and the strong peak at 5.4 ev peak to local ct excitation , in view of our results
we would conclude that the low energy excitation rather corresponds to the d - intersite zrs excitation .
the weak peak observed at 7.6 ev however has no correspondance in our calculations .
this peak may be associated with a local ct excitation into states with different ligand symmetry orthogonal to the zrs ligand states , which may become accessible once oxygen @xmath115 interactions are included .
cuo@xmath0 by kim _
_ in ref .. * ( b ) * calculated rixs spectrum for parameters corresponding to li@xmath0cuo@xmath0 convoluted with the @xmath20-dos . ] as pointed out in ref .
, the large valency of ge@xmath116 nearby the cuo planes increases the madelung energy difference between the cu and planar o sites in cugeo@xmath83 compared to licuo@xmath0 , leading to the values @xmath117 and 3.2 ev , respectively , for each system .
since this energy scale largely determines the energies of the d - intersite zrs and the separation in energy to the local ct excitation , the rixs spectrum and resonance profile should be largely different in the two cases .
in a first approach , as we did for li@xmath0cuo@xmath0 , we present in fig .
[ ge ] a non - convoluted rixs spectrum for parameters corresponding to cugeo@xmath1 , which are the same set of parameters as before apart from the change @xmath118=1.23 ev , and @xmath117 ev . the larger value of @xmath119 leads to several changes compared to the rixs spectrum for li@xmath0cuo@xmath0 parameters ( see fig .
[ wave1].a ) .
first , we note that the d - intersite zrs peak has moved higher in energy from the @xmath69 and core - hole ground states , lying at a photon energy transfer @xmath81 of 3.1 ev , and resonant at a higher incident photon energy @xmath120 ev .
the bonding- anti - bonding splitting is smaller and the resonant peaks are more sharp . in addition , the ct @xmath70 excitations move higher in energy , lying at @xmath121 ev and @xmath122 ev .
this energy cost arises from the higher oxygen site energies as well as the reduction in exchange energy gain with the cu spin .
, as defined in the text . ]
the cugeo@xmath83 structure in fig.[structure ] is different from the one of li@xmath0cuo@xmath0 : the copper chains run now along the c - direction and they are tilted from the main - axis planes .
we present in fig .
[ geconv](a ) the @xmath20 projected dos for cugeo@xmath83 .
the coordinates system for which the calculation has been done corresponds to the one presented in fig.[structure ] .
as one can see , the result is strongly dependent on polarization : the density of @xmath98 is peaked around @xmath123 ev , the @xmath99-density has its stronger weight around @xmath124 ev and is much broader . structure . ]
the corresponding rixs results are shown for photon polarizations along @xmath45 and @xmath125 in fig .
[ geconv](b ) and ( c ) , respectively .
the overall shape of the spectra does not change much compared to the unconvoluted results ( see fig .
[ ge ] ) : one can still identify clearly a low - energy zhang - rice peak and a brighter @xmath75 peak at higher energy as well as the poorly screened @xmath69 excitation peak .
the main difference is a shift on the incident photon energy axis , of @xmath126 ev for @xmath45 and @xmath127 ev for @xmath112 . partial density of states in cugeo@xmath83 calculated in wien2k . *
( b ) * rixs spectrum for parameters corresponding to cugeo@xmath1 , convoluted with the @xmath45-dos . * ( c ) * rixs spectrum for parameters corresponding to cugeo@xmath1 , convoluted with the @xmath112-dos . ] . _ ( right ) _ theoretical spectrum obtained for the same polarization . ] these results are compared to the recent experimental spectrum obtained by hill _
et al._@xcite in fig .
[ water_ge ] and [ ge_exp ] . in this experiment
a ct @xmath70 peak at @xmath128 ev and the zrs peak as well , closer to the elastic line , at about @xmath129 ev , were identified .
these two observed peaks correspond to the two peaks seen in fig .
[ ge ] . in fig .
[ water_ge ] we have used an energy - dependent damping for the lorentzian representing the delta function of eq.([rixs - formula ] ) .
this energy dependence of the line width accounts for the fact that the higher energy states lie in a high density continua . for @xmath130ev , a quadratic energy dependence of the damping rate was taken and assumed to saturate at higher raman shifts .
this form was chosen in order to account for the relative intensity and width of the zr and @xmath75 peaks .
we note that the slight discrepancy between theory and experiment in the position and intensity of the d - intersite zrs may be remedied by fine - tuning the cluster parameters , to more accurately reflect the material properties ( such as @xmath71 and the effective zr - hopping @xmath131 ) , as well as incorporating a more realistic resolution broadening parameter .
another difference between theory and experiment is the relative intensity of the peaks for different polarizations : while the peaks are more intense for a polarization along @xmath132 in the experiment , it is the opposite for the simulation .
this issue may be related to the two following points : ( i ) from a theoretical point of view the relative intensity of the peaks is sensitively linked to the relative peaks height of the dos ( since in eq.(1 ) it appears to be squared ) , thus a slight variation in the dft calculation might make a difference .
( ii ) due to experimental uncertainties a direct comparison of the measured absolute intensities for @xmath133 and @xmath134 is not possible .
however , for a fixed polarization the measured relative intensities of the zr and the ct feature can be compared . with polarization dependence .
_ ( right ) _ exact - diagonalization results obtained after convolution with the @xmath20 projected dos of fig.[geconv].(a ) ( see subsection [ convo ] ) ] finally we remark that the crystal structures of cugeo@xmath83 and li@xmath0cuo@xmath0 are strongly different ( see for instance refs . ) . in both cases
the cuo@xmath0 chains run along a principal axis of the crystal but in cugeo@xmath83 they are not oriented along the same plane , whereas for li@xmath0cuo@xmath0 they are .
this has strong implications concerning the interpretation of the experimental results : for cugeo@xmath83 the polarization - dependent experiments will give rixs spectra which are a result of convolution with the projected dos in a rotated basis .
the result of such a convolution is given in the right panel of fig.[ge_exp ] .
a full exact - diagonalization treatment of multi - band hubbard clusters has been used to calculate the polarization , incident photon energy , and transfered photon energy dependencies , of rixs in one - dimensional , edge - shared insulating cuprates . by building upon previous work ,
@xcite the rixs spectrum can be constructed using xps final - states as the intermediate states of inelastic x - ray scattering .
the intensity and resonance profile , as well as the identity and character of the main rixs peaks has been obtained using the eigenstates and eigenenergies of clusters with and without the core - hole .
these peaks correspond to non - locally screened d - intersite zrs @xmath93 , local ct @xmath75 , and poorly screened @xmath69 , each having intensities and resonance profile determined by the overlap of these states with the intermediate core - hole states .
the polarization dependent rixs spectra were obtained by convolving the raw rixs spectra with the copper @xmath20 projected dos determined by wien2k density functional calculations .
while the resulting rixs spectra includes the @xmath20 states as merely spectators to the re - arrangement of copper - oxygen valence charge density , a polarization and material dependent rixs emerges due to the peak structures and intensities of the @xmath20 orbitally - projected intermediate states . direct comparison ( see figs . [ li_comp],[ge_exp ] ) with recent data for two edge - shared cuprates , li@xmath0cuo@xmath0 and cugeo@xmath83,@xcite shows both qualitative and , more importantly , quantitative agreements between theory and experiment for the shape , energy , width , and resonance profile of rixs spectral features .
this indicates that the local cluster model can adequately reproduce the data for photon energy losses in the energy range of 2 - 7 ev .
we again remark that @xmath12 excitations are not captured in the model , and further , that spin - flip excitations , which require large clusters to observe bi - magnon spin rearrangements at long wavelengths , are also very weak .
better agreement of the calculated spectra with experiment may be obtained by using more refined cluster parameters , multi - pole couplings , and more accurate representation of resolution broadening .
a polarization dependent rixs experimental study , for li@xmath0cuo@xmath0 , should find significant differences in the resonant profiles for in - plane polarization and polarization perpendicular to the cuo@xmath0-plane .
[ liconv ] indicates that peak intensities are brighter and their resonance energies are shifted by @xmath135 ev for in - plane compared to perpendicular polarizations .
these predictions as well as others , like multiple resonant features resulting from @xmath20-projected states , open a way to infer from cu @xmath11edge and other indirect rixs experiments the unoccupied dos . in this way rixs can be seen as a complementary tool to core - level spectroscopies such as xps and xas , providing at the same time diverse and important spectral information on elementary excitations in strongly correlated materials . in summary
, we have explored the cu @xmath11edge spectrum for different edge - shared copper oxide systems , and have shown that the energy scales , peak intensities , and resonance profiles can be explained qualitatively within the context of exact diagonalization studies of the multi - band hubbard model .
the prospect of mapping out the full momentum , and resonance profile dependences would allow for a direct spectral check of the many roles played by local electronic correlations in the multi - band hubbard approach .
this remains a topic of future research .
the authors wish to thank j. p. hill , y .-
kim , z .- x .
shen , z. hussain , m. z. hasan , m. greven , j. hancock , and m. gingras for many useful discussions .
tpd and gas would like to acknowledge support of this work in part by nserc , cfi , onr grant n00014 - 05 - 1 - 0127 ( tpd ) , prea ( tpd ) , the alexander von humboldt foundation ( tpd ) and cifar ( gas ) .
jg gratefully acknowledges the financial support by the dfg .
tpd wishes to thank the pacific institute for theoretical physics for their hospitality .
fv would like to thank financial support from cifar .
99 a. damascelli , z. hussain , and z .- x .
shen , rev .
phys . * 75 * , 473 ( 2003 ) .
j. graf , g .- h .
gweon , a. lanzara , cond - mat/0610313 ; j. graf , g .- h .
gweon , k. mcelroy , s. y. zhou , c. jozwiak , e. rotenberg , a. bill , t. sasagawa , h. eisaki , s. uchida , h. takagi , d .- h .
lee , a. lanzara , phys .
* 98 * , 067004 ( 2007 ) ; t. valla , t. e. kidd , z .- h . pan , a. v. fedorov , w .- g .
yin , g.d .
johnson , cond - mat/0610249 ; b. p. xie , k .
yang , d. w. shen , j. f. zhao , h. w. ou , j. wei , s. y. gu , m. arita , s. qiao , h. namatame , m. taniguchi , n. kaneko , h. eisaki , z. q. yang , d.l .
feng , phys .
98 * , 147001 ( 2007 ) ; w. meevasana , x.j .
zhou , s. sahrakorpi , w.s .
lee , w.l .
yang , k. tanaka , n. mannella , t. yoshida , d. h. lu , y.l .
chen , r.h .
he , hsin lin , s. komiya , y. ando , f. zhou , w.x .
xiong , z. x. zhao , t. sasagawa , t. kakeshita , k. fujita , s. uchida , h. eisaki , a. fujimori , z. hussain , r. s. markiewicz , a. bansil , n. nagaosa , j. zaanen , t.p .
devereaux , z .- x .
shen , phys .
b * 75 * , 174506 ( 2007 ) t. p. devereaux and r. hackl , rev . mod
. phys . * 79 * , 175 ( 2007 ) .
p. m. platzmann and e. d. isaacs , phys .
b * 57 * , 11107 ( 1998 ) ; a. kotani and shin , rev .
phys . * 73 * , 203 ( 2001 ) .
f. c. zhang and t. m. rice , phys .
b * 37 * , 3759 ( 1988 ) ; h. eskes and g. a. sawatzky , phys . rev
. lett . * 61 * , 1415 ( 1988 ) ; h. eskes , l. h. tjeng , and g. a. sawatzky , phys .
b * 41 * , 288 ( 1990 ) ; h. eskes and g. a. sawatzky , phys . rev .
b * 44 * , 9656 ( 1991 ) .
y. mizuno , t. tohyama , s. maekawa , t. osafune , n. motoyama , h. eisaki , and s. uchida , phys .
b * 57 * , 5326 ( 1998 ) .
j. b. goodenough , phys . rev . * 100 * , 564 ( 1955 ) ; j. kanamori , j. phys . chem . solids * 10 * , 87 ( 1959 ) ; p. w. anderson , solid state phys . * 14 * , 99 ( 1963 ) .
young - june kim , j. p. hill , f. c. chou , d. casa , t. gog , and c. t. venkataraman , phys .
b * 69 * , 155105 ( 2004 ) .
d. qian , y. li , m.z .
hasan , d.m .
casa , t. gog , y .- d .
chuang , k. tsutsui , t. tohyama , s. maekawa , h. eisaki , and s. uchida , j. phys .
solids * 66 * 2212 , ( 2005 ) .
m. z. hasan , y .- d .
chuang , y. li , p. a. montano , z. hussain , g. dhalenne , a. revcolevschi , h. eisaki , n. motoyama , and s. uchida , int .
b * 17 * , 3519 ( 2003 ) .
s. suga , s. imada , a. higashiya , a. shigemoto , s. kasai , m. sing , h. fujiwara , a. sekiyama , a. yamasaki , c. kim , t. nomura , j. igarashi , m. yabashi , and t. ishikawa , phys .
b * 72 * , 081101 ( 2005 ) . m. v. zimmermann , j. p. hill , c .- c .
kao , t. ruf , g.a .
sawatzky , t. gog , c. venkataraman , t. masuda , i. tsukada , and k. uchinokura , preprint .
duda , j. downes , c. mcguiness , t. schmitt , a. augustsson , k. e. smith , g. dhalenne , and a. revcolevschi , phys . rev .
b * 61 * , 4186 ( 2000 ) kenji tsutsui , takami tohyama , and sadamichi maekawa , phys .
lett . * 91 * , 117001 ( 2003 ) ; k. ishii , k. tsutsui , y. endoh , t. tohyama , s. maekawa , m. hoesch , k. kuzushita , m. tsubota , t. inami , j. mizuki , y. murakami , and k. yamada , phys .
* 94 * , 207003 ( 2005 ) j. zaanen , g. a. sawatzky , and j. w. allen , phys .
lett . * 55 * , 418 ( 1985 ) m. a. van veenendaal and g. a. sawatzky , phys .
lett . * 70 * , 2459 ( 1993 ) m. a. van veenendaal , h. eskes , and g. a. sawatzky , phys . rev .
b * 47 * , 11462 ( 1993 ) ; m. van veenendaal * 74 * , 085118 ( 2006 ) ; k. okada and a. kotani , j. phys . soc .
japan , * 75 * , 044702 ( 2006 ) . due to the one - dimension geometry of the clusters , especially for the corner - shared case ,
the @xmath93 is not truely a zrs as it has more weight on the central , bridging oxygen .
however , the symmetry of this state remains zr - like , maintaining the zr signs of the ligands .
y. ohta , t. tohyama , and s. maekawa , phys .
b * 43 * , 2968 ( 1991 ) ; c. di castro , l. f. feiner , and m. grilli , phys . rev . lett . *
66 * , 3209 ( 1991 ) ; j. h. jefferson , h. eskes , and l. f. feiner , phys . rev .
b * 45 * , 7959 ( 1992 ) ; l. f. feiner , m. grilli , and c. di castro , phys .
b * 45 * , 10647 ( 1992 ) ; r. raimondi , j. h. jefferson , and l. f. feiner , phys .
b * 53 * , 8774 ( 1996 ) . c. de graaf and r. broer , phys . rev .
b * 62 * , 702 ( 2000 ) .
s. atzkern , m. knupfer , m. s. golden , j. fink , a. hubsch , c. waidacher , k. w. becker , m. weiden and c. geibel , phys . rev .
b * 64 * , 075112 ( 2001 ) .
k. okada and a. kotani , phys .
b * 63 * , 045103 ( 2001 )
. m. w. haverkort , a. tanaka , l. h. tjeng , g. a. sawatzky , cond - mat / arxiv:0705.4637 ( 2007 ) .
p. blaha , k. schwarz , g.k.h .
madsen , d. kvasnicka , and j. luitz , _
wien2k _ , _ an augmented plane wave + local orbitals program for calculating crystal properties_,isbn 3 - 9501031 - 1 - 2 ( tu wien , austria , 2001 )
. j. hill _
et al . _ , preprint m. braden , g. wilkendorf , j. lorenzana , m. an , g. j. mcintyre , m. behruzi , g. heger , g. dhalenne , and a. revcolevschi , phys . rev .
b * 54 * , 1105 ( 1996 ) f. sapia , j. rodriguez - carvajal , m. j. sanchis , r. ibez , a. beltrn , d. beltrn , solid state comm . *
74 * , 779 - 784 ( 1990 ) | we present calculations for resonant inelastic x - ray scattering ( rixs ) in edge - shared copper oxide systems , such as li@xmath0cuo@xmath0 and cugeo@xmath1 , appropriate for hard x - ray scattering such as the copper @xmath2-edge .
we perform exact diagonalizations of the multi - band hubbard model and determine the energies , orbital character and resonance profiles of excitations which can be probed via rixs .
we find excellent agreement with recent results on li@xmath3cuo@xmath3 and cugeo@xmath1 in the 2 - 7 ev photon energy loss range . |
the @xmath3 process is a rare decay , which has been recently studied by several authors in context of standard chiral perturbation theory ( s@xmath4pt ) , namely at the lowest order by knchlein , scherer and drechsel @xcite and to next - to - leading by bellucci and isidori @xcite and ametller et al . @xcite .
the experimental interest for such a process comes from the anticipation of large number of @xmath0 s to be produced at various facilities.ne about @xmath5 decays per year ] the goal of our computations is to add the result for the next - to - leading order in generalized chiral perturbation theory ( g@xmath4pt ) .
the motivation is that one of the important contributions involve the @xmath6 off - shell vertex which is very sensitive to the violation of the standard scheme and thus this decay provides a possibility of its eventual observation .
we have completed the calculations at the tree level , added 1pi one loop corrections , corrections to the @xmath6 vertex and phenomenological corrections to the resonant contribution .
these preliminary results we would like to present in this paper .
the amplitude of the process can be defined @xmath7 in the square of the amplitude summed over the polarizations @xmath8 we integrated out all of the independent lorentz invariants except the diphoton energy square @xmath9 our goal is to calculate the partial decay width @xmath10 of the @xmath0 particle as the function of the diphoton energy square @xmath11 . at the lowest order , the s@xmath4pt does not depend on any unknown free order parameters .
in contrast , there are two free parameters controlling the violation of the standard picture in the generalized scheme .
we have chosen them as @xmath12 and their ranges are @xmath13 .
we use abbreviations for @xmath14 , @xmath15 and @xmath16 .
the standard values of these parameters are @xmath17 and @xmath18 .
at the @xmath19 tree level , the amplitude has two contributions , with a pion and an eta propagator .
the first one is resonant , ` @xmath20-pole ' , the other is not , ` @xmath0-tail ' .
the standard values of the contributions to the partial decay rate and the maximum possible violation of the standard counting ( @xmath21 ) are represented in fig .
[ graph4 ] .
the pole of the resonant contribution at @xmath22 is transparent . while in the standard case it is fully dominant , in the generalized scheme the @xmath0-tail could be determining in the whole area @xmath23 .
the reason can be found in the @xmath6 vertex .
its contribution in the generalized amplitude can jump up to 16 times its standard value . the full decay width for the standard ( @xmath24 ) and generalized case ( @xmath17,@xmath25 and @xmath26,@xmath27 ) is displayed in fig .
[ graph6 ] .
it can be seen , that even in the conservative intermediate case the change is quite interesting .
there are four distinct contributions at the next - to - leading order : one loop corrections to the @xmath28-pole and the @xmath0-tail , one particle irreducible diagrams ( 1pi ) and counterterms . in the latter case we rely upon the results of @xcite .
their estimate from vector meson dominated counterterms indicates , that it causes only a slight decrease of the full decay width .
because the estimate is the same for both schemes , for our purpose of studying the differences between them we can leave it for later investigation . more important are the corrections to the @xmath0-tail diagram .
we did take into account the corrections to the @xmath6 vertex .
these involve loop corrections and counterterms with many unknown higher order parameters . as a first approximation
, we set these parameters equal to zero and estimated their effect through the remaining dependence on the renormalization scale .
the scale was moved in the range from the mass of the @xmath0 to the mass of @xmath29-meson .
we decided , similarly to @xcite , to correct the @xmath20-pole amplitude by a phenomenological parametrization of the @xmath30 vertex and fix the parameters from experimental @xmath30 data .
we made an estimate of its phase by expanding the @xmath30 one loop amplitude around the center of the dalitz plot . in the 1pi amplitude , we neglected the suppressed kaon loops .
[ graph2 ] represents the one loop corrected decay widths for the standard and the maximum violation of the standard scheme .
the dependence on the renormalization scale is used to estimate the uncertainty in the unknown higher order coupling constants .
we can see that the scale dependence is small in the standard counting and not too terrible in the generalized variant . in the case of the maximum violation of s@xmath4pt ,
the difference is big enough to not to be washed out by the uncertainty .
however , in the conservative case @xmath17,@xmath25 this is not true and the promising results from the tree level are lost .
we have analyzed the @xmath31 decay to the next - to - leading order of chiral perturbation theory in its both variants .
the tree level results are promising , the sensitivity to the change in parameters controlling the violation of the stndard @xmath4pt is considerable .
at the one loop level , we tried to estimate the uncertainty in the higher order couplings constants in the crucial @xmath6 vertex through their dependence on the renormalization scale .
although for big violation of the standard case the difference is preserved , for the more realistic conservative case the output is not satisfactory .
we would like to stress that these results are preliminary and there are several ways how to deal with the unknown order parameters .
one of them is to take into account the vector mesons , similarly to the counterterm estimate in @xcite .
other way is to treat the whole @xmath4pt expansion differently , with more caution , as developed in @xcite .
this approach , called ` resumed ' @xmath4pt could provide results similar to the tree level case even if the one loop corrections are involved . | calculations of @xmath0 @xmath1 @xmath2 decay in generalized chiral perturbation theory are presented . tree level and next - to - leading corrections are involved . sensitivity to violation of the standard counting
is discussed . + marin kolesr , ji novotn + _ charles university , faculty of mathematics and physics , v holeovikch 2 , 18000 praha 8 , czech republic _ |
the observational evidence have proved beyond doubt that the present universe is accelerating @xcite .
the exotic matter , the dark energy , causing the accelerated expansion of the universe is assumed to have a negative pressure .
various models were proposed aimed at understanding the properties of dark energy . among these , the cosmological
constant is the prominent one , but it faces the problems of fine tuning and also unable to explain the coincidence of dark energy density and dark matter density .
this leads to consideration of dynamic dark energy models .
among which holographic dark energy model @xcite based on the holographic principle is one of the prominent models discussing in the recent literature . according to the holographic principle @xcite
, the vacuum energy density is bounded.the significance of the principle lies in the constraint that the total energy inside a region of size l , should not exceed the mass of a black hole of the same size . from effective quantum field theory ,
an effective ir cut off can saturate the length scale , so that the dark energy density ( the vacuum energy density ) can be written as @xmath2 where c is a dimensionless numerical factor.the possible choices for the ir cut - off are the hubble horizon distance , particle horizon distance , event horizon distance and some generalized ir cut - off @xcite .
but the first two options could not support the accelerated expansion of the universe while the event horizon posed the problem of causality.thus a new model has been introduced by granda and oliveros @xciteby taking the ricci scalar as the ir cut - off , and later studied by many others @xcite .
this is seemed to be efficient in solving the coincidence problem , causality problem , and also the fine tuning problem .
a modified form of this model was studied in interaction with dark matter assuming a hidden non - gravitational coupling existing between them @xcite . in this paper
we will explore the thermodynamics , especially the status of the generalized second law of thermodynamics in a flat universe with holographic dark energy with ricci scalar as the ir cut - off , interacting with the dark matter .
thermodynamics of cosmological models is of great interest after the discovery of the black hole thermodynamics @xcite by bekenstein and hawking @xcite which says that the entropy of the black hole is proportional to the area of it s event horizon @xcite . through the hawking radiation
, the black hole could lose it s thermal energy which results in the decrease of its entropy and this is not allowed by the second law if black hole is assumed as an isolated system .
it is explained as even though the entropy of the black hole decreases the loss is compensated by the increase in the entropy of the matter outside the black hole boundary , eventually the total entropy of the black hole plus that of the surrounding is increasing which is termed as the generalized second law ( gsl ) @xcite.the black hole thermodynamics is applied in the cosmological holographic model by l.susskind @xciteand t hooft @xcite which says that the maximum entropy of a bounded region is proportional to its area rather than its volume .
gibbons and hawking @xcite studied the de - sitter model of the universe where apparent horizon and the cosmological event horizon are coincide with each other and found that the gsl is satisfied . assuming the whole universe as a thermodynamic system in which the validity of the first law of thermodynamics is undoubtedly assured , the present paper quests for the validity of the generalized second law of thermodynamics(gsl ) , in a universe filled with holographic dark energy interacting with the dark matter present .
the first law of thermodynamics and gsl is found to be satisfied for the apparent horizon while violated in the case of event horizon for dark energy with constant equation of state and in a universe with generalized chaplygin gas @xcite .
the event horizon describes a global concept of space - time , hence a difficulty is said to appear in understanding hawking radiation clearly @xcite.thus there are arguments in favor of apparent horizon as the real physical boundary @xcite .
hence we consider whether the gsl is satisfied by the apparent horizon and event horizon in the realm of holographic dark energy model , in order to discern the real physical boundary of the universe from the point of view of thermodynamics .
the friedmann equations for flat universe with frw metric can be written as @xmath3 where h is the hubble parameter , @xmath4 is the energy density of the dark matter and @xmath5 is that of the holographic dark energy .
the expression for the holographic dark energy with ricci scalar as the ir cut - off is @xmath6 where @xmath7 represent the derivative of h with respect to the cosmic time t and @xmath8 where @xmath9 and @xmath10 are parameters of the model .
the interaction between the dark energy and dark matter is studied through the conservation equations given below @xmath11 @xmath12 where @xmath13 is the pressure of the dark energy , @xmath14 is the interaction term , dot represents the derivative with respect to time .
dark matter is assumed to be pressure less .
the term @xmath14 can have mainly three forms , @xmath15 , @xmath16 and @xmath17 @xcite .
for the present study we have considered the form , @xmath18 . substituting , the dark energy density given by equation ( [ eqn : mhrde ] ) in the friedmann equation([eqn : friedmann1 ] ) , and changing the variable cosmic time t to the variable @xmath19 and differentiating the equation once more leads to the second order differential equation , @xmath20 where @xmath21 , @xmath22 is the present value of hubble parameter .
the general solution for this can be obtained as @xmath23 where the constants are evaluated using the initial conditions , @xmath24 and @xmath25 , are obtained as @xmath26 and @xmath27 are obtained as @xmath28 using the friedmann equation , the dark energy density parameter can be obtained as @xmath29 the pressure of the dark energy and its equation of state parameter are calculated as , @xmath30.\ ] ] and @xmath31 the evolution of the equation of state of the interacting holographic dark energy is plotted against redshift @xmath32 for the model parameters , @xmath33 and is shown in figure([fig : eos ] ) .
the figure shows that in the past stage of the universe the equation of state was nearly zero .
as the universe evolves the equation of state decreases and stabilizes as @xmath34 ] in a dark energy dominated universe , @xmath35 and @xmath36 , the coefficients in the above equation become @xmath37 and @xmath38 then the equation of state parameter become @xmath39 , which shows that for positive values of @xmath10 the equation of state @xmath40 corresponds to quintessence type behavior and for @xmath41 the equation of state @xmath42 corresponds to phantom behavior .
the deceleration parameter @xmath43 for the dark energy considered is obtained as @xmath44 the nature of deceleration parameter is studied by plotting @xmath43 ] against the redshift @xmath32 and the result is shown in the figure [ fig : q ] .
it is evident from the figure that during the initial times in the evolution of the universe the @xmath43 parameter was positive implies deceleration in the expansion .
as the universe expands the @xmath43 parameter enters the negative value region in the recent past , implying that the current acceleration was started in the recent past .
it is also seen that the @xmath43 parameter saturates as @xmath34 for dark energy dominated universe with @xmath36 , the deceleration parameter become @xmath45 this implies that @xmath46 corresponds to quintessence type behavior , @xmath47 and @xmath41 corresponds to phantom behavior , @xmath48
in this section the validity of the generalized second law of thermodynamics of the dark energy under thermal equilibrium is studied .
the validity of the gsl implies that the sum of the change in entropy of the dark sectors added with the entropy change of the cosmological horizon is greater than or equal to zero @xcite .
that is , @xmath49 where @xmath50 denotes the change in entropy of dark sectors with respect to cosmic time and @xmath51 is the change in horizon entropy with respect to cosmic time .
the study is done by considering that the dark sectors and the horizon were in thermal equilibrium so that temperature of the dark sectors is equal to the temperature of the horizon . in the following
we will analyse separately the gsl for universe with apparent horizon as boundary and event horizon as the boundary . here
the validity of the gsl is that sum of entropy of the dark energy , dark matter and the apparent horizon must increase with time @xcite .
apparent horizon do exist for all kinds of frw universes @xcite .
the apparent horizon is a causal horizon for the dynamical space time and is associated with gravitational entropy and surface gravity @xcite . for a universe with frw metric the apparent horizon radius can be obtained as @xmath52 for flat frw universe ( @xmath53 ) , the apparent horizon become @xmath54 which is same as the radius of the hubble horizon .
the temperature on the apparent horizon can be defined as @xmath55 , where @xmath56 is the surface gravity @xcite . for flat universe
the surface gravity depends on the horizon radius , and simple considerations will leads to the temperature as @xmath57 the entropy of the apparent horizon is @xmath58 with @xmath59 @xcite , where the area @xmath60 thus the entropy of the apparent horizon can be obtained as @xcite @xmath61 the entropy of the dark sectors can be found using gibbs equation given below @xmath62 where @xmath63 is the volume occupied by the dark entities and @xmath64 and the pressure @xmath65 has contribution only from the dark energy since dark matter is non - relativistic hence pressureless . thus using the above equations and the conservation equation of the dark sectors given in the previous section , the total entropy variation with respect to @xmath19 is obtained as @xmath66 where @xmath67 is the equation of state parameter of the dark energy ,
@xmath68 is the dark energy density parameter and @xmath43 is the deceleration parameter .
the gsl is valid if @xmath69 since @xmath70 , the validity of the gsl requires that , @xmath71.$ ] this condition can approximately be translated as @xmath72 where @xmath73 ( a comparatively smaller value ) . in the present case where @xmath74 is comparatively small hence @xmath75 this is a reasonable bound for @xmath76 in this context and will lead to the increase in the change of entropy hence the validity of the gsl .
the equation ( [ eqn : s1 ] ) can be further simplified using the relation for the @xmath43 parameter , @xmath77 the equation ( [ eqn : s1 ] ) can be re - written as for the parameters @xmath33 thick continuous line , @xmath78 thin continuous line,@xmath79 dashed line and with the interaction coupling constant b=0.001 inside apparent horizon under thermal equilibrium conditions ] for the parameters @xmath80 thick continuous line , @xmath81 thin continuous line,@xmath82 dashed line and with the interaction coupling constant b=0.001 inside apparent horizon under thermal equilibrium conditions ] @xmath83 this equation shows that , as far as @xmath84 the gsl is satisfied for almost all kinds of dark energy , irrespective of the sign of @xmath43 which is positive for decelerated expansion ( radiation or matter dominated era ) and negative for accelerated expansion of the universe .
deceleration parameter @xmath85 ( corresponds to @xmath40 ) in the quintessence phase and @xmath86 ( corresponds to @xmath42 ) in the phantom phase of expansion .
so both in the quintessence phase of expansion and in the phantom phase the gsl is satisfied with apparent horizon as the boundary of the universe .
current cosmological observations hint that @xmath67 may be as lower as -1.5 @xcite . in the dark energy model we have considered equation ( [ eqn : qpar ] )
implies that , in dark energy dominated case the universe is in the quintessence phase when @xmath46 and when @xmath87 the universe enter the phantom phase of expansion as @xmath34 equation ( [ eqn : sprime1 ] ) shows that the gsl is satisfied for both positive and negative values of @xmath10 parameter .
we have studied the evolution of @xmath88 with @xmath89 using the equation of @xmath43 derived in the last section to check the above conclusions .
the plot [ fig : svar1 ] showing the behavior of the @xmath88 using the equation ( [ eqn : s1 ] ) for various choices of the model parameters @xmath90 and it shows that the gsl is satisfied always in conformation with equation ( [ eqn : sprime1 ] ) . in this section we analyse the gsl with event horizon as the boundary under thermal equilibrium condition .
the event horizon distance can be evaluated using the standard relation , @xmath91 on integration the event horizon is turn out to be @xmath92\ ] ] for positive @xmath10 values and @xmath93\ ] ] for negative values of @xmath94 the functions @xmath95 $ ] are the hypergeometric functions .
the constants @xmath96 are different for different values of @xmath97 and the suffix @xmath98 refers to positive ( negative ) values of @xmath94 the values of the various constants is found to be @xmath99 for @xmath100 respectively and @xmath101 for all the parameter sets . for negative values of @xmath10 the constants are found to be @xmath102 for @xmath103 respectively and @xmath104 in this case the temperature is @xmath105 and area of the horizon is @xmath106 using the gibb s relation the rate of change of entropy of dark energy plus matter is @xmath107 for the parameters @xmath33 thick continuous line , @xmath78 thin continuous line,@xmath108 dashed line with the interaction coupling constant b=0.001 inside event horizon under thermal equilibrium conditions ] for the parameters @xmath80 thick continuous line , @xmath81 thin continuous line,@xmath109 dashed line , with the interaction coupling constant @xmath110 inside event horizon under thermal equilibrium conditions ] substituting the temperature @xmath111 and using the relation @xmath112@xcite the above equation become , @xmath113 considering that the fluid inside the horizon satisfies the dominant energy condition @xmath114 @xcite , the above equation shows that the rate of change of entropy of dark energy plus dark matter within the event horizon decreases as far as @xmath115 adding the rate of change of horizon entropy to the above equation we get the rate of change of the total entropy as @xmath116\ ] ] the above equation can be modified by using @xmath117 we get @xmath118.\ ] ] for @xmath119 and @xmath120 the gsl is satisfied i.e. @xmath121 if @xmath122 this can integrated to , if one avoid the constant of integration , as @xmath123 the exact condition for the validity of gsl become @xmath124 this shows that the gsl is valid only if the universe satisfies the dominant energy condition @xmath125 @xcite in the case of increasing horizon radius , i.e. in the quintessence phase . using the friedmann equation and conservation equation a more exact relation for the validity of gsl from the condition @xmath126
can be obtained as @xmath127 in a de sitter universe in which the apparent horizon and event horizons coincide such that @xmath128 , the equation of state parameter @xmath129 and @xmath130 the above condition reduces to @xmath123 using the relation for @xmath43 in terms of equation of state and dark energy mass parameter , the above equation can be translated in to the form @xmath131 for de sitter @xmath132 , so the above condition critically satisfied with @xmath133 and the gsl is satisfied . in such a universe
the equation of state @xmath134 for the holographic dark energy described in section 2 , it was found that for positive values of @xmath10 the equation of state @xmath135 and for -ve values of @xmath10 the equation of state , @xmath136 we made a numerical analysis on the evolution of the condition ( [ eqn : horcond1 ] ) by plotting @xmath137 with the redshift @xmath138 the plots shows that for positive values of @xmath10 parameter ( see figure [ fig : ehpob ] ) the gsl is partially satisfied at the event horizon on the other hand for negative values ( see figure [ fig : ehneb ] ) the gsl is completely violated at the event horizon .
for the further analysis we consider the universe with apparent horizon as the boundary .
a complete thermal non - equilibrium condition means that there is no interaction between the dark sectors hence they have different temperatures that is,@xmath139 where @xmath140 is the temperature of the dark energy , @xmath141 is the temperature of the dark matter , and @xmath142 is the temperature of the horizon . using the gibb s relation the rate of change of the total entropy comprising that of entropy of the dark energy , dark matter and horizon
, can be written as , @xmath143\ ] ] which can be simplified using the basic definition of @xmath43 and friedmann equations to @xmath144.\ ] ] as a first case we will consider the validity of the gsl in which the dark sectors were in equilibrium , @xmath145 then the second term in the above equation will vanish .
it is clear from the remaining equation that the validity of the gsl demands that @xmath146 provided @xmath147 this implies that for an accelerating universe with @xmath148 the dark energy temperature is greater than horizon temperature , @xmath149 for the special case @xmath133 , corresponds to de sitter type universe in which @xmath150 , where @xmath151 is the cosmological constant , the change in entropy is zero as a result the total entropy of the universe remains constant . for @xmath152 ,
the gsl is satisfied provided @xmath153 the condition @xmath152 implies that the universe is in the phantom phase @xcite and for such phases of expansion the above equation shows that the dark energy temperature is less than the dark matter temperature . for @xmath154=1.25 under thermal non - equilibrium condition ] as a further case let us now avoid the dark matter contribution to the total entropy of the universe .
the equation ( [ eqn : sfull1 ] ) now become , @xmath155 from this relation it can be shown that the gsl is satisfied if @xmath156 for @xmath154=1.25 under thermal non - equilibrium condition ] in an accelerated expanding universe , i.e. @xmath157 and if the expansion is such that the universe will not enter the phantom behavior so that @xmath158 and @xmath159 then the above ratio implies that @xmath160 that is the dark energy temperature is greater than that of the horizon . in the cases of @xmath161 and @xmath152 corresponds to the phantom phase also the @xmath162 so in a dark energy dominated universe the temperature of the dark energy is greater than the temperature of the dark matter for the gsl to be valid .
let us now assume that the temperature of the dark energy , @xmath163 especially assume that @xmath164 with @xmath165 then the entropy of the dark energy is @xmath166 where we have used the standard relation for entropy @xmath167 @xcite .
the evolution of @xmath168 with @xmath89 is shown in figure [ fig : sde - x.pdf ] and it is clear that the entropy of the dark energy is decreasing as the universe expands .
the total entropy of the universe , which can be obtained by adding the horizon entropy to the above equation , we get ( avoiding the matter contribution ) , @xmath169 the evolution of this total entropy is plotted in figure [ fig : entropy2 ] , shows that the total entropy is increasing as the universe expands , and thus satisfying the gsl .
this shows that the decrease in the entropy of the dark energy is compensated by the increase in the horizon entropy .
this confirms our previous analyses that if the temperature of the dark energy is greater than the horizon temperature , the total entropy of the universe will always increases which guarantees the validity of the gsl .
in this paper we have analysed the gsl in a flat universe with holographic dark energy and non - relativistic dark matter .
the analysis were done both under equilibrium and non - equilibrium conditions . in equilibrium conditions
the following are our main conclusions . in the case of taking apparent horizon as the boundary of the universe
, we have obtained that rate of change of total entropy is proportional to @xmath0 hence the gsl is infact valid for any kind of dark energy .
we have verified it with holographic dark energy with ricci scalar as the ir cut - off .
so it can be concluded that the apparent horizon is a perfect thermodynamic boundary .
we have also analysed the status of the gsl with event horizon as the boundary and found that in the presence of holographic dark energy the gsl is only partially satisfied for positive values of the model parameter @xmath10 but completely unsatisfied for negative values of the parameter . in literature
there are studies regarding the status of the gsl at the apparent horizon . by considering viscous dark energy in non - flat universe ,
it was found in reference @xcite that the gsl is valid at the apparent horizon . in reference
@xcite , the authors analyzed the validity of the gsl at the apparent horizon with logarithmic corrected entropy relation for the horizon and showed that the gsl is satisfied at the boundary if the model parameter @xmath170 and also holds for @xmath171 if @xmath172 so the validity of the gsl at the apparent horizon seems to be a general fact as advocated by our analysis .
there are other studies in the literature regarding the status of the gsl at the event horizon .
for example in references @xcite , the authors argued that the gsl is only partially satisfied at the event horizon of the universe .
we have extended our analysis of the validity of the gsl with thermal non - equilibrium conditions .
if there is partial non - equilibrium , such that the temperature of the dark sectors are equal and is different from the horizon temperature , the validity of the gsl demands that the temperature of the dark sector is greater than the horizon temperature if the expansion is the quintessence phase . on the other hand
if the universe is in the phantom phase of expansion the dark sector temperature is less than the horizon temperature .
but in a dark energy dominated universe ( avoiding the dark matter contribution ) , equivalent to be a condition of full non - equilibrium where all the components have different temperatures , the temperature of the dark energy is greater than the dark matter temperature both in quintessence and phantom phases of expansion
. there are attempts to study the thermodynamics of the horizon under non - equilibrium conditions .
for example in reference @xcite , the authors have anlaysed the conditions for the validity of the gsl under non - equilibrium condition , however no specific conclusions or criterion were obtained for the validity of the gsl .
so it seems difficult to analyses the status of gsl under non - equilibrium conditions .
however we carried out the analysis by assuming the validity of the gsl at the apparent horizon , and have shown that the gsl is valid if the dark energy temperature is greater than the horizon temperature .
99 perlmutter s 1999 _ astrophys . j. _ * 517 * 565 .
shariff m and abdul jawad 2012 _ eur.phys.j.c _ * 72 * 2097 .
koivisto t and mota d f 2006 _ phys .
d _ * 73 * 083502 .
daniel s f 2008 _ phys .
d _ * 77 * 103513 .
fedeli c , moscardini l and bartelmann m 2009 _ astron.astrophys .
_ * 500 * 667 .
ujjal debnath , 2012 _ astrophys .
space sci . _ * 337 * 503 .
davies p c w 1988 _ class . quantum grav . _ * 5 * , 1349 .
yungui gong , bin wang and anzhong wang 2007 _ jcap _ * 01 * 024 .
bin wang , yungui gong and elcio abdalla 2006 _ phys .
rev.d _ * 74 * 083520 .
diego pavon and ninfa radicella 2013 _ gen .
gravtit . _ * 45 * 63 .
bekenstein j d 1974 _ phys .
d _ * 9 * 3292 .
bekenstein j d 1975 _ phys .
* 12 * 3077 .
susskind l 1995 _ j. math .
phys . _ * 36 * 6377 .
stephens c r , gt hooft and whiting b f 1994 _ class . quan .
_ * 11 * 621 .
rong - gen cai , li - mingcao and ya - peng hu 2009 _ class .
quantum grav . _ * 26 * 155018 .
zhang x and wu f q 2005 _ phys .
d _ * 72 * 043524 .
huang q g and li m 2005 _ jcap _ * 0503 * 001 .
zhang x 2005 _ int .
d _ * 14 * 1597 .
zhang x 2010 _ phys .
b _ * 683 * 81 .
caldwell r r 2002 _ phys .
b _ * 545 * 23 .
chattopdhay s 2011 _ eur.phys.j.plus _ * 122 * 130 .
fu t f , zhang j z , chen j q , and zhang x 2012 _ eur.phys.j.plus _ * 72 * 1932 .
setare m r 2007 _ jcap _ * 01 * 023 .
li m 2004 _ phys .
b _ * 603 * 1 .
malekjani m , khodam - mohammadi a and nazari - pooya n 2011 _ astrophys .
space sci . _ * 332 * 515 . huang q g and gong y 2004 _ jcap _ * 0408 * 006 .
granda l n and oliveros a 2009 _ phys .
b _ * 671 * 199 .
mathew t k , suresh j and nair d 2013 _ int . j. mod .
* 22 * 1350056 .
chimento l p and richarte m g 2011 _ phys .
rev d _ * 84 * 123507 .
chimento l g and richarte m g 2012 _ phys .
d _ * 85 * 127301 .
knop r a et .
2003 _ astrophys
. j. _ * 598 * 102 .
karami k , sheykhi a , sahraei n and ghaffari s 2011 _ euro .
* 93 * 29002 .
karami k , ghaffari s and soltanzadeh m m 2010 _ class .
quantum grav . _
* 27 * 205021 .
german izquierdo and diego pavon 2006 _ phys .
lett b _ * 633 * 420 .
mathew t k , aiswarya r. and vidya k s 2013 _ eur . phys .
j. c _ * 73 * 2619 .
pollock m d and singh t p 1989 _ class .
quantum grav . _ * 6 * 901 .
cai r g and kim s p 2005 _ jhep _ * 0502 * 050 .
hayward s a , mukohyana s , and ashworth m c 1999 _ phys .
a _ * 256 * 347 .
hayward s a 1998 _ class .
quantum grav . _ * 15 * 3147 .
bak d and rey s j 2000 _ class .
quantum grav . _
* 17 * l83 .
gibbons g and hawking s w 1977 _ phys .
d _ * 15 * 2738 padmanabhan t 2002 _ class .
_ * 19 * 5387 .
davies t m , davies p c w and lineweaver c h 2003 _ class .
quanum grav .
_ * 20 * 2753 .
debnath u , chattopadhyay s , hussain i , jamil m and myrzakulov r 2012 _ eur .
j. c _ * 72 * 1875 .
kolb e w and turner m s , `` the early universe '' , addison - wesley pub . com . | in this paper we have considered the holographic dark energy and studied it s cosmology and thermodynamics .
we have analysed the generalized second law ( gsl ) of thermodynamics in a flat universe consists of interacting dark energy and dark matter .
we did the analysis both under thermal equilibrium and non - equilibrium conditions .
if the apparent horizon is taken as the boundary of the universe , we have shown that the rate of change of the total entropy of the universe is proportional to @xmath0 which in fact shows that the gsl is valid at the apparent horizon irrespective of the sign of @xmath1 the deceleration parameter .
hence for any form of dark energy the apparent horizon can be considered as a perfect thermodynamic boundary of the universe .
we made a confirmation of this conclusion by using the holographic dark energy .
when event horizon is taken as the boundary , we found that the gsl is only partially satisfied .
the analysis under non - equilibrium conditions revealed that the gsl is satisfied if the dark energy temperature is greater than the temperature of dark matter . |
renormalizable quantum field theories like qed , qcd or pseudoscalar yukawa theory are characterised by anomalous dimensions which determine the asymptotic properties of the green functions as all momenta are scaled .
specifically the anomalous dimension @xmath2 of a quantum field is determined through the asymptotic behaviour of the propagator @xmath3 when @xmath4 .
the calculation of @xmath2 as a power series in the renormalized coupling constant @xmath0 , arising in the trilinear interaction lagrangian , has occupied much time and effort and is known up to order @xmath5 for certain models .
it would be nice if one could calculate @xmath2 to all orders of @xmath0 by summing subclasses of diagrams , corresponding to particular topologies , without excessive effort .
this paper is devoted to outlining a method by which this might be achieved .
it is based on conformal scaling of green functions at asymptotic momenta , with anomalous dimensions providing the self - regulation of the field theory @xcite in question .
such a proposal makes eminently good sense at a zero @xmath6 of the beta function on the positive real line and we simply assume that this applies in all that follows .
however it must be pointed out that for the models considered later there is no indication of such a zero and with its corresponding coupling constant . the basic idea behind the method is to eliminate the renormalization constants as far as it is possible and to write the renormalized green functions in terms of one another as a power series in the coupling constant with ever more complicated topologies .
this produces a set of self - consistent schwinger - dyson equations of the full propagator in terms of itself and the proper vertex part , though there exist infinitely many different topological terms in the skeleton expansion of course . in the end one
is forced to truncate these topological contributions , but in any case it is rather easy to show ( see below ) that the only self - consistent behaviour is one where the anomalous vertex and propagator dimensions are oppositely related .
the equations in principle then yield an equation for the anomalous dimensions in terms of the renormalized coupling constant . in the past
it has proved relatively simple to sum over particular topologies where the vertex function remains undressed , but in this paper we shall emphasize the importance of considering the full three - point vertex ( not just in some kinematic limit ) when studying these equations ; this is the novel aspect of our work and by this means one may hope to improve on the rainbow @xcite , ladder and chain @xcite approximations of diagram sets the favourite ones studied thus far and with great aplomb by broadhurst and kreimer . to illustrate what is involved ,
consider the case of the renormalizable @xmath7 interaction in six dimensions , rather than @xmath8 theory @xcite , ignoring triple @xmath9 interactions for the moment .
including renormalization constants , the equation for the @xmath10 and @xmath9 propagators , @xmath11 and @xmath12 respectively , expressed in terms of full green functions , in the massless case reads @xmath13 @xmath14 for each field , we know that in renormalized perturbation theory the sum of the series will produce the large @xmath15 scaling behaviour , @xmath16 , where @xmath2 is given as a series in @xmath17 , which can be computed appearing in @xmath18 behaves like unity when @xmath19 ; this is because the imaginary part of @xmath20 equals @xmath21 and this generalised function reduces to @xmath22 for the free case @xmath23 . ]
order by order in perturbation theory , albeit with greater and greater pain as the power of @xmath17 rises .
the purpose of this paper is to look for relations between @xmath2 and @xmath17 , which correspond to particular truncations of various topological contributions to the self - energy ; such relations are normally given by a transcendental equation that can be found via the dyson - schwinger equations .
but how do we find them by suitably manipulating eqs ( 1 ) and ( 2 ) ?
for the moment neglect the first term on the right , which involves the renormalization constant and seems to scale as @xmath24 ; the remaining terms have the form @xmath25 and we must ask how they sum up to the scaling form on the left - hand side . it is possible , but extremely unlikely , that each of these terms behaves as @xmath24 aside from logarithms and that each contains an infinity ( which is subtracted off by the @xmath26-factor on the right ) and that they somehow combine to produce the anomalous scaling ; if that were true there would be little point in using the skeleton expansion as a means of transcending perturbation theory .
a more likely scenario , which we shall _ assume _ hereafter , is that each of the contributions @xmath27 scale in exactly the same way .
then it is not hard to work out what must be the scaling behaviour of the proper vertex @xmath28 to produce @xmath29 at large @xmath24 .
one easily establishes that @xmath30 will ensure that _ all _ of the topological contributions @xmath27 to the two - point function yield the same scaling as the external momentum @xmath31 .
[ without losing too much generality one may fix the vertex function @xmath32 either at the symmetrical euclidean point @xmath33 * or * at zero - momentum transfer @xmath34 , but that still leaves vast freedom in the dynamics of @xmath28 in ( 2 ) , through its dependence on the momentum ratios . ] as for justifying why we drop the renormalization terms @xmath35 , we note that the wave function renormalization constant may be generally defined by @xmath36 and then @xmath26 can well vanish applies to composite particles and agrees with this observation , but becomes problematic for positive @xmath2 . ] for negative @xmath2 , in which case dropping it from the self - consistent but nonperturbative schwinger - dyson equations is not entirely absurd in the asymptotic limit .
specifically , let us contemplate the situation where one the fields , say @xmath9 , remains undressed ( some sort of quenched approximation in which closed @xmath10-loop graphs are disregarded , signifying that induced multisigma interactions can be dropped ) .
thus its corresponding @xmath37 , so the scaling of @xmath28 just becomes tied to the asymptotic behaviour of the field @xmath10 : they are inverse to one another .
this connection is unsurprising especially in qed where the ward identity @xmath38 at zero photon momentum , already indicates that the scaling properties of the charged field and the soft vertex are intimately tied ( ` @xmath39 ' in a perturbative context ) . in section 2
we summarize what is known about the vertex function in @xmath40 and yukawa theory , the two models upon which we focus .
there we look for a nonperturbative version of @xmath28 which reduces to first order perturbation theory , _ with the correct singularities_. then in section 3 we show how this can be used to determine the relation between @xmath2 and @xmath17 even in the simplest kind of truncation encompassing the chain and rainbow summations with various model vertex functions .
one of the most elementary exercises in quantum field theory is to work out the first order corrections to the propagator and vertex part , where we normally encounter infinities that must be renormalized .
working in 2@xmath41 dimensions ( for isolating the infinities ) leads one to the typical expressions for the self - energy @xmath42 and proper vertex correction @xmath43 , @xmath44^{2-\ell}}\ ] ] @xmath45^{3-\ell}},\ ] ] where @xmath46 or 3 in the integer limit and @xmath47 are symmetry factors including @xmath48 .
subtraction of the pole terms by renormalization constants produces logarithms of the denominators of each of the right hand sides , normalized at some mass scale @xmath49 , which is how dimensional transmutation and the renormalization prescription enters
. one can recognize the resulting expressions as the first terms of an expansion which strictly makes sense at a zero of the beta - function when we proceed to the nonperturbative regime of the green functions in powers of @xmath17 ; for example for the massless two - point function , where the logarithmic term is easily computed , we end up straightforwardly with @xmath50 in keeping with @xmath51 . for the three - point function ,
renormalized at zero @xmath9 momentum , the logarithm is actually a dilogarithm , being given by the integral @xmath52\end{aligned}\ ] ] whose singularities characterize the ` triangle ' graph .
it simplifies to pure logarithms when @xmath53 , being proportional to @xmath54,\ ] ] which one should note is the order @xmath2 term of the expression @xmath55.\ ] ] this bears an uncanny resemblance to the ratio @xmath56/(p_2 ^ 2-p_1 ^ 2)$ ] that one encounters when `` solving '' for the longitudinal 3-point vertex in gauge theories ; we ll come back to this point presently . in any case
it strongly suggests that the nonperturbative form of the vertex possessing triangular topology exponentiates to @xmath57^{\gamma_\gamma},\ ] ] where the ( symmetric in @xmath58 ) spectral function @xmath59 equals 1 up to first order in @xmath17 and @xmath60 .
if we restrict ourselves to such topology , neglect the dressing of @xmath9 and reinsert ( 4 ) into ( 1 ) , this will produce a self - consistent equation for @xmath61 in terms of @xmath17 , as we already know that uniform scaling of @xmath27 requires @xmath62 when @xmath9 is quenched .
the feynman parametric form of @xmath43 in the limit as @xmath63 , ( but _ not _
@xmath64 ) is very suggestive of another lehmann - like representation which one encounters in the gauge technique . thus by changing variable to @xmath65 in the self - energy , one may convert from feynman to dispersive form : @xmath66 in the particular case that @xmath67 for the vertex function , which is often true , one can make a similar conversion in @xmath43 as @xmath53 : @xmath68^{3-\ell}}\ ] ] @xmath69,\ ] ] @xmath70 thus we anticipate that the vertex possesses a simplified dispersive - like representation when one of the momenta is lightlike , and this property might prove rather useful .
further progress may need additional truncation of the skeleton expansion and other practical simplifications .
these remarks apply to all theories with basic trilinear interactions . before looking at nonperturbative aspects of the models @xmath71 and @xmath72 ,
let us first note the order @xmath17 results for @xmath73 since they supply helpful weak - coupling limits for the two and three point functions involving massless fields @xmath74 and @xmath75 . in @xmath76
one trivially finds that @xmath77 where the limit @xmath78 must be taken .
renormalizing at @xmath79 , one ends up with @xmath80 likewise the vertex correction is just expression ( 4 ) , with @xmath81 .
it may be expressed @xcite as an appell function , but better still , it can be converted into a symmetrical sum of three hypergeometric functions in arbitrary dimensions @xcite ; we will utilise this presently . turning to @xmath72 theory ,
the inverse renormalized @xmath75 propagator to that order reads , @xmath82,\ ] ] while the fully off - shell proper vertex part correction is given by @xmath83)\gamma_5,\ ] ] where , of the two scalar functions @xmath84 and @xmath85 , only the former carries the ultraviolet divergence .
( a massive theory , apart from modifying @xmath84 and @xmath85 , would have led to further terms like @xmath86\gamma_5 $ ] . )
one readily finds that @xmath87\ ] ] and @xmath88\end{aligned}\ ] ] to be taken in the limit @xmath89 . in eq .
( 10 ) , @xmath18 is nothing but the kllen function , namely @xmath90=p^4+p'^4+(p - p')^4 - 2p^2p'^2 - 2p^2(p - p')^2 -2p'^2(p - p')^2.\ ] ] it is interesting to look at the case @xmath91 in ( 9 ) and ( 10 ) before worrying about renormalization ; one gets @xmath92}{p'^2-p^2 } \left(\frac{1}{(p'+k)^2}-\frac{1}{(p+k)^2}\right)\right]\gamma_5.\ ] ] again we notice the structure @xmath93 in the finite part , even though we are not dealing with a gauge theory . anyhow , renormalizing so that @xmath94 for @xmath79 , the finite @xmath85-type term remains unaffected , and we are left with the lightlike limit @xmath95 \right]\gamma_5.\ ] ] our aim is to identify these expressions as first order in @xmath17 parts of some nonperturbative construct , so as to capture at the very least the triangular topology of the full vertex function , and then see how far we can take it from there . let us start by reducing the perturbative vertex to a more manageable form .
the first order expression for @xmath96 involves hypergeometric functions of two variables when the dimension and masses are arbitrary .
however it can be simplified for @xmath97 in a very elegant way and davydychev @xcite has shown the way to do this : draw three lengths of size @xmath98 and let @xmath99 if a euclidean triangle can be drawn with those sides , 0 otherwise . then @xmath100 is four times the square of the area of such a triangle .
the internal angles @xmath101 of the triangle are given in an obvious notation by @xmath102 , etc .
and of course @xmath103 .
the result for @xmath104,\ ] ] which leads to the hypergeometric answer , @xmath105\end{aligned}\ ] ] can be converted into the beautiful form , @xmath106.\ ] ] another useful expression ( in that limits can be taken more easily ) is obtained by adopting an integral representation of the hypergeometric functions ; thus @xmath107\!.\ ] ] davydychev has gone further and expressed @xmath43 as a series in @xmath108 leading to polylog functions ls@xmath109 , but we shall not require that expansion .
note that the residue at @xmath110 of eq.(12 ) or ( 13 ) vanishes ( as it must since the vertex is convergent in 4-d ) , and that the residue in 6-d at @xmath111 reduces to @xmath112 , because of the identity , @xmath113 another interesting situation arises when @xmath114 , whereupon @xmath115 .
thus @xmath116}{[p'^2-p^2]},\ ] ] which allows one to take the limit @xmath117 , namely @xmath118 as @xmath78 .
we now suggest a way of going nonperturbative " , which captures the essence of the vertex triangular topology .
we firstly observe that the asymptotic form of the inverse propagator @xmath119 can be gotten directly from the self - energy @xmath120 in @xmath121 dimensions simply be making the replacement @xmath122 in the dimensionally continued result , apart from an overall factor that must be carefully chosen to accord with the renormalization condition .
this procedure will lead from ( 6 ) to @xmath123 with @xmath124 to first order . applying the same procedure to the @xmath125 vertex
, we will end up with @xmath126,\ ] ] where @xmath127 to first order . realising that @xmath128 and @xmath129 we can readily establish that @xmath130 when @xmath131 . also one may check that an expansion of ( 15 ) to order @xmath17 reproduces perturbation theory , including terms like @xmath132 which appear after renormalization .
thus we are emboldened to regard ( 15 ) as a decent nonperturbative vertex that incorporates triangular topological contributions , but whether we can easily make use of it is entirely another matter since its form is analytically complicated .
perhaps a more amenable form of ( 15 ) is @xmath133,\ ] ] because , without too much trouble , it allows us to take the lightlike limit , @xmath134\!,\ ] ] a result which we foresaw earlier .
the special limit when one leg carries zero momentum , @xmath135 is then readily found .
also the renormalized perturbative answer stated at the end of the last subsection falls out upon expansion to first order in @xmath136 , provided the anomalous dimension is correctly identified .
let us perform a similar procedure on @xmath137 . here
the full vertex , consists of two terms : @xmath138)\gamma_5,\ ] ] whose first order in @xmath17 terms are summarised in ( 9 ) and ( 10 ) . in making the substitution @xmath139 so as to obtain a nonperturbative expression
, it is very easy to handle the propagator ( 7 ) and arrive at @xmath140 also it is possible to exponentiate the ( renormalized ) self - energy like terms in the vertex parts @xmath84 and @xmath85 arising in eqs ( 9 ) and ( 10 ) : @xmath141 on the other hand the full triangular topology integral ( 12 ) or ( 13 ) will produces a _ finite _ result @xmath142 .\ ] ] one may verify that the vertex scaling behaviour is reproduced in @xmath143 and that it vanishes for @xmath144 , as it should in zeroth order .
anyhow , combining the terms , we arrive at the nonperturbative yukawa vertex parts , @xmath145 @xmath146 + ( p - p')^2\gamma_{fb},\ ] ] where @xmath147\gamma_f + 2(-(p - p')^2/\mu^2)^{\gamma_\gamma}-(-p^2/\mu^2)^{\gamma_\gamma } - ( -p'^2/\mu^2)^{\gamma_\gamma}$ ] .
the question is how to apply all this .
one s first inclination is to substitute the vertex ( 15 ) or ( 16 ) into ( 1 ) and ( 2 ) so as to find the relation between anomalous dimension and coupling constant a relation which is normally found by a tedious process of perturbative renormalization and is worked out to order @xmath148 at least .
our proposal is that nonperturbative forms of propagator and vertex avoid the need for renormalization since the skeleton expansion is automatically regularized at the physical dimension @xmath149 or 6 ; as we shall see , such a procedure will lead to a transcendental relation between @xmath2 and @xmath17 .
this happens even when one truncates to the first contribution to the self - energy , having the form @xmath150 although one ought properly to consider the entire series of skeleton terms with their ever more intricated topologies ( and matching vertices ) .
nonetheless , considering even the first term of the skeleton series is a substantial improvement on past efforts @xcite and is worthy of study .
the task of evaluating the first term @xmath151 is rather daunting : one is required to integrate the product of two dressed propagators with the _ square _ of expression ( 15 ) or expression ( 16 ) , and this is technically very demanding .
numerical calculations are useless from that viewpoint because one is interested in obtaining the analytical connection between @xmath2 and @xmath17 , even if comparison with perturbation theory eventually necessitates a power series expansion in @xmath17 . a number of helpful auxiliary integrals are collected in the appendix , but we are forced to admit that the full - blown integration producing @xmath152 is presently beyond our technical reach . _ faute de mieux _ we are forced to approximate the skeleton feynman integral by one which is doable and which captures the essence of the idea : the main thing is to ensure that the non - perturbative vertex has the correct scaling behaviour , symmetry properties and analytic behaviour as far as possible , and that it should by itself regularize the intermediate momentum integral . in the following
we shall attempt to use a few approximations to the nonperturbative triangular vertex that reflect its main features , in order to extract the relation between anomalous dimension and coupling constant .
thus we will examine a number of models in which @xmath152 is free of infinities and is automatically regularized it all too easy to construct models which have the correct scaling property but which nevertheless contain infinities but where the vertex singularities are not quite correct . here
we make the choice @xmath153 and leave the @xmath9 propagator undressed .
this has the virtue of simplicity ; it possesses symmetry at the @xmath10 legs and correct scaling but is otherwise awry in its analytical properties and especially its vertex singularities . ignoring these defects , and using the results in the appendix
, we obtain the self - consistency relation @xmath154 or @xmath155 this may be contrasted with the rainbow approximation where the self - consistency relation instead reads , @xmath156 to obtain a perturbative expansion of ( 21 ) we take a series in @xmath2 or @xmath157 as needed to arrive at @xmath158 compared with @xcite @xmath159 we now consider a vertex which better captures the analytical behaviour of the triangular topology but which is necessarily more complicated than the previous model . here
we try to mimic some of the dependence on @xmath160 which arises in ( 11 ) by letting @xmath161/\delta_e.\ ] ] making use of the equations in the appendix , we may arrive at the self - consistency relation , @xmath162 \nonumber \\ & = & -\frac{a}{6}\left[\frac{1}{\gamma}+\frac{11}{6}+\frac{41\gamma}{36 } + \cdots \right].\end{aligned}\ ] ] this corresponds to the series @xmath163 before making any approximations , we may note that the @xmath152 contribution to the inverse fermion propagator @xmath165 , with @xmath166)\gamma_5\ ] ] can be written as @xmath167.\ ] ] therefore , if we use a quenched @xmath9 approximation , the @xmath152 term leads to the following self - consistent relation for the anomalous dimension , @xmath168\ ] ] into which we may feed various models for the vertex parts @xmath169 .
of course we would dearly have loved to make use of their nonperturbative forms given in the previous section ( consistent with triangular topology ) but find the resulting computation too hard analytically ; so we are obliged to model something resembling the true vertex that is within our capabilities .
noting expressions ( 9 ) and ( 10 ) and ensuring correct scaling , we take @xmath170\left(\frac{p^2k^2}{\mu^2(p+k)^2 } \right)^\gamma.\ ] ] inserting this into ( 24 ) we end up with @xmath171,\qquad a\equiv \frac{g^2}{16\pi^2},\ ] ] corresponding to @xmath172 this result should be compared with the rainbow and chain approximations @xcite which , for yukawa theory , read @xmath173 we saw earlier that in the lightlike limit of one of the momenta , _ not necessarily the limit of zero momentum transfer _ , the vertex function assumed the form of the difference of two self - energies , at least to first order in perturbation theory .
let us therefore make the approximation @xmath174 in quenched @xmath125 theory , since we have already disregarded the @xmath9 field dressing . using the usual spectral form of the @xmath10 field propagator , we may therefore substitute @xmath175 in the schwinger - dyson equation , @xmath176 recalling that @xmath177 , the spectral equation reduces to @xmath178 where @xmath179 $ ] is the first order self - energy for a @xmath10 field of mass @xmath180 , to be taken in the limit as @xmath78 .
this has essentially the same form as in the gauge technique @xcite for qed , except that it is no longer true that @xmath181 and this is just what one needs !
the point is that the self - energy carries the infinity ( in its real part ) leaving us with the representation @xmath182+\frac { ( p^2-w^2)^2}{\pi}\int\frac{\im\sigma(s , w)\,ds}{(s - w^2)^2(s - p^2-i\epsilon)},\ ] ] while @xmath183 and they both combine neatly to produce a factor @xmath184 in the numerator of ( 28 ) . therefore taking the imaginary part of ( 28 ) , one arrives at @xmath185 @xmath186 ( 29 ) may be solved by use the ansatz @xmath187 , yielding the sought - after relation @xmath188 only exact to order @xmath157 .
evidently @xmath189 has the same anomalous dimension as its spectral function ( or imaginary part ) @xmath190 .
all the above model conclusions should be treated with great caution and some scepticism .
we supplied models of the true vertex whose scaling coincided to second order with known results for the anomalous dimension , but which possessed incorrect singularities .
it would have been quite easy to change radically the results of our calculations by inputting equally plausible vertex anstze .
so all we can purport to have demonstrated to the critical reader is that it is possible , _ in principle _ , to regulate the skeleton expansion by using nonperturbative propagators and vertices all the while staying in integer dimensions . in the end
we dare only claim that the scheme outlined in the introduction is a viable method for discovering the anomalous scaling properties of the field theory in question .
after all , the skeleton expansion has recently proved its worth in a similar context @xcite and we have no reason to suspect that it will fail us in the present circumstances .
this said we are at a computational impasse in our approach in attempting to include the proper vertex with its full complement of singularities ; unfortunately we see no easy way out of this difficulty if we shy away from numerical methods .
we thank the australian research council for providing financial support for this project , under grant number a00000780 .
we also benefitted from conversations with dirk kreimer and m kalmykov .
here we present a number of auxiliary integrals which assist in determining the connection between @xmath2 and @xmath17 .
the first one of these is @xmath191.\ ] ] the case @xmath192 is rather well - known to practioners in this field and can be found , using feynman parametric techniques , to equal @xmath193 for @xmath194 we must resort to another method in order to find @xmath195 .
namely we go to the frame where @xmath196 , @xmath197 , so @xmath198 @xmath199 hence @xmath200 but the standard texts inform us that @xmath201 and @xmath202 putting this all together we end up with @xmath203 in particular , for 6d , the result reads @xmath204 therefore if one takes @xmath205 non - integer for the present , the limit as @xmath206 will produce a vanishing result leading to @xmath207 .
we take this result to be correct even for integer @xmath205 , paralleling the treatment of tadpole integrals in dimensional regularization .
99 mack g and todorov it 1973 , _ phys .
* d6 * 1764 palchik mya and fradkin es 1999 , _ dokl . phys . _ * 44 * 754 + zaikin vn and palchik mya 2001 , _ theor . math . phys . _ * 128 * 1181 delbourgo r , kalloniatis a and thompson g 1996 , _ phys .
* d54 * 1095 + delbourgo r , elliott d and mcanally d 1997 , _ phys .
_ * d55 * 5230 fukuda r and kugo t 1976 , _ nucl . phys . _ * b117 * 250 + higashijima k 1984 , _ phys . rev . _
* d29 * 1228 + roberts cd and williams ag 1994 , _ prog .
phys . _ * 33 * 477 broadhurst dj and kreimer d 1999 ,
_ j. symb .
* 27 * 581 + broadhurst dj and kreimer d 2001 , _ nucl .
phys . _ * b600 * 403 macfarlane aj and woo g 1974 , _ nucl . phys . _ * b77 * 91 boos ee and davydychev ai 1991 , _ theor .
phys . _ * 89 * 1052 + davydychev ai 1992 , _ j. phys .
_ * 25a * 5587 davydychev ai 2000 , _ phys .
* d61 * 087701 delbourgo r 1979 , _
nuovo cim .
_ * 49a * 484 broadhurst dj , delbourgo r and kreimer d 1996 , _ phys .
lett . _ * b366 * 421 | a self - consistent treatment of two and three point functions in models with trilinear interactions forces them to have opposite anomalous dimensions .
we indicate how the anomalous dimension can be extracted non - perturbatively by solving and suitably truncating the topologies of the full dyson - schwinger set of equations .
the first step requires a sensible ansatz for the full vertex part , which conforms to first order perturbation theory at least .
we model this vertex to obtain typical transcendental relations between anomalous dimension and coupling constant @xmath0 which coincide with known results to order @xmath1 . |
an _ unfolding _ of a polyhedron is obtained by cutting the surface of the polyhedron and flattening it in the plane as a simple non - overlapping polygon .
an _ edge unfolding _ considers only cuts made along edges , while general unfoldings allow cuts anywhere on the surface .
a polyhedron is _ orthogonal _ if each of its faces is perpendicular to a coordinate axis .
an orthogonal polyhedron can not be unfolded using edge cuts alone ( the simplest example is a box sitting on top of a larger box ) .
a _ grid unfolding _ of a polyhedron @xmath0 allows additional cuts along grid edges created by slicing @xmath0 with axis - aligned planes that pass through at least one polyhedron vertex , which we refer to as _
grid planes_. few nontrivial subclasses of orthogonal polyhedra are known to have grid unfoldings : orthotubes @xcite , orthostacks composed of orthogonally convex slabs @xcite , well - separated orthotrees @xcite and terrains @xcite .
a @xmath1-_refined grid unfolding _ , defined for an orthogonal polyhedron and for some integer @xmath2 , allows additional cuts created by at most @xmath1 planes parallel to and sandwiched between adjacent grid planes .
a remarkable breakthrough is the @xmath1-refined grid unfolding of any orthogonal polyhedron homeomorphic to a sphere , where @xmath1 is polynomial in the number of polyhedron vertices @xcite . for constant @xmath1 ,
only two classes of orthogonal polyhedra are known to have a @xmath1-refined grid unfolding : orthostacks @xcite , and manhattan towers @xcite .
informally , an orthogonal polyhedron @xmath0 is an _
orthogrid _ if it is composed of extrusions of simple polygons ( which we call _ slabs _ ) stacked together so that a ( left ) vertex belongs to a unique slab .
it turns out that this restriction forces adjacent slabs to cross each other and collectively form a ( non - uniform ) orthogonal grid structure , hence the name of `` orthogrid '' .
see ahead to fig .
[ fig : defs2 ] for two orthogrid examples .
[ [ our - result . ] ] our result .
+ + + + + + + + + + + in this paper we introduce a new class of orthogonal polyhedra , called _ orthogrids _ , and show that orthogrids have a @xmath1-refined grid unfolding for some constant integer @xmath2 .
our restriction to orthogrids is necessary to guarantee that some front / back surface piece is available at each vertex , so it can be used in unfolding as needed .
let @xmath3 be an orthogonal polyhedron with the surface homeomorphic to a sphere ( i.e. , genus zero ) . classify the faces of @xmath3 according to the direction of their outward normal : _ front _
( @xmath4 ) , _ back _ ( @xmath5 ) , _ top _ ( @xmath6 ) , _ bottom _ ( @xmath7 ) , _ right _ ( @xmath8 ) and _ left _ ( @xmath9 ) .
the vertices of @xmath3 fall into similar categories , with the difference that a top or bottom vertex is also a left or right vertex , depending on its two incident faces .
we take the @xmath10-axis to define the vertical direction ; _ vertical _ faces are parallel to the @xmath11-plane or the @xmath12 plane . clockwise
( cw ) and counterclockwise ( ccw ) directions are defined with respect to a viewpoint at @xmath13 .
the _ gridded _ version of @xmath3 adds edges to the surface of @xmath3 by intersecting @xmath3 with planes parallel to cartesian coordinate planes through every vertex .
we distinguish between an original vertex of the polyhedron , which we call a _ corner vertex _ or simply a _
vertex _ , and the points of the grid ( some of which may be vertices ) .
let @xmath14 be planes passing through every vertex of @xmath3 , orthogonal to the @xmath15-axis . by convention
, @xmath16 is the plane with the largest @xmath15-coordinate . _
@xmath17 is the portion of @xmath3 bounded by planes @xmath18 and @xmath19 , and is composed of one or more connected components called _
slabs_. the _ front _ ( _ back _ ) face of a slab is the face with normal pointing in the @xmath4 ( @xmath5 ) direction , and it may be partly on the surface of @xmath3 and partly interior to @xmath3 .
the top , right , bottom and left faces surrounding a slab form a _
band_. the intersection between a band @xmath20 and a bounding plane @xmath18 is a cycle of edges called a _ rim _ , which we denote by @xmath21 .
each band @xmath20 in a layer @xmath17 has two rims , @xmath21 and @xmath22 .
we use the notation @xmath23 $ ] to denote the closed area enclosed by rim @xmath21 .
( note that @xmath23 $ ] is either the front or the back face of the slab surrounded by @xmath20 . ) two bands @xmath24 and @xmath20 separated by a plane @xmath18 are _ adjacent _ if there is a point common to @xmath25 $ ] and @xmath23 $ ] .
if a vertex @xmath26 belongs to single slab of @xmath3 , then @xmath26 is called _ exposed _ ; otherwise , @xmath26 is _ unexposed .
_ in the example from fig .
[ fig : defs2]b for instance , vertex @xmath27 is exposed , but vertices @xmath28 , @xmath29 and @xmath30 are unexposed . unexposed . ] [ [ orthogrids . ] ] orthogrids .
+ + + + + + + + + + + a genus zero orthogonal polyhedron @xmath3 is an _ orthogrid _
if all left vertices of @xmath0 are exposed .
we further assume that @xmath3 has no _ y - dents _ , meaning that each @xmath15-perpendicular 2d cross section of @xmath0 has genus zero .
note that the meaning of `` left '' in the definition of an orthogrid is tied to a coordinate system , therefore a polyhedron with all right vertices exposed is also an orthogrid ( simply reverse the direction of the @xmath15-axis ) .
[ fig : defs2 ] shows two examples of orthogrids : one with all vertices exposed ( a ) , and one with some right vertices unexposed ( b ) .
single band unfolding is a basic tool used in several unfolding papers @xcite .
the key idea is to identify a path on the box surface that cycles around the top , right , bottom and left faces of the box at least once , and can be flattened out into an @xmath31-monotone orthogonal path in the plane .
we refer to this path as the _ unfolding spiral _ , or simply the _ spiral _ ; we denote the spiral on the box surface @xmath32 , and the flattened version of the spiral in the plane @xmath33 . .
( b ) flattened spiral @xmath33 , slightly thickened to mark face transitions ( c ) thickened spiral that covers the entire band ( d ) 3d hand orientation . ] figure [ fig : unf0]a shows the directed path followed by @xmath32 : it starts at the entering point @xmath34 on the top edge of the front rim @xmath35 and spirals in clockwise direction around the top , right , bottom , and left faces toward the back rim @xmath36 .
we call this spiral strip from @xmath34 up to the point it hits @xmath36 the _ entering strip_. we require that the entering strip cycles around the band _ at least once _ , before moving towards the back rim @xmath36 . when it reaches @xmath36 , @xmath32 turns around by crossing the back face upwards toward the top edge of @xmath36 ; we refer to this back face strip as the _ turnaround _ strip , to emphasize its role in the unfolding . from @xmath36 ,
the unfolding path @xmath32 spirals in a counterclockwise direction back to @xmath35 , ending at the exiting point @xmath37 lying next to @xmath34 on the front rim .
we call this strip piece extending from @xmath36 back to @xmath37 the _ exiting strip_. thus the entering strip and the exiting strip spiral in opposite directions . when @xmath32 is cut out , unfolded and laid horizontally in the plane
, it forms a monotonic staircase @xmath33 as depicted in fig .
[ fig : unf0]b .
the spiral can be subsequently thickened so that it covers the top , right , bottom and front faces entirely , as depicted in fig .
[ fig : unf0]c . marked along the flattened spiral path in fig .
[ fig : unf0]d is an l - shaped indicator ( or simply an @xmath38-indicator ) with two pointers corresponding to the coordinate axes in the plane : the _ hand pointer _ pointing in the direction of the positive @xmath31-axis , and the _ head pointer _ pointing in the direction of the positive @xmath15-axis .
when mapped back onto the 3d box shape , the @xmath38-indicator traces the path taken by @xmath32 .
this makes it easier to visualize the orientation of @xmath33 just by looking at the @xmath38-indicator on @xmath32 : if the @xmath38-indicator glides in the direction of the hand pointer along a piece of @xmath32 , that piece flattens out horizontally to the right in the plane ; if the @xmath38-indicator glides in the direction opposite to the hand pointer along a piece of @xmath32 , that piece flattens out horizontally to the left in the plane .
this is what we will try to avoid ( with small exceptions ) so that @xmath33 maintains its @xmath31-monotonicity .
if the @xmath38-indicator glides in a direction orthogonal to the hand pointer along a piece of @xmath32 , that piece flattens out vertically in the plane . in this latter case ,
the up or down direction is irrelevant , because we seek monotonicity only in the horizontal direction . for this reason
, we will henceforth refer to the @xmath38-indicator as hand , to remind the reader that it is really the hand pointer that affects the monotonicity property of @xmath33 .
nevertheless , the head pointer will be useful in tracing the paths taken by the unfolding spiral on the surface of @xmath3 . in fig .
[ fig : unf0]d for example , one can imagine the hand sliding across the surface in the direction indicated by the arrows ; the head pointer is simply an aid in this visualization process .
( b ) @xmath39 ( c ) @xmath40 ( d ) @xmath41 ( a@xmath42 ) @xmath43 ( b@xmath42 ) @xmath44 ( c@xmath42 ) @xmath45 ( d@xmath42 ) @xmath46 . ] the type of unfolding depicted in fig .
[ fig : unf0 ] and again in fig .
[ fig : unf]a is denoted by @xmath47 .
the symbol @xmath48 is used to indicate that the entering strip starts on the front rim of the band and cycles clockwise around the band .
the subscript @xmath49 is used to indicate that the ( black ) entering strip ( rather than the ( blue ) exiting strip ) runs alongside the rim containing @xmath34 and @xmath37 ; the only place it does nt run alongside the rim is at the small gap at @xmath37 where the exiting spiral leaves the band .
sometimes the unfolding of a band @xmath24 will be required to begin on the back rim of @xmath24 ; we use the symbol @xmath50 to indicate that the unfolding begins on the back rim .
the @xmath51 unfolding is simply an @xmath11-reflection of the @xmath47 unfolding , and is depicted in fig .
[ fig : unf]a@xmath42 .
the only difference between @xmath47 and @xmath51 is that @xmath34 and @xmath37 are located on the front rim in the first case , and on the back rim in the latter case ; the fact that the entering strip starts by making a clockwise cycle around the rim containing @xmath34 applies to both cases .
alternate types of unfolding are @xmath52 ( fig .
[ fig : unf]b ) , @xmath53 ( fig . [ fig : unf]c ) , @xmath54 ( fig . [ fig : unf]d ) , and their corresponding primed @xmath11-reflections @xmath55 ( fig .
[ fig : unf]b@xmath42 ) , @xmath56 ( fig . [ fig : unf]c@xmath42 ) and @xmath57 ( fig . [ fig : unf]d@xmath42 ) . here
the symbol @xmath58 ( @xmath59 ) is used to indicate that the entering strip starts on the front ( back ) rim of the band and cycles counter - clockwise around the band ; and the subscript @xmath60 is used to indicate that the ( blue ) exiting strip runs alongside the rim containing @xmath34 and @xmath37 ( with the exception of the small gap between @xmath34 and @xmath37 ) . for ease of reference we introduce the following definitions .
the band rim containing the entering point @xmath34 is called the _ entering rim _ ; the opposite band rim is called the _ turnaround _ rim , to indicate the fact that the turnaround is executed on the facet of that rim .
note that , if the unfolding type for a band is @xmath48 or @xmath58 , the entering rim is the front rim and the turnaround rim is the back rim ; and if the unfolding type is @xmath50 or @xmath59 , then the entering rim is the back rim and the turnaround rim is the front rim .
call the strip that runs alongside the front rim of a band the _ front strip_. note that the front strip could be either on the entering spiral or on the exiting spiral .
similarly , call the strip that runs alongside the back rim of a band the _ back strip_. the unfolding directions of the front and back strips corresponding to the eight unfolding types depicted in fig .
[ fig : unf ] are summarized in table [ tab : frontback ] .
observe that in the case of an orthogrid , any two adjacent bands exhibit a left face orthogonally crossing a top face of the other band .
( this is because all left vertices of an orthogrid are exposed . )
our unfolding algorithm uses this property in transitioning from one band to the other and back .
the overall structure of the unfolding algorithm is as follows : 1 .
compute a band unfolding tree @xmath61 . select the root of @xmath61 to be a band of smallest y - coordinate , with ties broken arbitrarily ( [ sec : unftree ] ) .
2 . determine forward and return transition paths corresponding to each arc in @xmath61 , and assign an unfolding label to each band in @xmath61 ( [ sec : transitions ] ) .
3 . starting with the root band , unfold each band as a conceptual unit , but interrupt the unfolding each time a forward transition path to a child is encountered .
the unfolding follows the forward transition path to the child band , the child band is recursively unfolded , then the unfolding returns along the return transition path back to the parent , resuming the parent band unfolding from the point it left off . the resulting strip is laid out horizontally in the plane ( [ sec : alg ] ) .
4 . attach the vertical front and back faces of @xmath3 below and above appropriate horizontal sections of the unfolding strip determined in the previous step ( [ sec : frontback ] ) .
we proceed to presenting the details of each of these steps .
the unfolding tree @xmath61 has one node for each band and anchors connecting pairs of adjacent bands as follows . for each band @xmath24 and each band @xmath20
adjacent to @xmath24 , we add arc @xmath62 to @xmath61 and select the _ anchor _
point connecting @xmath24 to @xmath20 to be the leftmost topmost intersection point between @xmath24 and @xmath20 .
it can be verified that the structure @xmath61 constructed this way os a tree that spans all bands . to distinguish edges in @xmath61 from edges of the orthogonal polyhedron @xmath3
, we will refer to the first as _ arcs _ , and to the latter simply as _
edges_. a child is a _ front child _ ( _ back child _ ) if it is adjacent to the front ( back ) of its parent .
an anchor connecting a parent band @xmath24 to a child band @xmath20 falls in one of the following two categories : * the anchor is a point @xmath1 at the intersection between a horizontal edge of @xmath24 and a vertical edge of @xmath20 ( see fig . [
fig : anchors]a ) .
we call @xmath1 a c1-_anchor _ , and @xmath20 a c1-_child_. * the anchor is a point @xmath1 at the intersection between a vertical edge of @xmath24 and a horizontal edge of @xmath20 ( see fig . [
fig : anchors]b ) .
we call @xmath1 a c2-_anchor _ , and @xmath20 a c2-_child_.
in this section we present three key tools used in implementing the recursive unfolding procedure sketched in [ sec : unfidea ] : _ forced turns _ ( [ sec : forcedturns ] ) , _ unfolding variations _ ( [ sec : kinunf ] ) and _ forward and return transition paths _ ( [ sec : transitions ] ) . in the unfolding procedure of @xmath3
, we say that a band @xmath20 has been _ visited _ if an unfolding path has been traced on @xmath20 by hand .
we seek to maintain two invariants throughout the unfolding procedure : 1 .
the hand is parallel to the @xmath31-axis on a horizontal face of a band , pointing in the direction of the unfolding .
if a vertical rim edge @xmath63 of a band @xmath24 contains at least one anchor point attaching an _ unvisited _ child of @xmath24 to @xmath24 , then the hand is parallel to the @xmath15-axis ( pointing to either @xmath4 or @xmath5 ) while moving along @xmath63 .
as we shall shortly see , invariant ( i2 ) enables recursive unfolding of @xmath24 s children attached to @xmath63 .
invariants ( i1 ) and ( i2 ) will sometimes require the hand to change orientation when moving from a horizontal face to a vertical face of a band , or vice - versa .
such a reorientation corresponds to a @xmath64-turn of the strip @xmath33 in the plane .
( recall that @xmath33 is the flattened plane version of the unfolding strip @xmath32 . )
we call such reorientations _ forced turns _ , to distinguish them from the normal turns resulting from the basic unfolding method described in [ sec : boxunf ] .
( a ) horizontal - to - vertical , and ( b ) vertical - to - horizontal . ]
forced turns will be executed only at left vertices of @xmath3 , which by definition are exposed .
let @xmath29 be a convex vertex of a band @xmath24 where a forced turn occurs .
the turn for the case when the hand is on a horizontal ( vertical ) face of @xmath24 upon meeting @xmath29 is depicted in fig .
[ fig : turns]a(b ) , for both @xmath32 ( left ) and @xmath33 ( right ) .
the coordinate axes from fig .
[ fig : turns]a indicate that the top left vertex @xmath29 is encountered in a counterclockwise traversal of the band ; however , a simple reorientation of the coordinate axes shows that this turn also applies to the case when @xmath29 is a bottom left vertex encountered in a clockwise band traversal .
( a ) horizontal - to - vertical , and ( b ) vertical - to - horizontal . ] fig .
[ fig : turns2 ] a(b ) shows the forced turn for the case where @xmath29 is a reflex vertex , and the hand is on a horizontal ( vertical ) face of @xmath24 upon meeting @xmath29 .
note that the paths traced by @xmath33 in fig .
[ fig : turns2 ] are not monotone in the @xmath31-direction , but the extent of the turn steps are adjusted so that @xmath33 does not self - intersect ( the steps can be arbitrarily small , but are exaggerated in the figure for the sake of clarity ) .
the forced turns depicted in figs .
[ fig : turns]a and [ fig : turns2]a indicate that the following property holds : just after a forced horizontal - to - vertical turn , the hand points away from the @xmath65-plane containing @xmath29 if @xmath29 is convex ( fig . [ fig : turns]a ) , and toward the @xmath65-plane containing @xmath29 if @xmath29 is reflex ( fig . [ fig : turns2]a ) . also observe that the clockwise / counterclockwise unfolding direction is the same before and after a forced turn .
this is necessary so that the unfolding advances along the path depicted for the single band in [ sec : boxunf ] , rather than backtracking . to make it possible for vertical - to - horizontal turns to also preserve the unfolding direction , we require one more invariant to be maintained throughout the unfolding procedure : 1
each vertical - to - horizontal forced turn at a corner vertex @xmath29 reorients the hand parallel to the @xmath31-axis , pointing in the the unfolding direction .
just before a vertical - to - horizontal turn is forced at a vertex @xmath29 , the hand points toward the @xmath65-plane containing @xmath29 if @xmath29 is convex ( fig . [ fig : turns]b ) , and away from the @xmath65-plane containing @xmath29 if @xmath29 is reflex ( fig .
[ fig : turns2]b ) . note that the vertical - to - horizontal turns depicted in figs .
[ fig : turns]b and [ fig : turns2]b for @xmath33 are mirrors of the corresponding horizontal - to - vertical turns from figs .
[ fig : turns]a and [ fig : turns2]a , with the unfolding direction reversed .
also note that each horizontal - to - vertical forced turn results in the hand pointing in a direction opposite to the one required to perform the counterpart vertical - to - horizontal turn ( compare figs .
[ fig : turns]a and [ fig : turns]b , and similarly for figs .
[ fig : turns2]a and [ fig : turns2]b ) . as we shall later see , a horizontal - to - vertical turn at a corner vertex @xmath29 of a band @xmath24 is performed only if an unvisited child of @xmath24 is attached to the vertical edge @xmath63 of @xmath24 incident to @xmath29 ( so a forced turn is necessary to maintain invariant i2 ) .
this will enable us to use one such unvisited child to reverse the direction of the hand if necessary . throughout the paper , the terms _ reverse _ and _ opposite _ associated with a vector @xmath66 indicate the scalar @xmath67 multiplied by @xmath66 .
in this section we present one variation to the basic unfolding types presented in
[ sec : boxunf ] , to accommodate for the fact that sometimes the unfolding of a band must begin on a vertical face of the band .
we refer to the unfolding types depicted in fig .
[ fig : unf ] as @xmath31-unfoldings , to indicate the fact that the unfolding starts and ends with the hand parallel to the @xmath31-axis .
we introduce a variation called @xmath15-_unfolding _ , which begins and ends with the hand parallel to the @xmath15-axis ( see ahead the entering and exiting hand in @xmath20 s unfolding from fig . [
fig : c1-transitions ] ) .
consider an unfolding that starts on a vertical face of a band , with the hand parallel to the @xmath15-axis ( see starting point marked @xmath34 in fig . [
fig : reverseturns ] ) .
let @xmath29 be the first vertex encountered by the hand during the unfolding .
our discussion for the rest of this section is based on the assumption that invariant ( i3 ) holds , meaning that upon reaching @xmath29 , the hand is oriented in such a way that a forced turn at @xmath29 maintains the unfolding direction .
after the forced turn , the unfolding spiral follows the same path as in the case of an @xmath31-unfolding , up to the point where the exiting strip reaches vertex @xmath29 on its return trip . at this point
the exiting strip retraces the path taken by the entering strip around @xmath29 and along the vertical face incident to @xmath29 . fig .
[ fig : reverseturns ] shows the retracing path around a convex vertex @xmath29 ( the reflex vertex case is similar ) .
note that the nesting of the unfolding strips around @xmath29 matches the nesting of the strips on the incident band faces , so no crossing occurs .
also note that turns taken in opposite directions at @xmath29 also have opposite hand orientations on both incident ( vertical and horizontal ) band faces .
we remind the reader that the extent of the turn steps are exaggerated for clarity in our figures , but in fact the turns can be made arbitrarily close to the corner vertex .
similarly , the entering and exiting point can be placed arbitrarily close to each other .
the following property will be crucial in proving the correctness of the unfolding algorithm .
the entering hand and the exiting hand on a band have opposite orientations . [
prop : handexit ] corresponding to each arc in @xmath61 connecting a parent band @xmath24 to a child band @xmath20 , we select two transition paths : a _ forward transition path _ from @xmath24 to @xmath20 , and a _ return transition path _ from @xmath20 back to @xmath24 .
when the unfolding strip for @xmath24 reaches a forward transition path to an unvisited child @xmath20 , the unfolding follows this path to @xmath20 , band @xmath20 is recursively unfolded , then unfolding follows the return transition path back to @xmath24 .
each of these two transitions must maintain invariants ( i1)-(i3 ) .
in addition , the return transition must be executed in such a way that the unfolding of @xmath24 can be resumed at the point it left off , as if the interruption did not occur .
in other words , the orientation of the hand on @xmath24 following the return transition must be identical to the one on @xmath24 prior to the forward transition .
this section is concerned with determining the forward and return connecting paths corresponding to each arc in @xmath61 , and assigning an unfolding label to each band in @xmath61 that is consistent with these transitions .
consider an arbitrary arc @xmath68 connecting parent band @xmath24 to child band @xmath20 .
let @xmath18 be the plane separating @xmath24 and @xmath20 .
in the following we discuss the transitions between @xmath24 and @xmath20 corresponding to each of the two anchor classes . in this case
the anchor connecting child @xmath20 to parent @xmath24 is a point @xmath1 at the intersection between a horizontal edge of @xmath24 and a vertical edge of @xmath20 , as depicted in fig .
[ fig : anchors]a .
consider first the case in which the unfolding proceeds clockwise along the top edge of @xmath24 upon reaching @xmath1 . by invariant ( i1 ) ,
the hand is parallel to the @xmath31-axis on the top edge of @xmath24 .
let @xmath63 be the vertical edge of @xmath20 incident to @xmath1 .
after the forward transition to @xmath20 , the hand must be parallel to the @xmath15-axis on @xmath20 ( so that invariant ( i2 ) is satisfied in case unvisited children of @xmath20 are attached along @xmath63 ) and pointing in the direction required by invariant ( i3 ) .
unexposed ( b ) @xmath29 exposed . ] the forward transitions from @xmath24 to @xmath20 are depicted in fig .
[ fig : c1-transitions ] ( follow the paths from @xmath24 to @xmath20 in the direction indicated by the arrows , ignoring for the moment the return transition paths from @xmath20 to @xmath24 ) .
the forward transition depends on the nature of the bottom endpoint @xmath29 of @xmath63 : 1 .
if @xmath29 is convex , then the hand must point toward @xmath18 upon reaching @xmath29 ( invariant i3 ) .
this is accomplished by the forward transition depicted in fig .
[ fig : c1-transitions]a .
( flattened spiral @xmath33 is shown below @xmath32 . ) note that the transition uses a second point @xmath69 at the at the intersection between @xmath24 and @xmath20 on the open vertical segment @xmath70 , to flip the orientation of the hand . the intersection point @xmath69 always exists , because the left vertex @xmath29 is exposed ( by the definition of an orthogrid ) .
if @xmath29 is reflex , then the hand must point away from @xmath18 upon reaching @xmath29 ( invariant i3 ) ; this property is enforced along the entire edge segment @xmath71 by the forward transition depicted in fig .
[ fig : c1-transitions]b ( @xmath33 shown below @xmath32 ) . for the cases depicted in fig .
[ fig : c1-transitions ] , the entry and exit points for @xmath20 are selected so that they support a recursive @xmath15-unfolding of @xmath20 ; these points are marked in fig .
[ fig : c1-transitions ] for each situation identified above .
note that upon exiting @xmath20 at @xmath37 , the hand is parallel to the @xmath15-axis and pointing in the direction opposite to the one used upon entering @xmath20 at @xmath34 ( by property [ prop : handexit ] ) .
this enables the unfolding strip to make the complementary return transition from @xmath20 back to @xmath24 depicted in fig .
[ fig : c1-transitions ] ( follow the return path from @xmath20 to @xmath24 in the direction of the arrows ) .
note that each of these return transitions preserve the orientation of the hand on @xmath24 as required by invariant ( i3 ) . our discussion so far was based on the assumption that the unfolding proceeded clockwise along the top edge of @xmath24 before reaching @xmath1 .
this assumption helps visualize the transitions between the @xmath24 and @xmath20 , however the transitions are general : the case when the unfolding strip proceeds counterclockwise on @xmath24 upon meeting @xmath1 is similar to the clockwise case , with forward and return path switching roles and traversed by the hand in opposite direction .
next we assign an unfolding label to @xmath20 consistent with the transitions discussed above .
the label we assign to @xmath20 depends on whether @xmath20 is a front or back child of @xmath24 .
we discuss three related cases ; the unfolding labels for the second and third case will be derived from the unfolding label established in the first case . [
[ case-1 . ] ] case 1 .
+ + + + + + + @xmath20 is a front child of @xmath24 and the front strip of @xmath24 runs clockwise just before reaching @xmath1 .
we assign to @xmath20 the label @xmath45 to indicate the fact that @xmath20 s unfolding starts at the back of @xmath20 , and the entering strip runs alongside the back rim of @xmath20 in counterclockwise direction .
note that this label is consistent with the transitions from fig .
[ fig : c1-transitions ] . [ [ case-2 . ] ]
+ + + + + + + @xmath20 is a front child of @xmath24 and the front strip of @xmath24 runs counterclockwise just before reaching @xmath1 .
then the same transitions as in case 1 occur , with the forward and return transition strips switching roles and traversed by the hand in opposite direction .
note that the direction opposite to the one indicated by the return strip matches the direction of the forward strip , therefore the unfolding direction is as established for the clockwise case ( case 1 ) .
the only change in the unfolding label from case 1 is switching indices @xmath34 and @xmath37 , to indicate a switch in the roles of the entering and exiting strips . thus the unfolding label for @xmath20 is @xmath46 .
[ [ case-3 . ] ] case 3 .
+ + + + + + + @xmath20 is a back child of @xmath24 . note that an @xmath11-mirror of cases 1 and 2 above depicts the case in which @xmath20 is a back child of @xmath24 ; this transformation preserves the unfolding direction and the notion of left / right , but reverses the notion of front / back .
consequently , @xmath20 s unfolding label for the case when the back strip runs clockwise ( counterclockwise ) just before reaching @xmath1 can be obtained from @xmath20 s unfolding label for the case when the front rim runs clockwise ( counterclockwise ) just before reaching @xmath1 , by replacing @xmath59 by @xmath58 .
see table [ tab : c1labels ] for a summary of all unfolding labels .
c2-transitions apply when the anchor connecting an unvisited child @xmath20 of parent @xmath24 is a point @xmath1 at the intersection between a vertical edge of @xmath24 and a horizontal edge of @xmath20 , as depicted in fig .
[ fig : anchors]b . by invariant ( i2 ) ,
the hand is parallel to the @xmath15-axis upon reaching @xmath1 . after making the forward transition to the horizontal face of @xmath20
, the hand must be parallel to the @xmath31-axis , so that invariant ( i1 ) is satisfied .
we now show how to execute such a transition .
( b ) hand pointing away from @xmath1 . ] 1 .
if the hand on @xmath24 points toward @xmath1 , then the forward transition from @xmath24 to @xmath20 uses a small l - shaped strip from the front / back face of @xmath20 containing @xmath1 , as depicted in fig .
[ fig : c2-transitions]a .
( follow the path from @xmath24 to @xmath20 in the direction of the arrows , ignoring for the moment the return path from @xmath20 to @xmath24 . )
note that fig .
[ fig : c2-transitions ] depicts only the case where @xmath20 is a front child and the unfolding proceeds clockwise along @xmath24 .
however , we will see that all other cases can be derived from this case .
if the hand on @xmath24 points away from @xmath1 , then the forward transition from @xmath24 to @xmath20 is similar to the transition from the previous case , with the unfolding going in opposite direction on @xmath20 .
the transition for this case is depicted in fig .
[ fig : c2-transitions]b .
each of these forward transitions positions the hand parallel to the @xmath31-axis on a horizontal face of @xmath20 , so invariant ( i1 ) is satisfied . by property [ prop : handexit ]
, the hand will be parallel to the @xmath31-axis after an @xmath31-unfolding of @xmath20 , pointing in the direction opposite to the one upon entering @xmath20 .
this property enables the return transitions from @xmath20 to @xmath24 depicted in fig .
[ fig : c2-transitions ] . ( a )
hand pointing toward @xmath1 ( b ) hand pointing away from @xmath1 , rim point @xmath69 exists . ]
we will see that the unfolding algorithm ( discussed later in [ sec : alg ] ) may sometimes require a c2-child band to flip the orientation of the hand for the parent band , so that it is properly positioned for an upcoming forced turn ( as required by invariant i3 ) . in this case , both forward and return transitions are slightly different .
let @xmath29 be the corner vertex of @xmath24 such that @xmath71 is on the rim of @xmath24 and extends from @xmath1 in the unfolding direction of @xmath24 . by invariant
( i3 ) , a hand reversal may be necessary in two situations : ( i ) if the hand points toward @xmath29 and @xmath29 is reflex , and ( ii ) if the hand points away from @xmath29 , and @xmath29 is convex .
the forward and return transitions for case ( i ) are depicted in fig .
[ fig : reverse - c2-transitions]a ( @xmath32 at the top and @xmath33 at the bottom ) , and enable an @xmath31-unfolding of @xmath20 . in case ( ii ) we use in the transition to @xmath20 a small neighborhood around a second point @xmath69 at the intersection between @xmath24 and @xmath20 , as depicted in fig .
[ fig : reverse - c2-transitions]b . the intersection point @xmath69 always exists , because all left vertices are exposed ( by the definition of an orthogrid ) .
note that the return transitions from fig .
[ fig : reverse - c2-transitions ] reverse the orientation of the hand on @xmath24 , as required .
[ fig : c2-transitions ] and [ fig : reverse - c2-transitions ] depict the situation in which the unfolding proceeded clockwise on @xmath24 upon reaching @xmath1 .
the case when the unfolding strip proceeds counterclockwise on @xmath24 upon meeting @xmath1 is similar to the clockwise case , with forward and return path switching roles and traversed by the hand in opposite direction .
finally , a vertical @xmath11-mirror of the above cases depicts the situation in which @xmath20 is a back child of @xmath24 .
we are now ready to assign an unfolding label to @xmath20 consistent with the transitions identified in this section .
recall that @xmath20 s label depends on the unfolding direction and the orientation of the hand on @xmath24 . as in the case of the c1 anchor class , here we discuss the situation where @xmath20 is a front child of @xmath24 and the front strip of @xmath24 runs clockwise just before reaching the anchor @xmath1 attaching @xmath20 to @xmath24 ; all other cases are symmetric . if the hand points towards @xmath1 as it reaches @xmath1 ( as depicted in fig .
[ fig : c2-transitions]a ) , we assign to @xmath20 the label @xmath45 to indicate the fact that @xmath20 s unfolding starts at the back of @xmath20 in counterclockwise direction , and the entering strip runs alongside the back rim of @xmath20 .
note that this label is consistent with the transitions from fig .
[ fig : c2-transitions]a .
if the hand points away from @xmath1 as it reaches @xmath1 ( as depicted in fig .
[ fig : c2-transitions]b ) , we assign to @xmath20 the label @xmath44 to indicate the fact that @xmath20 s unfolding starts at the back of @xmath20 in clockwise direction , and the exiting strip runs alongside the back rim of @xmath20 .
this label is consistent with the transitions from fig .
[ fig : c2-transitions]b .
the unfolding labels for the situation when @xmath20 is a front child of @xmath24 and the front strip of @xmath24 runs counterclockwise just before reaching @xmath1 are derived from the unfolding labels for the situation discussed above , with @xmath34 and @xmath37 switching roles ( because the forward and return transition strips switch roles ) .
finally , the cases where @xmath20 is a back child of @xmath24 are @xmath11-mirrors of the corresponding cases with @xmath20 as a front child , and the unfolding labels can be obtained by simply dropping the prime symbol to indicate that @xmath20 s unfolding starts on the from rim of @xmath20 .
the unfolding labels for all cases are listed in table [ tab : c2labels ] .
later in [ sec : alg ] we will define our main unfolding procedure unfold for a band @xmath24 . this procedure unfolds @xmath24 and all its descendants in the unfolding tree @xmath61 , starting at entering point @xmath34 on @xmath24 and ending at exiting point @xmath37 on @xmath24 .
we assume that the hand is already positioned at entering point @xmath34 on @xmath24 at the beginning of the unfold procedure , and the clockwise / counterclockwise direction of @xmath24 is determined by @xmath24 s unfolding label .
after recursively unfolding all descendants of @xmath24 in @xmath61 ( details to be filled in later ) , the procedure ends at @xmath37 with the hand pointing in the direction opposite to the one indicated by the entering hand . in this section
we make use of this unfold procedure and the tools developed in the previous section to recursively unfold a child @xmath20 of a parent band @xmath24 , as described in the algorithm unfoldchild from table [ tab : c1unfold ] .
the case when @xmath20 is a c2-child that must reverse the direction of the hand on @xmath24 is handled by the procedure unfoldchildreverse defined in table [ tab : c2unfoldrev ] .
having developed the procedure to recursively unfold a single child @xmath20 of a parent band @xmath24 , we now proceed to unfolding all children anchored to a rim of a band @xmath24 , as described in the unfoldrimchildren algorithm from table [ tab : rimunfold ] .
this procedure will be invoked for both the entering and the turnaround rim of @xmath24 . in the case of the turnaround rim @xmath35 of @xmath24
, the unfoldrimchildren procedure may need to handle one special child marked _ turnaround _ by the main unfolding procedure .
this happens when @xmath24 has children anchored to @xmath35 , so a vertical strip from @xmath72 $ ] may be unavailable for the turnaround ( as in the single box unfolding discussed in [ sec : kinunf ] ) . in this case
one particular child @xmath20 of @xmath24 attached to @xmath35 will be charged with reversing the unfolding direction of @xmath24 .
( note that the unfolding strip resulting from the recursive unfolding of @xmath20 plays the role of the turnaround strip in the single band unfolding from fig .
[ fig : unf0]b . )
let unfoldchildretrace be a procedure identical to unfoldchild , but with each return transition modified so that it retraces the forward transition .
the result is that the hand reverses its direction on the parent band .
this procedure will be used in unfolding turnaround children . at the start of the unfoldrimchildren procedure
, the hand can be on a horizontal face of @xmath24 parallel to the @xmath31-axis ( by invariant i1 ) , or on a vertical face of @xmath24 parallel to the @xmath15-axis or the @xmath10-axis ( by invariant i2 ) .
the hand begins cycling around @xmath35 , and processes three types of events that occur during this cycle .
the actions taken in response to each of these events are summarized in table [ tab : rimunfold ] .
next we discuss the details of each event .
[ [ event-1 . ] ] event 1 .
+ + + + + + + + the first type of event occurs when the hand reaches a corner vertex @xmath29 of @xmath24 .
let @xmath73 and @xmath74 be the two faces of @xmath24 incident to @xmath29 , with the hand currently on @xmath73 . at this point ,
the hand is about to make a transition from @xmath73 to @xmath74 . if @xmath73 is horizontal , then the hand is parallel to the @xmath31-axis .
if the vertical rim edge @xmath75 incident to @xmath29 contains no anchor points attaching an unvisited child of @xmath24 , then no action is taken : @xmath24 s unfolding proceeds onto @xmath74 , with the hand parallel to the @xmath10-axis .
otherwise , if @xmath75 contains at least one anchor point attaching an unvisited child of @xmath24 , then the hand must be reoriented parallel to the @xmath15-axis on @xmath74 , so that the invariant ( i2 ) holds .
this is accomplished by a forced turn at @xmath29 .
if @xmath73 is vertical , invariant ( i2 ) tells us that the hand may be parallel to the @xmath10-axis or to the @xmath15-axis .
in the first case , no action is necessary : the hand continues to @xmath74 , where it becomes parallel to the @xmath31-axis , so invariant ( i1 ) is maintained . in the latter case ,
one of the forced turns depicted in figs .
[ fig : turns]b and [ fig : turns2]b must be taken at @xmath29 , depending on whether @xmath29 is convex or reflex . by invariant ( i3 ) , such a forced turn at @xmath29 reorients the hand parallel to the @xmath31-axis , pointing in the unfolding direction . thus invariant ( i1 )
. if this is the first time that @xmath29 gets encountered by the hand , then the unfolding strip marks the first forced turn at @xmath29 .
it is however possible that this is the second time the hand encounters @xmath29 while positioned parallel to the @xmath15-axis ; this situation occurs when the unfolding of @xmath24 starts at a point @xmath34 on @xmath76 and there are children attached to @xmath76 on both sides of @xmath34 ( see fig . [
fig : verticalstart ] ) . in this case
the hand simply traces the path taken by the first forced turn at @xmath29 , so that it becomes parallel to the @xmath31-axis on @xmath74 . on a vertical rim edge ( a ) first visit of @xmath29 ( b ) second visit of @xmath29 . ] in the example from fig .
[ fig : verticalstart ] , the anchor point @xmath1 that attaches @xmath24 to its parent is located on the vertical edge @xmath77 of @xmath24 .
the unfolding of @xmath24 begins at point @xmath34 located on the vertical segment @xmath71 just below @xmath1 , and proceeds downwards with the hand parallel to the @xmath15-axis ; children attached to @xmath24 by anchor points located on the rim segment @xmath78 are recursively unfolded as they are encountered on the way down to @xmath29 .
a forced turn is taken at corner vertex @xmath29 to reorient the hand parallel to the @xmath31-axis .
now note that there may be other children of @xmath24 attached to @xmath24 by anchor points located on the rim segment @xmath79 above @xmath1 .
these children get visited as the unfolding strip moves downwards from @xmath30 , after a forced turn at @xmath30 that reorients the hand parallel to the @xmath15-axis .
once the unfolding strip passes @xmath34 on the way down to @xmath29 , it simply retraces the path previously taken around @xmath29 , until it reaches the the next horizontal face of @xmath24 . at this point , the unfolding strip has visited all children of @xmath24 and the unfoldrimchildren procedure terminates .
[ [ event-2 . ] ] event 2 .
+ + + + + + + + the second type of event occurs when the hand reaches an anchor attaching an unvisited child @xmath20 of @xmath24 . at this point
the parent band unfolding is suspended and @xmath20 is recursively unfolded .
if @xmath20 is a turnaround child , the unfolding direction of @xmath24 is reversed ; otherwise the unfolding resumes on @xmath24 from the point it left off .
if @xmath20 is a c2-child , then the decision whether @xmath20 must reverse the orientation of the hand on @xmath24 so as to satisfy invariant ( i3 ) must be taken at this point .
[ [ event-3 . ] ] event 3 .
+ + + + + + + + the third and final type of event occurs when all rim children of @xmath24 have been visited . in this case
it may be necessary to proceed to the closest horizontal face of @xmath24 , to ensure that the hand ends up parallel to the @xmath31-axis .
this is necessary if the unfolding strip needs to follow the unfolding path towards the other rim of @xmath24 ( as in the single box unfolding from s[sec : kinunf ] ) in order to complete the unfolding of @xmath24 .
in this section we describe the algorithm for recursively unfolding a band and all its descendants .
let @xmath24 be a band to unfold , initially the root band .
the unfolding of @xmath24 begins at entering point @xmath34 and follows the basic unfolding path tied to @xmath24 s unfolding label , as discussed in
[ sec : kinunf ] .
let @xmath35 be the rim of @xmath24 containing @xmath34 , and let @xmath36 be the opposite ( turnaround ) rim of @xmath24 .
the procedure unfold from table [ tab : unfoldrims ] describes our method for recursively unfolding all children attached to the band @xmath24 , starting at the entering point @xmath34 on @xmath24 . in the following
we briefly discuss this procedure .
if @xmath24 has an @xmath80 label , then @xmath24 s entering spiral visits @xmath24 s children attached to @xmath35 , and @xmath24 s exiting spiral visits @xmath24 s children attached to @xmath36 .
after recursively unfolding all @xmath24 s children anchored on @xmath35 ( step 2 ) , the spiral proceeds towards the back rim @xmath36 ( step 3 ) where it needs to turn around and reverse the unfolding direction for @xmath24 .
if @xmath24 has no children attached to @xmath36 , then the facet @xmath72 $ ] is exposed , therefore a vertical strip from @xmath72 $ ] can be used for the turnaround , as in the single band unfolding case ( see the if branch of step 4 ) .
note however that such a vertical strip may be unavailable if @xmath20 has children attached to @xmath36 .
we resolve this issue by using a child @xmath20 attached to @xmath36 to reverse the unfolding direction of @xmath24 .
thus the entering spiral must continue cycling in the unfolding direction up to the first child encountered , which reverses @xmath24 s unfolding direction as required ( see the else branch of step 4 ) .
note that @xmath20 s unfolding subscript must be identical the @xmath24 s unfolding subscript , so that the relative position of entering and exiting spiral is the same in both ( it may help to view the entering spiral of @xmath20 as an extension of the entering spiral of @xmath24 , and the exiting spiral of @xmath24 as an extension of the exiting spiral of @xmath20 ) .
after @xmath20 is recursively unfolded , the exiting spiral for @xmath24 moves alongside the turnaround rim @xmath36 ( because @xmath24 s label is @xmath80 ) and recursively unfolds all children attached to @xmath36 , then retraces the path on @xmath24 back to @xmath35 .
if @xmath24 has a @xmath81 label , then @xmath24 s exiting spiral visits @xmath24 s children attached to @xmath35 , and @xmath24 s entering spiral visits @xmath24 s children attached to @xmath36 . in this case , the child @xmath20 of @xmath24 _ last _ visited by the entering spiral reverses the unfolding direction for @xmath24 , after all other children attached to @xmath36 have been recursively unfolded by the entering spiral ( see the else branch of step 7 ) .
this way , the exiting spiral can proceed immediately back to @xmath35 after the turnaround ( step 8) , and continue unfolding all children attached to @xmath35 ( step 9 ) .
the last strip of the exiting spiral depends on whether the exit point @xmath37 is located on a horizontal or a vertical rim edge of @xmath24 .
if @xmath37 is located on a horizontal rim edge , then the hand simply cycles around in the direction of the hand back to @xmath37 ( see the else branch of step 10 ) .
if @xmath37 is on a vertical rim edge of @xmath24 , it must be that @xmath24 is a c1-child and the exiting hand must be repositioned parallel to the @xmath15-axis , so that property [ prop : handexit ] is satisfied .
( one such situation is depicted in fig .
[ fig : verticalstart ] . )
this can be accomplished by retracing the very first forced turn taken on @xmath24 , as described in
[ sec : kinunf ] ( see the if branch of step 10 ) .
if @xmath82 is the root band of the unfolding tree @xmath61 corresponding to an orthogrid @xmath3 , then unfold(@xmath82 ) produces an unfolding spiral @xmath32 cycling around all band faces of @xmath3 .
the planar piece @xmath33 has the basic staircase shape illustrated in fig .
[ fig : unf0 ] , with only two exceptions .
the first exception occurs at the forced turns around corner vertices depicted in fig .
[ fig : turns2 ] .
we thicken the spiral pieces corresponding to forced turns so that no front / back surface piece gets trapped between such a spiral piece and the band faces incident to the turn vertex .
the second exception occurs at the transitions between bands depicted in figs [ fig : c1-transitions]a , [ fig : c2-transitions ] and [ fig : reverse - c2-transitions]b .
again we thicken the spiral pieces corresponding to transitions between bands so that no front / back surface piece gets trapped between such a spiral piece and the band faces involved in the transitions .
the remaining pieces of @xmath32 ( which correspond to monotone pieces of @xmath33 ) can be thickened in the @xmath15-direction without overlap so that @xmath32 covers all of the top , bottom , left and right faces of the orthogrid .
we now show that the remaining front and back surface pieces that are not part of @xmath32 can be attached orthogonally to @xmath33 without overlap .
consider the horizontal boundary segments of @xmath33 that correspond to boundary segments of @xmath32 adjacent to front and back surface pieces that are not part of @xmath32 .
imagine these boundary segments emanate lightrays downwards .
then all unattached front / back pieces are illuminated .
it can be verified that these pieces can now be attached to the corresponding illuminating rim segments of @xmath33 without overlap .
we believe that the techniques described in this paper can be extended to unfold any orthogonal polyhedra of genus zero with constant refinement of the gridded surface
. however , the details involved in handling unexposed left vertices are much more involved .
therese biedl , erik demaine , martin demaine , anna lubiw , mark overmars , joseph orourke , steve robbins , and sue whitesides .
unfolding some classes of orthogonal polyhedra . in _ proceedings of the 10th canadian conference on computational geometry _ , montral , canada , august 1998 .
mirela damian , robin flatland , henk meijer , and joseph orourke .
unfolding well - separated orthotrees . in _ abstracts from the 15th annual fall workshop on computational geometry _ , philadelphia , pa , november 2005 . mirela damian and henk meijer .
edge - unfolding orthostacks with orthogonally convex slabs . in _ abstracts from the 14th annual fall workshop on computational geometry _ ,
pages 2021 , cambridge , ma , november 2004 . | we define a new class of orthogonal polyhedra , called _ orthogrids _ , that can be unfolded without overlap with constant refinement of the gridded surface .
= 1 |
the basic setup for our problem is as follows : consider the ground state of a periodic critical system in one dimension with the total size @xmath21 .
then we make a projective local measurement in the subsystem @xmath7 with the length @xmath22 of this system in particular local basis . after the measurement
the subsystem @xmath7 will be decoupled ( unentangled ) from the complement @xmath23 .
however , the wave function of the subsystem @xmath23 after measurement , i.e. @xmath24 , is highly entangled . in other words
if we take a subsystem @xmath25 it will be entangled with @xmath6 , where @xmath26 . from now on we will take the length of the subsystem @xmath5 equal to @xmath2 and for simplicity we will take @xmath5 and @xmath6 in a way that they are adjacent simply connected domains . here
we are interested to calculate the entanglement entropy of the subsystem @xmath5 with respect to @xmath6 , i.e. @xmath27 , in a critical quantum chain .
to calculate such a quantity it is first useful to think about the euclidean version of the system after measurement . for a periodic boundary conditions [ htb ] [ fig1 ] .
the removed slit domain @xmath7 has length @xmath22 and we are interested in calculating entanglement entropy of the region @xmath5 with length @xmath22 with the complement in the quantum spin chain
. the twist operator can be put at point @xmath28.,title="fig:",scaledwidth=30.0% ] using the transfer matrix approach one can simply map the system after measurement to a cylinder with a slit on it as it is demonstrated in figure 1 , see @xcite . in general this system is not necessarily an example of boundary cft except for those cases that the subsystem @xmath7 after measurement picks up a configuration which renormalizes to a boundary cft
. we will discuss this important issue later in the case of critical spin chains .
now if we consider that the outcome of the measurement leads us to a boundary cft , then based on the cardy - calabrese argument @xcite the calculation of the entanglement entropy @xmath27 should boil down to the calculation of the one point correlation function of twist operator sitting on the border between @xmath5 and @xmath6 on the cylinder . in other words
we have : @xmath29 where @xmath30 is the twist operator with conformal weight @xmath31 and @xmath4 is the central charge .
the one point function of the twist operator on the cylinder with a slit can be easily found by mapping the system to the upper half plane by using the map @xmath32 and the following well - known formula of cft @xcite : @xmath33 where @xmath34 is the one point function on the upper half plane @xmath35 .
@xmath36 is the lattice spacing and we took the coordinate of the twist operator at @xmath37 .
putting all together we will have @xmath38 where @xmath39 is a constant and the dots are subleading terms .
notice that in the above formula @xmath22 can not go to zero because cylinder with a slit is topologically different from cylinder . to get the result for before measurement case one can simply put @xmath40 which is the smallest scale in the system ,
then we will have @xmath41 which is the well - known result for the bipartite entanglement entropy in cft @xcite . in the next sections we will verify the validity of the above equation in two important cases : klein - gordon field theory ( harmonic oscillators ) and the critical xx model .
in this section we would like to calculate bipartite entanglement entropy of the ground state of coupled harmonic oscillators after measuring the position of a string of oscillators . consider the hamiltonian of @xmath21-coupled harmonic oscillators , with coordinates @xmath42 and conjugated momenta @xmath43 : @xmath44 the ground state of the above hamiltonian has the following form @xmath45
one can calculate the two point correlators @xmath46 and @xmath47 using the @xmath19 matrix , where @xmath48 is the reduced density matrix of domain @xmath49 .
the squared root of this matrix , as well as its inverse , can be split up into coordinates of the subsystems @xmath49 and @xmath50 , i. e. , @xmath51 the spectra of the matrix @xmath52 , can be used to calculate the entanglement entropy , see @xcite and reference therein , @xmath53}.\ ] ] now if we do measurement on the position of all the oscillators @xmath54 they will take some definite values ( for example , @xmath55=0 ) and eventually will get decoupled from the rest of the oscillators . in other words the final state will be the same as ( [ groundswave ] ) but instead of @xmath56 we need to consider @xmath57 which is a subblock of the matrix @xmath56 corresponding to the oscillators in the subsystem @xmath23 .
this means that we now have a new gaussian wave function and one can calculate its bipartite entanglement entropy with the formula ( [ ee harmonic oscillators ] ) .
the results for the short - range harmonic oscillators ( discrete klein - gordon field theory ) are shown in the figure 2 which show perfect agreement with the equation ( [ sb for pbc ] ) if we consider @xmath58 which is the central charge of the free field theory .
[ htb ] [ fig2 ] for short - range coupled harmonic oscillators with total length @xmath59 and the measurement region sizes @xmath60 and @xmath61 with respect to @xmath62 , where @xmath63 .
the full line is the function ( [ sb for pbc ] ) with @xmath58 and @xmath64.,title="fig:",scaledwidth=45.0% ] since the above results are independent of the outcome of the measurement of the the position of the oscillators @xmath54 one can use them to find a lower bound for the localizable entanglement in this system as follows @xmath65 notice that the above result is based on this fact that the scaling properties of the entanglement entropy after partial measurement in harmonic oscillators is independent of the outcome of the measurement .
this is not necessarily correct in generic systems .
[ htb ] [ fig3 ] with length @xmath2 after measurement on subsystem with length @xmath22 with ferromagnetic outcome for xx model in a magnetic field . from top to bottom we have @xmath66 and @xmath67 . in our numerics
@xmath68 is fixed .
the full line is the function ( [ sb for pbc ] ) with @xmath69 and @xmath58 and @xmath70 and the arrows are at @xmath71 from left to right for @xmath66 and @xmath67 .,title="fig:",scaledwidth=45.0% ]
in this section we would like to calculate bipartite entanglement entropy after partial projective measurement in xx model in a magnetic field .
the method that we use can work equally for all the other spin chains that can be mapped to free fermions .
the hamiltonian of xx model is as follows : @xmath72 after using simple jordan - wigner transformation , i.e. @xmath73 , the hamiltonian will have the following form @xmath74 the entanglement entropy of a subsystem can be calculated easily , see @xcite , by first calculating the reduced density matrix @xmath75 and then diagonalizing it , the final formula is [ htb ] [ fig4 ] with length @xmath2 after measurement on subsystem with length @xmath22 with antiferromagnetic outcome for xx model in half filling . in our numerics
@xmath68 is fixed .
the full line is the function ( [ sb for pbc ] ) with @xmath69 and @xmath58 and @xmath76.,title="fig:",scaledwidth=45.0% ] @xmath77}.\ ] ] the matrix @xmath78 is related to the correlation matrix @xmath79 as @xmath80 . in our case
the correlation matrix for infinite system has the form @xmath81 , where @xmath82 .
looking to the jordan - wigner transformation it is easy to see that any measurement in the @xmath16 basis on particular site can be translated to the measuring the number of fermions in that site . in other words
if one measures the @xmath16 in all the sites of a subsystem with size @xmath22 the outcome will be one of @xmath83 possible configurations which can be easily translated to the configurations made of presence or absence of fermions in the sites of the subsystem .
for simplicity we first consider that the outcome of the measurement is a string of @xmath22 fermion occupied sites .
this can be calculated easily by using grassmann numbers by first calculating the reduced density matrix for a sub - system with length @xmath84 and then finding the reduced density matrix of the subsystem with length @xmath2 with the assumption that the outcome of the measurement in the subsystem @xmath22 is a string of filled sites .
the method is explained with full detail in the appendix .
the results for different filling factors are shown in the figure 3 .
there are some comments in order : first of all because of the @xmath85 symmetry of the xx model the number of particles in the system is conserved and for this reason as far as @xmath86 is small it is very difficult to have a string of sites with fermions . in other words for example in the xx model with @xmath87
the most probable outcome is an antiferromagnetic string rather than ferromagnetic string . in @xcite
it is already conjectured that for the half filling case most probably the ferromagnetic configuration is not going to lead to a boundary conformal field theory and so the very first assumption that we used is going to fail . despite this argument we found surprisingly that for @xmath88 the formula ( [ sb for pbc ] ) works perfectly .
one can also check the results for those cases that the outcome of the measurement is antiferromagnetic string .
it is expected that this case leads to dirichlet boundary condition in the bosonization language and so it is related the boundary cft .
the results presented in the figure 4 indeed show that the formula ( [ sb for pbc ] ) works perfectly also in this case .
since the ferromagnetic and antiferromagnetic configurations are at the two extreme sides of the all possible configurations based on the above numerical results one can conjecture that independent of the outcome of the measurement in the @xmath16 basis the formula ( [ sb for pbc ] ) will work if the @xmath2 is bigger enough than @xmath22 .
note that since the number of configurations increases exponentially this conjecture is beyond what we can check numerically .
one can also do all the cft calculations in the case of open boundary condition , the only difference is that now we have a strip with a slit instead of a cylinder .
one can simply map a strip with a slit to upper half plane by using the map @xmath89 .
after some algebra we have @xmath90 the limit @xmath91 simply gives us the well - known formula @xmath92 , see @xcite .
we have checked the validity of this formula in the case of short - range harmonic oscillators using the same method that we hired for the periodic bc .
the results shown in figure 5 are fairly compatible with the formula ( [ sb for obc ] ) .
we noticed that although the log formula is perfectly compatible there is almost 8 percent deviation from central charge that might be due to finite size effects .
[ htb ] [ fig5 ] for a system ( short - range coupled harmonic oscillators ) with total length @xmath59 and the measurement region sizes @xmath93 and @xmath61 with respect to @xmath62 , where @xmath94 .
the full line is the function ( [ sb for obc ] ) with @xmath58 and @xmath95.,title="fig:",scaledwidth=45.0% ]
we described a general set up for calculating bipartite entanglement entropy after local projective measurement in critical one dimensional quantum systems .
exact formulas were derived for bipartite entanglement entropy after
conformal measurements in the case of periodic and open boundary conditions .
the formulas were checked in explicit examples of free bosonic system and xx model in a magnetic field .
we noticed that since bosonization of xxz model in the @xmath16 basis leads to free bosonic system the antiferromagnetic outcome of the measurement in this basis should be compatible with the results of free bosonic system @xcite . in the case of xx model in a magnetic field we showed that if we do our measurements in the @xmath16 basis independent of having ferromagnetic or antiferromagnetic outcome for our measurements the bipartite entanglement entropy in particular regimes can be described with cft formulas .
we also derived a lower bound for the localizable entanglement in the case of harmonic oscillators .
there are interesting questions remained to be answered : first of all in our free fermion approach we were able to handle just @xmath16 basis , it is important to look to the other basis especially @xmath96 basis by using exact diagonalization methods .
another interesting question is investigating the same questions in the case of non - critical quantum systems and the validity of the area law .
some of these questions will be discussed in a forthcoming paper @xcite . finally understanding the problem in the holographic set up
@xcite will surely help to extend some of these results to higher dimensions .
[ [ acknowledgments . ] ] acknowledgments .
+ + + + + + + + + + + + + + + + we thank m. g. nezhadhaghighi for early collaboration on the subject . appendix
in this supplementary note we would like to present a method which can be used to measure the entanglement entropy after measuring the fermion occupation numbers in a subsystem in a generic free fermion system .
the method is based on grassmann numbers and it is a generalization of the work @xcite . although the method can be also used for the most generic free fermion system @xcite here we will concentrate on a system with the following reduced density matrix : @xmath97 the matrix @xmath78 is related to the correlation matrix @xmath79 as @xmath80 . finally one can find the entanglement entropy using the following formula , see for example @xcite ; @xmath98}.\end{aligned}\ ] ] to calculate the reduced density matrix after measurement we need to first define fermionic coherent states .
it can be defined as follows where @xmath100 s are grassmann numbers following the properties : @xmath101 and @xmath102 .
then it is easy to show that @xmath103 using the above formula one can derive the following useful formula : @xmath104 with the same method one can also define another kind of fermionic coherent state as @xmath105 where @xmath106 s are grassmann numbers .
then it is easy to show that @xmath107 and consequently @xmath108 the two formulas ( [ matrix elements1 ] ) and ( [ matrix elements2 ] ) are the main formulas one can use to calculate reduced density matrix after measurement .
we first discuss how the method works for a string of empty sites .
a string of empty sites means that in the equation ( [ matrix elements1 ] ) we need to put all the @xmath100 s with @xmath109 inside the measuring subsystem equal to zero .
this is equivalent to keeping just all the elements of the matrix @xmath110 outside the subsystem .
then we take the logarithm of this matrix and find a new matrix @xmath111 which is going to be our new entanglement hamiltonian and then one can easily calculate the entanglement entropy of that system with the method we explained at the beginning of this section . to do the same calculation for an string of occupied sites instead of equation ( [ matrix elements1 ] ) one needs to use equation ( [ matrix elements2 ] ) and follow the same method as we just explained . in principle
this method can be applied for any possible configuration by just appropriate using the equations ( [ matrix elements1 ] ) and ( [ matrix elements2 ] ) .
m. srednicki , phys .
lett . * 71 * , 666(1993 ) and j. eisert , m. cramer , m. b. plenio , rev .
phys . * 82 * , 277 ( 2010 ) c. holzhey , f. larsen , and f. wilczek , nucl .
b , * 424 * , 443 ( 1994 )
. g. vidal , j. i. latorre , e. rico , a. kitaev , phys .
* 90 * , 227902 ( 2003 ) j. cardy , phys .
* 106 * , 150404 ( 2011 ) .
d. a. abanin and e. demler , phys .
lett . * 109 * , 020504 ( 2012 ) .
a. j. daley , h. pichler , j. schachenmayer , and p. zoller , phys .
109 * , 020505 ( 2012 ) .
m. greiner , `` entanglement detection through interference of quantum many - body twins in ultra cold atoms '' kitp conference 02/06/2015 j. ignacio cirac [ arxiv:1205.3742 ] f. verstraete , m. popp and j. i. cirac , phys .
lett . * 92 * , 027901(2004 ) . s. o. skrvseth , s. d. bartlett , phys . rev .
a * 80 * , 022316 ( 2009 ) ; t. b. wahl , d. perez - garcia , j. i. cirac , phys . rev . a * 86 * , 062314 ( 2012 ) f. sciarrino , e. nagali , f. de martini , m. gavenda , and r. filip phys
a * 79 * , 060304 ( 2009 ) j. simon , w. s. bakr , r. ma , m. e. tai , p. m. preiss , m. greiner , nature * 472 * , 307 ( 2011 ) r. islam , c. senko , w. c. campbell , s. korenblit , j. smith , a. lee , e. e. edwards , c. c. wang , j. k. freericks , c. monroe , science * 340 * , 583 ( 2013 ) | we derive exact formulas for bipartite von neumann entanglement entropy after partial projective local measurement in @xmath0 dimensional conformal field theories with periodic and open boundary conditions . after defining the set up we will check numerically the validity of our results in the case of klein - gordon field theory ( coupled harmonic oscillators ) and spin-@xmath1 xx chain in a magnetic field .
the agreement between analytical results and the numerical calculations is very good .
we also find a lower bound for localizable entanglement in coupled harmonic oscillators . in the last couple of decades
bipartite entanglement entropy attracted a lot of attention in the high energy physics and the many body condensed matter physics mostly due to the area law property of bipartite von neumann entanglement entropy which is the most celebrated measure of entanglement @xcite .
in one dimensional quantum systems while the area law is usually valid just for massive ( gapped ) systems for the critical systems the entanglement entropy of subsystem with the size @xmath2 of the ground state follows the logarithmic formula @xmath3 , where @xmath4 is the central charge of the underlying conformal field theory ( cft ) describing the low energy behavior of the critical system @xcite . since the central charge of the system usually can fix the universality class of the system calculating entanglement entropy for a system can give a lot of insight about the possible universality class of the system .
in particular , since numerical calculation of the entanglement entropy by using the techniques of dmrg in one dimension is now a well - known method @xcite one can easily study the critical and non - critical properties of the quantum systems in one dimension by studying the entanglement entropy .
it is worth mentioning that entanglement entropy is not the only quantity which gives directly the central charge of the system . among other measures one can name , mutual informations @xcite and quantum contours @xcite ,
see also other related works @xcite .
the former one is based on local measurements in particular basis , so called conformal basis @xcite .
one of the simple properties of the entanglement entropy which makes it more appealing in numerical calculations with respect to other measures such as shannon and rnyi mutual informations is that it is completely independent of the basis that we write the wave function .
however , since it is a very non - local quantity it is not an easy quantity to be measured in the experiment , for recent developments see @xcite . although bipartite entanglement entropy has been studied in length in many quantum systems there are few studies regarding entanglement in multipartite systems @xcite .
one of a few entanglement measures regarding tripartite systems is localizable entanglement introduced in @xcite , see @xcite for review .
the localizable entanglement is defined as the maximal amount of entanglement that can be localized , on average , by doing local measurements in part of the system . in other words
the localizable entanglement between the two parts @xmath5 and @xmath6 after doing local measurement in the rest of the system @xmath7 is defined as @xmath8 where @xmath9 is the set of all possible outcomes ( @xmath10 ) of the measurements and @xmath11 is the chosen entanglement measure . because of the maximization over all the possible local measurements the localizable entanglement is a very difficult quantity to calculate . however , in those cases that @xmath5 and @xmath6 are single spins and @xmath12 is the concurrence one can derive interesting lower bounds to the localizable entanglement , see @xcite and references therein and for other related works see also @xcite . the quantity has been also measured experimentally in a system of two photons in a noisy surrounding in @xcite . due to the complexity of the definition of localizable entanglement , to the best of our knowledge ,
the case of @xmath5 and @xmath6 not being just single particles but many particle subsystems has not been addressed in the literature . in this work instead of calculating the very complicated quantity of localizable entanglement
we calculate @xmath13 appearing in the definition ( [ localizable ] ) for particular natural basis . in the case of harmonic oscillators
our exact results are useful to find lower bound for the localizable entanglement .
consider the ground state of a many body system ( for example a spin chain ) written in particular basis , i.e. @xmath14 , where for example in the case of spin chain @xmath15 are all the possible spin configurations in the @xmath16 direction
. if we do local projective measurement of a local quantity ( for example @xmath16 in the spin chain ) in some part of the system that part will take definite spin direction and will be disentangled from the rest .
in other words one can write @xmath17 , where @xmath18 is the projective measurement operator of the subsystem with the outcome @xmath19 and the sum is now over the spins of the rest of the system .
notice that if we start with the ground state of a spin chain the resulting state after measurement is nothing to do with the ground state of the rest of the system .
the entanglement content of the new wave function will be dependent on the basis that we chose for our measurement and of course to the outcome of the measurement , in other words , the entanglement content of the @xmath20 is dependent on @xmath18 . in a recent development @xcite
it is shown that although , in general , doing measurement in arbitrary basis is not compatible with cft set up there are some natural basis ( conformal basis ) that are closely related to boundary cft .
in other words if one makes a measurement in these basis one can still hope to preserve some of the universal properties of the system and be able to calculate the entanglement entropy .
we noticed that engineering quantum spin chains and making projective measurement on natural local basis is now possible with optical lattices and ion trapping techniques , see @xcite and references therein .
however , calculating entanglement entropy after partial local measurement in many body quantum chains has not been investigated yet .
analytical calculation of bipartite von neumann entanglement entropy after projective measurement in natural conformal basis is the main purpose of this article . to do that we first define our set up and fix our assumptions and
then we will analytically calculate entanglement entropy using twist operators of cft .
exact formulas will be derived and then we will check the validity of our results by explicit examples in the field theories such as klein - gordon field theory ( coupled harmonic oscillators ) and in quantum spin chains such as , xx model in a magnetic field . |
one of the not yet solved puzzles of cuprate high - temperature superconductors ( htsc ) is a nature of underdoped normal state - the pseudogap regime@xcite .
perhaps most powerful experimental tool to access electronic properties of the pseudogap state is angular resolved photoemission spectroscopy ( arpes)@xcite .
common understanding is the fluctuating origin of the pseudogap state , however type of the fluctuations is still under discussion .
whether these are superconducting fluctuations@xcite or some order parameter fluctuations ( afm(sdw ) , cdw , stripes , etc.)@xcite coexisting or competing with cooper pairing is up to now undecided .
there are several prototype compounds among high - t@xmath0 systems e.g. hole doped bi@xmath7sr@xmath7cacu@xmath7o@xmath8 ( bi2212 ) system or the electron doped one nd@xmath9ce@xmath10cuo@xmath3 ( ncco ) . now
a lot of experimental arpes data on bi2212 and ncco is available ( see the review ref . ) .
for instance fermi surface ( fs ) maps , quasiparticle band dispersions and even self - energy lineshapes mapped onto some models are obtained from modern arpes measurements @xcite . into this list of prototype componds
should be of course included the first ever high - t@xmath0 hole doped system la@xmath9sr@xmath11cuo@xmath3 ( lsco ) which was also investigated both in theory and experiment with great details.@xcite within the normal underdoped phase ( pseudogap regime ) a number of interesting physical phenomena were discovered .
for example , fs is partially `` destructed '' in the vicinity of the so - called `` hot - spots '' ( points of crossing between fs and afm umklapp surface ) .
`` shadow bands '' ( partial folding of band dispersion ) appear apparently as a result of short range afm order .
formation of the so called fermi `` arcs '' around brillouin zone ( bz ) diagonal reminiscent of the parts of noninteracting fs is experimentally detected in numerous arpes experiments@xcite . despite apparently the same physics behind
, pseudogap regime demonstrates some material specific features . for bi2212 fermi `` arcs '' go almost up to bz border where they are strongly blurred .
the ncco also has fermi `` arcs '' but fs `` destruction '' looks different .
the hot - spots are well observed in ncco while towards the bz border fs is almost restored as a non interacting one.@xcite the present paper is devoted to pseudogap behavior in the underdoped lsco and its comparison with bi2212 and ncco . according to common knowledge high - t@xmath0 systems
are usually doped mott insulators , effectively described by the hubbard model .
most common method to solve the hubbard model in our days is dynamical mean - field theory ( dmft)@xcite .
its exactness in the infinite spatial dimensions limit makes dmft a local approach . since high - t@xmath0 compounds as it is well established have quasi two - dimensional nature spatial fluctuations play important role for their physics . to overcome this difficulty we introduced dmft+@xmath12 computational scheme@xcite which supplies conventional dmft with `` external '' @xmath5-dependent self - energy .
main assumption of the dmft+@xmath12 is additive form of the self - energy which allows one to keep conventional dmft self - consistent set of equations .
the dmft+@xmath12 approach was used to address the pseudogap problem@xcite , electron - phonon coupling in strongly correlated systems@xcite and disorder induced metal - insulator transition within the hubbard - anderson model@xcite . for the pseudogap state
this self - energy @xmath12 describes the interaction of correlated electrons with non - local ( quasi ) static short - ranged collective heisenberg - like antiferromagnetic ( afm or sdw - like ) spin fluctuations@xcite .
dmft+@xmath12 approximation was also shown to be appropriate to describe two - particle properties e.g. optical conductivity.@xcite as a possible way of theoretical simulation of the pseudogap regime for real materials we proposed lda+dmft+@xmath4 hybrid method@xcite .
it combines first principle one - electron density functional theory calculations within local density approximation ( dft / lda ) @xcite with dmft+@xmath4.@xcite lda+dmft+@xmath4 method allowed us to obtain fermi arcs and `` hot - spots '' behavior for both electron doped e.g. nd@xmath13ce@xmath14cuo@xmath3@xcite ( ncco ) and pr@xmath13ce@xmath14cuo@xmath3 ( pcco)@xcite as well as hole doped bi@xmath7sr@xmath7cacu@xmath7o@xmath8 ( bi2212)@xcite high - t@xmath0 cuprates .
pseudogap behavior of dynamic optical conductivity within lda+dmft+@xmath4@xcite was also discussed for b12212@xcite and ncco@xcite . this communication reports lda+dmft+@xmath4 computations of fermi surface and spectral functions for hole underdoped la@xmath1sr@xmath2cuo@xmath3 ( lsco ) system supported by high energy , high resolution bulk sensitive arpes@xcite .
the la@xmath7cuo@xmath3 system has base - centered orthorhombic crystal structure with space group @xmath15 with two formula units per cell@xcite .
corresponding lattice parameters are a=5.3346 , a=5.4148 and c=13.1172 .
atomic positions are the following : la(0.0,-0.0083,0.3616 ) , cu(0,0,0 ) , o(1/4,1/4,-0.0084 ) , o2(0.0,0.0404,0.1837 ) . as a first step of lda+dmft+@xmath4 method we performed density functional theory calculations in the local density approximation ( lda ) for these crystallographic data .
band structure was obtained within the linearized muffin - tin orbitals ( lmto ) method@xcite .
it is well known for these compounds that fermi level is crossed by antibonding o2@xmath16-cu3@xmath17 partially filled orbital of @xmath18 symmetry .
tight - binding parameters for this band were calculated by -th order lmto ( nmto ) method@xcite as ( in ev units ) @xmath19 , @xmath20 , @xmath21 , @xmath22 .
these values agree well with previous studies.@xcite coulomb interaction value on effective cu-3@xmath17(@xmath18 ) orbital was calculated by constrained lda approach@xcite and was found to be @xmath23=1.1 ev .
these lda obtained parameters are used to set up corresponding one - band hubbard model .
the second step is further treatment of the above defined hubbard model within the dynamical mean - field theory ( dmft ) self - consistent set of equations@xcite supplied by `` external '' momentum - dependent self - energy @xmath4@xcite .
using the additive form of self - energy ( main approximation of the scheme neglecting the interference between hubbard interaction and pseudogap fluctuations , which allows one to preserve conventional dmft equations ) one can define lda+dmft+@xmath4 green function as : @xmath24 where bare electron dispersion @xmath25 is defined by lda calculated hopping parameters listed above . to calculate @xmath4 we used a two - dimensional pseudogap model@xcite describing nonlocal correlations induced by ( quasi ) static short - range collective heisenberg - like afm spin fluctuations .
thus we introduce correlation length dependence of the pseudogap fluctuations into conventional dmft loop .
there are two points which make dmft+@xmath4 different from the usual dmft .
first , momentum dependent @xmath4 is recalculated on each dmft iteration ( @xmath26)$ ] is in fact the function of dmft chemical potential and dmft self - energy ) .
second , dmft+@xmath4 lattice problem is defined at each dmft iteration as : @xmath27 after numerical self - consistency is reached we get the green s function ( [ gk ] ) with corresponding @xmath28 and @xmath29 taken on the last dmft iteration .
all further computational details can be found e.g. in refs . .
as an `` impurity solver '' for dmft equations numerical renormalization group ( nrg@xcite ) was employed .
temperature of dmft(nrg ) computations was taken to be 0.011 ev and electron concentration used was n=0.86 .
self - energy @xmath30 due to pseudogap fluctuations depends , in general , on two parameters : the pseudogap amplitude @xmath31 , and the correlation length @xmath32.@xcite the value of @xmath31 was calculated as in ref .
@xmath33 where local densities @xmath34 and double occupancy @xmath35 were calculated within the standard dmft@xcite .
behavior of @xmath31 as a function of hopping integrals and coulomb interaction was studied in our previous work @xcite , while @xmath31 as a function of occupancy @xmath36 was investigated in ref . .
for @xmath32 we believe it is more safe to take experimental values . in this work
the value of @xmath31 was calculated to be 0.275 ev and @xmath32 was taken to be 10@xmath37 , where @xmath37 - lattice constant.@xcite
the high - energy arpes measurements were carried out at bl25su in spring- 8 , using incident photons of 500 ev , on single crystal samples . the normal to the cleaved sample surface was set almost parallel to the axis of the lens of the analyzer and the sample was set to about 45@xmath38 from the incident light direction .
the photoelectrons within polar angles of about @xmath39 around the normal to the sample were simultaneously collected using a gammadatascienta ses200 analyzer , thereby covering more than a whole brillouin zone along the directions of the analyzer slit .
the fermi surface mapping was performed by changing the angle along the perpendicular direction to the analyzer slit .
the base pressure was about 4 x 10@xmath40pa .
the ( 001 ) clean surface was obtained by cleaving the samples in situ in vacuum at the measuring temperature of 20 k. the overall energy resolution was set to 100 and 170 mev for high - resolution measurements and fermi surface mapping , respectively .
the angular resolution was @xmath410.1 ( @xmath410.15 ) for the perpendicular ( parallel ) direction to the analyzer slit .
these values correspond to the momentum resolution of @xmath410.024@xmath42 ( @xmath410.036@xmath42 ) at h@xmath43=500 ev , where @xmath37 is twice the cu - o bondlength within the cuo@xmath7 plane . by virtue of the
longer photoelectron mean free path of @xmath4412 at the kinetic energy of @xmath44500 ev than that for conventional arpes at @xmath4520 - 60 ev , the bulk contribution to the spectral weight is estimated as about 60% .
the position of the fermi level was calibrated with pd spectra .
( @xmath18 ) band dispersion along high symmetry directions of square brillouin zone computed with lda+dmft+@xmath46 .
the fermi level is zero .
[ contour_plot ] ] in case of finite temperature and interaction strength we have to take into account the finite life time effects of quasiparticles .
thus instead of just a dispersions @xmath25 one should rather work with the spectral function @xmath47 given by @xmath48 with retarded green s function @xmath49 obtained via lda+dmft+@xmath4 scheme@xcite .
of course there are considerable lifetime effects originating from the @xmath12 corresponding to interaction with afm fluctuations ( substituted in our approach by the quenched random field ) . in fig .
[ contour_plot ] contour map of spectral function ( [ specf ] ) obtained from lda+dmft+@xmath4 for cu-3@xmath17(@xmath18 ) band is presented .
the width of the spectral function is inversely proportional to the lifetime .
around ( @xmath6,0 ) point one can clearly see splitting of the spectra by afm pseudogap fluctuations of the order of 2@xmath31 .
also afm nature of the pseudogap fluctuations leads to formation of the `` shadow '' band which is much weaker in intensity and becoming the real quasiparticle band in case of complete folding in case of long - range afm order .
figure [ sf ] displays experimental energy distribution curves ( edc ) on the panel ( a ) along ( 0,0)(@xmath6,0 ) direction . around
( @xmath6,0)-point as shown by stars certain deviation of the @xmath47 maxima from the fermi level a kind of `` turn - back '' is observed .
we attribute such behavior of the @xmath47 to pseudogap fluctuations .
similar theoretical behavior is shown on panel ( b)@xcite of fig . [ sf ] as calculated by our lda+dmft+@xmath4 approach ( see also fig .
[ contour_plot ] ) .
the same behavior is also observed as traced by circles in experimental arpes momentum distribution curves ( mdc ) demonstrated on panel ( c ) of fig . [ sf ] .
spectral functions ( b ) along ( 0,0)-(@xmath50 ) high symmetry direction ; ( c ) arpes mdc curves around ( @xmath50 ) point for lsco at x=0.14 .
on panels ( a ) , ( b ) and ( c ) filled circles guide the motion of the @xmath51 maxima .
the fermi level is zero .
[ sf ] ] the bulk - sensitive high - energy arpes data for la@xmath1sr@xmath2cuo@xmath52 show a clear `` turn - back '' structure of the edc peak as a function of momentum near ( 0,@xmath53 ) , which were not seen in the previous low - energy arpes data for la@xmath13sr@xmath14cuo@xmath52 @xcite .
the contour map of the spectral weight in the vicinity of @xmath54 seems to be essentially similar in overall features for this doping level between the high - energy and low - energy arpes studies .
computations ( right panel ) red crosses on the left panel correspond to experimental @xmath55 values .
[ fermisurface],title="fig : " ] + experimental and theoretical fermi surface maps are shown in fig . [ fermisurface ] at panels ( a ) and ( b ) correspondingly .
both pictures reveal strong scattering around ( @xmath6,0)-point which we associate with scattering in the vicinity of the so - called `` hot - spots '' ( cross - points of a fermi and afm umklapp surfaces ) which are close to the ( @xmath6,0)@xcite . such strong scattering comes from scattering processes with momentum transfer of the order of * q*=(@xmath56)@xcite , corresponding to afm pseudogap fluctuations . along nodal directions
we observe typical fermi arcs .
they are pretty well seen in the theoretical data while for experiment we have just narrow traces of them .
our lda+dmft+@xmath4 hybrid approach was shown to be an effective numerical tool to describe short - range ordered state in quasi - two dimensional systems.@xcite material specific model parameters such as hopping integrals ( which define bare electronic band dispersion of effective cu@xmath57 orbital ) were calculated via lda based nmto method.@xcite coulomb interaction parameter @xmath23 was obtained from constrained lda method .
pseudogap amplitude @xmath31 was computed using lda+dmft scheme.@xcite supplementing conventional dmft self - energy by @xmath30 describes non - local dynamical correlations due to short ranged collective heisenberg like afm spin fluctuations . in this work we performed lda+dmft+@xmath4 calculations for hole - doped la@xmath58sr@xmath2cuo@xmath3 compound in the pseudogap regime . because of fluctuations of afm short - range order we clearly observe formation of the so called `` shadow bands '' as partially folded bare dispersion .
pseudogap is formed around ( @xmath6,0 ) point which is qualitatively the same as in bi2212@xcite , ncco@xcite and pcco@xcite . as to fermi surface of lsco
it is alike that obtained for bi2212@xcite .
namely the `` hot spots '' are not well resolved since the crossing point of the bare fermi surface and afm umklapp surface are very close to brillouin zone border .
this fact is essentially due to the shape and size of lda fermi surface . in this respect ,
situation here is different from that for ncco@xcite and pcco@xcite , where `` hot spots '' are clearly seen . to support these theoretical results we present here the new high energy , high resolution arpes data for lsco .
typical pseudogap - like effects of fermi surface destruction were observed in both theory and experiment .
the same is true for spectral functions .
the overall semiquantitative agreement between theory and experiments basically supports our general picture of the pseudogap state as due to strong scattering of carriers by short - range afm order fluctuations .
we thank thomas pruschke for providing us the nrg code .
this work is supported by rfbr grants 08 - 02 - 00021 , 08 - 02 - 91200 , ras programs `` quantum physics of condensed matter '' and `` strongly correlated electrons solids '' . in is supported by grants of president of mk-614.2009.2(in ) and russian science support foundation . as and
ss acknowledge jsps for the financial support for the japan - russia collaborating research in 2008 - 09 .
they also acknowledge mext , japan for the support in grant - in - aid for scientific research ( 18104007 , 15gs0123 , 18684015 ) .
t. timusk , b. statt , rep .
progr . , *
62 * , 61 ( 1999 ) ; m.v.sadovskii , usp .
nauk * 171 * 539 ( 2001 ) [ phys .
usp . * 44 * , 515 ( 2001 ) ] ; m.v.sadovskii , in `` strings , branes , lattices , networks , pseudogaps and dust '' , scientific world , moscow , 2007 , p. 357
( in russian ) , english version : cond - mat/0408489 . t. yoshida , x. j. zhou , k. tanaka , w. l. yang , z. hussain , z .- x .
shen , a. fujimori , s. sahrakorpi , m. lindroos , r. s. markiewicz , a. bansil , seiki komiya , yoichi ando , h. eisaki , t. kakeshita , and s. uchida , phys .
b * 74 * , 224510 ( 2006 ) .
e.z.kuchinskii , i.a.nekrasov , z.i.pchelkina , m.v.sadovskii , zh .
. fiz . * 131 * , 908 ( 2007 ) [ jetp * 104 * , 792 ( 2007 ) ] ; i.a.nekrasov , e.z.kuchinskii , z.v.pchelkina and m.v.sadovskii , physica c * 460 - 462 * , 997 ( 2007 ) .
e.e.kokorina , e.z.kuchinskii , i.a.nekrasov , z.v.pchelkina , m.v.sadovskii , a.sekiyama , s.suga , m.tsunekawa .
fiz . * 134 * , 968 ( 2008 ) [ jetp * 107 * , 818 ( 2008 ) ] ; i.a.nekrasov _ et al .
_ , j. phys .
solids * 69 * , 3269 ( 2008 ) .
k. held , i. a. nekrasov , g. keller , v. eyert , n. blumer , a. k. mcmahan , r. t. scalettar , th .
pruschke , v. i. anisimov , and d. vollhardt , psi - k newsletter * 56 * , 65 ( 2003 ) [ psi-k.dl.ac.uk/newsletters/news_56/highlight_56.pdf ] ; k. held , adv . phys . * 56 * , 829 ( 2007 ) i.a .
nekrasov , n.s .
pavlov , e.z .
kuchinskii , m.v .
sadovskii , z.v .
pchelkina , v.b .
zabolotnyy , j. geck , b.buchner , s.v .
borisenko , d.s .
inosov , a.a .
kordyuk , m. lambacher , a. erb , phys .
b * 80 * 115124(r ) ( 2009 ) .
k.g.wilson , rev .
phys . * 47 * , 773 ( 1975 ) ; h.r.krishna-murthy , j.w.wilkins , and k.g.wilson , phys .
b * 21 * , 1003 ( 1980 ) ; _ ibid . _ * 21 * , 1044 ( 1980 ) .
r.bulla , a.c.hewson and th.pruschke , j.phys.:condens.matter * 10 * , 8365 ( 1998 ) ; | slightly underdoped high - t@xmath0 system la@xmath1sr@xmath2cuo@xmath3 ( lsco ) is studied by means of high energy high resolution angular resolved photoemission spectroscopy ( arpes ) and combined computational scheme lda+dmft+@xmath4 .
corresponding one band hubbard model is solved via dynamical mean - field theory ( dmft ) , while model parameters needed are obtained from first principles within local density approximation ( lda ) .
an `` external '' @xmath5-dependent self - energy @xmath4 describes interaction of correlated electrons with antiferromagnetic ( afm ) pseudogap fluctuations .
experimental and theoretical data clearly show `` destruction '' of the lsco fermi surface in the vicinity of the ( @xmath6,0 ) point and formation of `` fermi arcs '' in the nodal directions .
arpes energy distribution curves ( edc ) as well as momentum distribution curves ( mdc ) demonstrate deviation of the quasiparticle band from the fermi level around ( @xmath6,0 ) point .
the same behavior of spectral functions follows from theoretical calculations suggesting afm origin of the pseudogap state . |
one of the reasons spinel oxides ab@xmath9o@xmath10 have been of interest is the various spin structures they show at low temperatures .
@xcite cubic spinels _
_ m__cr@xmath0o@xmath1 ( _ _ m__=mn , co ) have drawn much attention recently in relation with multiferroics since it was found that their spin configuration leads ferroelectric .
@xcite the magnetic phase of _ _ m__cr@xmath0o@xmath1 changes from paramagnetic to ferrimagnetic and from ferrimagnetic to spiral orders as the temperature decreases , even becoming glass - like at low temperatures .
the complexity in the spin structure of cubic spinels results from geometrical frustration among the spins of b site ions , which form the lattice of corner - sharing tetrahedra .
_ m _ ( cr ) ions occupy the a ( b ) sites that locate at the center of a tetrahedron ( octahedron ) in the oxygen lattice .
both the _ m _ ions and cr ions are magnetic , and their magnetization directions are opposed in the ordered states . the antiferromagnetic exchange interaction @xmath11 among the spins of the b site ions does not induce ordered states due to the aforementioned frustration .
@xcite however , the addition of the exchange interaction @xmath12 between the a and b site ions can give rise to a magnetic order . the relative strength of @xmath12 to @xmath11
is the key factor determining the state of the magnetic order , including the spin canting angle .
the nel configuration , a collinear ferrimagnet , is expected to be the ground state when @xmath11 is much weaker than @xmath12 . as @xmath11 increases , both the _ m _ and cr spins become canted and the magnetic structure becomes complex .
a classical theory predicts that the ground state of a cubic spinel is determined by the parameter @xmath13 , where @xmath14 and @xmath15 are the spin magnitudes at the a and b sites , respectively .
@xcite the nel configuration is the ground state when @xmath16 and the ferrimagnetic spiral is locally stable when @xmath17 .
it becomes unstable when @xmath18 .
an early neutron diffraction experiment showed that the ferrimagnetic spiral is the ground spin state of the cubic spinel mncr@xmath0o@xmath1 .
@xcite two different spin canting angle values were measured for the cr ions , while only one value was measured for the mn ions .
the cone angles of the mn ions , cr(i ) , and cr(ii ) in the ferrimagnetic spiral were @xmath19 , @xmath20 , and @xmath21 , respectively .
the approximate @xmath22 value estimated by these measurements is 1.6 , which predicts that the ferrimagnetic spiral is unstable .
in contrast , the cone angle of mn ion spins measured by nmr was @xmath23 or @xmath24 , while those of cr(i ) and cr(ii ) spins were @xmath25 and @xmath26 , respectively .
@xcite theory poorly matches these numbers , but it predicts a very unstable spiral state in general .
the question as to whether the ferrimagnetic spiral is the stable ground state in the cubic spinels was raised again by a recent neutron diffraction experiment involving @xmath27 .
@xcite it was reported that the ferrimagnetic state is long - range ordered for all temperatures below @xmath28 k , and that the spiral component appears in the plane orthogonal to the direction of the ferrimagnetic order below @xmath29 k. however , it is only short - range ordered .
the present study investigates the characteristics of the ordered spin state of @xmath30 at a low temperature by nuclear magnetic resonance ( nmr ) .
first , the cone angles of mn and cr ion spins , to which the previous nmr and neutron diffraction results gave different values , were measured .
information pertaining to accurate cone angles is important to understand not only the ground state of the spin order but also the electrical polarization .
@xcite we measured the nuclear spin - spin relaxation time @xmath31 as a function of the temperature to study the change in the spin fluctuation with temperature .
the result indicates that the fluctuation of the spiral component increases rapidly as the temperature approaches the phase transition temperature of the spiral order @xmath32 , above which the fluctuation is too fast for the spiral component to be measured by even local probes such as neutron diffraction or nmr .
the temperature - dependence of the nmr signal intensity reveals that the spiral phase is mixed with the ferrimagnetic phase at temperatures below @xmath32 .
a polycrystalline mncr@xmath0o@xmath1 sample was synthesized by a solid - state reaction from a molar ratio mixture of mno and cr@xmath0o@xmath33 powders .
the mixture was sintered at @xmath34c for 12 hours in an ar environment , for 12 additional hours at @xmath35c , and finally for 24 hours at @xmath36c .
a single - crystalline sample was grown by the flux method using a mixture of mnco@xmath33 , cr@xmath0o@xmath33 , pbf@xmath0 and pbo ( molar ratio = 1:1:2:2 ) .
nmr signals were obtained by the conventional spin echo method using a custom - made spectrometer in the temperature range of 4 to 20 k. to estimate the spin canting angle , the resonance frequency was measured for various magnetic fields up to 4 t. the nuclear spin - spin relaxation time , @xmath31 , was obtained by varying the time delay between the @xmath37 and @xmath38 pulses .
the @xmath39cr nmr spectrum was measured in the frequency range from 60 to 70 mhz , and the @xmath40mn nmr spectrum was assessed from 530 to 560 mhz . as the spectral width was very broad , the signal intensity was measured as a function of the frequency after selective excitation .
the magnetization versus the temperature curves show a discrepancy in the transition temperatures of the ferrimagnetic spiral and the collinear ferrimagnet of the polycrystalline and single - crystalline samples . in fig .
[ magnetizationcurve ] , the thick solid and dashed lines represent the field cooling ( fc ) and zero field cooling ( zfc ) m(t ) curves of the polycrystalline sample , respectively , and the thin solid and dotted lines represent those of the single crystal .
@xmath41 is relatively well defined by the abrupt increase in the magnetization in both samples of approximately 40 k for the polycrystalline sample and 50 k for the single - crystalline sample .
@xmath32 of the polycrystalline sample is more clearly defined by the abrupt decrease in the magnetization at 20 k compared to that of the single - crystalline sample , whose magnetization decreases smoothly at approximately 12 k only in the zfc case .
the m(t ) curves of our polycrystalline and single - crystalline samples are in good agreement with those in previous reports .
@xcite one of the reasons for the difference in the characteristics of the two samples is the site disorder in the single - crystalline sample .
x - ray absorption spectroscopy showed that the mn ions occupy only the a sites and the cr ions occupy the b sites in our polycrystalline sample , whereas both ions are found in both sites in the single - crystalline sample .
@xcite all of the experimental data described below were obtained from the polycrystalline sample .
figure [ zerofield ] shows the @xmath39cr nmr spectrum obtained in a zero field at 6.5 k for several different echo times .
the spectrum obtained at the echo time of @xmath42 shows a very well - defined single peak whose width is about 5 mhz .
the single peak centered around 67 mhz at a short echo time of @xmath42 appears to split into a double peak as the echo time supasses @xmath43 .
this is not a splitting but a suppression of the spectral intensity around the center due to the frequency - dependent nuclear spin - spin relaxation rate . in an ordered magnetic insulator containing a high concentration of identical magnetic nuclear spins , the suhl - nakmura ( sn ) interaction , in which nuclear spins are indirectly coupled by virtual magnons ,
is expected to play a major role in the nmr relaxation at low temperatures .
it is known that the sn interaction generates a field- and frequency - dependent @xmath31 with a minimum occurring in the center of the spectrum , as the majority of nuclear spins precess at this frequency .
@xcite the double peak feature of the spectrum obtained at @xmath43 disappears in the spectrum obtained at the same echo time , however , in a magnetic field , as denoted by the open circles in the figure .
this is consistent with the fact that the spin - spin relaxation rate due to the sn interaction decreases as the field increases .
the dependence of @xmath31 on the frequency and field confirms that the sn interaction is the main source of cr nuclear interaction in mncr@xmath0o@xmath1 .
this most likely explains why a double peak was observed in the previous nmr work , rather than the difference in the sample quality .
@xcite the difference of only 3 @xmath44 in the canting angles of the two cr spins associated with the two peaks in the previous nmr report supports this claim , because the experimental error of the canting angle as estimated by nmr is larger than this in general .
the @xmath40mn nmr spectrum showed a well - defined single peak centered around 550 mhz in a zero field at the liquid - helium temperature .
the spin canting angles of the mn@xmath45 and cr@xmath46 ions relative to the magnetization direction are determined by the shift of the spectrum with an external field .
the nmr resonance frequency @xmath47 is proportional to the magnitude of the total field , which is the vector sum of the hyperfine field @xmath48 and external field @xmath49 .
this is expressed as follows : @xmath50 where @xmath51 is the gyromagnetic ratio , and @xmath52 is the angle between @xmath48 and @xmath49 .
the direction of the hyperfine field is antiparallel to the local magnetization in most magnetic materials , providing the minus sign in the equation .
as the hyperfine fields of the mn@xmath45 and cr@xmath46 ions in mncr@xmath0o@xmath1 are more than one order of magnitude larger than the external magnetic field used in the experiment , the first - order approximation of the total field can be taken .
the slope of the frequency shift with external field is then determined as @xmath53 . in fig .
[ coneangle ] , the center frequencies of the cr and mn nmr spectra obtained at 4.2 k are plotted as a function of the external field . both of the frequencies change linearly in the experimental field range , as expected .
the resonance frequency of the mn spectrum decreases as the field increases , whereas that of the cr spectrum increases .
this visually shows that the spin directions of the mn and cr ions are opposite to each other , because the signs of the hyperfine fields of the ions are identical . from the slope of the linear fit to the data ,
the spin canting angles of the mn and cr ions were determined to be @xmath54 and @xmath55 , respectively .
this is contrary to the previous neutron diffraction or nmr experiments that reported two different values of spin canting angles for cr ions .
the canting angle of the mn spin is identical to that in one of the previous nmr reports @xcite but twice as large as the value of the neutron result @xcite .
the canting angle of the cr spin is consistent with the previous nmr measurement @xcite and one of the values given by the neutron diffraction @xcite . considering that no single value of @xmath22 can result in these cone angles
, the classical theory fails to explain the result .
however , both of the values corresponding to the mn and cr spin canting angles indicate that the ferrimagnetic spiral configuration is unstable in mncr@xmath0o@xmath1 .
this reminds us of the fact that a long range order of the ferrimagnetic component accompanies a short range order of the spiral component below @xmath7 in mncr@xmath0o@xmath1.@xcite figure [ t2 ] shows the nuclear spin - spin relaxation rate @xmath6 of cr ions at temperatures ranging from 6.5 k to 14.5 k. the relaxation rate increases relatively slowly with the temperature below 11 k , above which the slope becomes steep .
the main interaction a cr ion nucleus experiences is the interaction with the electron spins of magnetic ions and the prominent relaxation source is sn interaction that is mediated by spin wave as mentioned above .
the relaxation due to sn interaction is normally temperature independent .
therefore , the weakly temperature dependent relaxation with the rate of @xmath56 near 7 k should be mainly due to sn interaction .
the additional relaxation increasing with temperature indicates that the spin fluctuation becomes too large to be described in the framework of spin waves as the temperature approaches @xmath7 .
the previous observation that the spiral component of the spin order is unstable and short ordered implies that it is the spiral component of the cr ion spins that fluctuates .
this interpretation is also consistent with the saturation magnetization plotted together with the nuclear relaxation rate in fig .
the value of the saturation magnetization stays at @xmath57 independent of temperature even when the temperature crosses @xmath7 , where the spiral component is generated or vanishes .
this means that the component along the easy axis remains the same while the spiral component perpendicular to it is averaged out by fluctuation crossing @xmath7 , leaving only the ferrimagnetic order .
the nmr signal intensity is in general a function of the temperature , @xmath31 , and the number of nuclei .
the data points in fig .
[ domain ] , where the cr nmr signal intensity vs. the temperature is plotted , were obtained from the raw experimental data after temperature and @xmath31 correction .
thus , they represent the number of the nuclei producing the signal .
the signal intensity obtained while warming the sample stays constant while that obtained while cooling it changes .
it is worthwhile to note that the nmr signal is observed not in the ferrimagnetic phase but only in the spiral phase of mncr@xmath0o@xmath1 .
therefore , the corrected signal intensity in the figure is proportional to the volume of the ferrimagnetic spiral phase . upon warming ,
the entire volume of the mncr@xmath0o@xmath1 sample maintains its ferrimagnetic spiral phase until the change to the ferrimagnetic phase at @xmath7 . upon cooling , however , mncr@xmath0o@xmath1 remains in the ferrimagnetic phase well below @xmath7 . in the temperature region where the signal intensity changes , the two phases coexist .
depending on the temperature change history , the ferrimagnetic spiral phase is embedded in the matrix of the collinear ferrimagnet phase .
this mixed phase might have caused some error in the measurement of various quantities in the previous neutron diffraction and nmr experiments .
the temperature hysteresis in the volume of the ferrimagnetic spiral phase can be ascribed to a first - order transition at @xmath7 .
an esr work on the similar cubic spinel cocr@xmath9o@xmath10 showed an abrupt shift of the frequency at @xmath32 , also indicating a first - order transition .
@xcite the experimental evidence is in conflict to the second - order transition on which the classical theory is based .
the mn nmr signal intensity also showed a similar temperature hysteresis .
the spin canting angles of mn and cr ions and the nuclear spin - spin relaxation rates were measured in this study .
only one canting angle of cr spins was observed , contrary to that observed in previous neutron and nmr experiments .
the measured canting angles predict an unstable ferrimagnetic spiral state at a low temperature .
this instability is consistent with the measurement of the spin - spin relaxation rate .
the relaxation rate increases more rapidly as the temperature increases until it appears to diverge at @xmath7 .
the rapid increase in the relaxation rate near @xmath7 can be explained by the fluctuation of the spiral component . the nmr signal is not observed due to this strong relaxation near @xmath7 .
the nmr signal is unobservable above @xmath7 as well , where the magnetic phase is collinear ferrimagnetic .
the magnetization remains the same , crossing @xmath7 , indicating that the canted spins in the ferrimagnetic spiral phase do not line up along one direction , entering the ferrimagnetic phase ; instead , the fluctuation of the spiral component averages out to leave only the magnetic component along the easy axis .
the fluctuation accelerates as the temperature approaches @xmath7 , and past @xmath7 , it becomes fast enough to make the spiral component unobservable in neutron diffraction experiments but not fast enough to leave an averaged hyperfine field to nuclei in the time scale of nuclear spin precession , which is on the order of @xmath8 s. the temperature hysteresis of the spiral volume fraction indicates that the spiral and collinear ferrimagnetic phases are generally mixed below @xmath7 . | the canting angles and fluctuation of the magnetic ion spins of spinel oxide mncr@xmath0o@xmath1 were studied by nuclear magnetic resonance ( nmr ) at low temperatures , which has a collinear ferrimagnetic order below @xmath2 and a ferrimagnetic spiral order below @xmath3 .
contrary to previous reports , only one spin canting angle of cr ions was observed .
the spin canting angles of mn and cr ions in the ferrimagnetic spiral obtained at a liquid - he temperature were @xmath4 and @xmath5 , respectively .
the nuclear spin - spin relaxation was determined by the suhl - nakamura interaction at low temperatures but the relaxation rate @xmath6 increases rapidly as the temperature approaches @xmath7 .
this indicates that the fluctuation of the spiral component becomes faster as the temperature increases but not fast enough to leave an averaged hyperfine field to nuclei in the time scale of nuclear spin precession in the ferrimagnetic phase , which is on the order of @xmath8 s. the spiral volume fraction measured for various temperatures reveals that the collinear and the spiral ferrimagnetic phases are mixed below the transition temperature of the spiral order .
the temperature hysteresis in the volume fraction implies that this transition has first - order characteristics . |
although david rosen @xcite introduced as early as 1954 an infinite family of continued fractions which generalize the nearest - integer continued fraction , it is only very recently that the metrical properties of these so - called rosen fractions haven been investigated ; see e.g. @xcite , @xcite , @xcite and @xcite . in this paper
we will introduce @xmath0-rosen fractions , and study their metrical properties for special choices of @xmath0 .
these choices resemble nakada s @xmath0-expansions , in fact for @xmath1 these are nakada s @xmath0-expansions ; see also @xcite . to be more precise ,
let @xmath2 , @xmath3 , and @xmath4 .
then we define for @xmath5 $ ] the map @xmath6 \to [ \lambda ( \alpha -1 ) , \lambda\alpha)$ ] by @xmath7 and @xmath8 .
here , @xmath9 denotes the _ floor _ ( or _ entier _ ) of @xmath10 , i.e. , the greatest integer smaller than or equal to @xmath10 . in order to have positive digits
, we demand that @xmath11 . setting @xmath12 ( with @xmath13 ) , @xmath14 , and more generally @xmath15 for @xmath16 , one obtains for @xmath17 $ ] an expression of the form @xmath18 where @xmath19 and @xmath20 . setting @xmath21,\ ] ]
we will show in section [ conv ] that @xmath22 and for convenience we will write @xmath23.\ ] ] we call @xmath24 the @xmath25th @xmath0-rosen convergent of @xmath26 , and ( [ cf ] ) the @xmath0-rosen fraction of @xmath26 .
the case @xmath27 yields the rosen fractions , while the case @xmath28 is the rosen fraction equivalent of the classical regular continued fraction expansion ( rcf ) . in case @xmath1 ( and @xmath29 ) , the above defined @xmath0-rosen fractions are in fact nakada s @xmath0-expansions ( and the case @xmath30 is the rcf ) .
already from @xcite it is clear that in order to construct the underlying ergodic system for any @xmath0-rosen fraction and the planar natural extension for the associated interval map @xmath31 , it is fundamental to understand the orbit under @xmath31 of the two endpoints @xmath32 and @xmath33 of @xmath34 $ ] .
although the situation is in general more complicated than the ` classical case ' from @xcite , the natural extension together with the invariant measure can be given , and it is shown that this dynamical system is weakly bernoulli . using the natural extension , metrical properties of the @xmath0-rosen fractions will be given in section [ metrical properties ] .
in this section we find the `` smallest '' domain @xmath35 on which the map @xmath36 is bijective a.e .. we will deal with the general case , resembling nakada s @xmath0-expansions , i.e. , @xmath37 and @xmath38 for some fixed @xmath2 , @xmath39 ( the case @xmath1 is in fact the case of nakada s @xmath0-expansions ; see also @xcite ) . as in @xcite
, we need to discern between odd and even @xmath40 s , but some properties are shared by both cases , and these are collected here first . for @xmath41 $ ]
, setting @xmath42 it immediately follows from @xmath43 that @xmath44 , @xmath45 , and @xmath46 if @xmath47 . for a matrix @xmath48 , with @xmath49
, we define the corresponding mbius ( or fractional linear ) transformation by @xmath50 consequently , considering @xmath51 as a mbius transformation , we find that @xmath52.\ ] ] it follows that the numerators and denominators of the @xmath0-rosen convergents of @xmath26 from ( [ convergents ] ) satisfy the usual recurrence relations ( [ recurrence relations ] ) ; see also @xcite , p. 1279 .
furthermore , since @xmath53 we have that @xmath54 let @xmath55 be the left - endpoint of the interval on which the continued fraction map @xmath56 was defined in ( [ alfamap ] ) , @xmath57 its right - endpoint and let @xmath58\mid \varepsilon_1(x)=\varepsilon,\ d_1(x)=d\,\},\ ] ] be the cylinders of order @xmath59 of numbers with same first digits given by ( [ cf ] ) .
if we set @xmath60 for all @xmath61 , then the cylinders are given by the following table @xmath62 & ( \delta_1,r_0 ] \end{array}\ ] ] where we have used that @xmath63 since @xmath64 for @xmath39 .
note that we have by definition that @xmath65 for all @xmath66 , @xmath67 . setting @xmath68 , @xmath69 , @xmath70
, we have that @xmath71 in case @xmath72 , we write @xmath73 instead of @xmath74 , for @xmath75 . in @xcite
, it was shown that @xmath76 , & \text{if $ q=2p$}\\ \text{[}\ , ( -1\ , : \ , 1)^h , \ ,
-1\ , : \ , 2 , \ , ( -1 \ , : \ , 1 ) ^h \ , \text { ] } \ , , & \text{if $ q=2h+3 $ } , \end{cases}\ ] ] from which it immediately follows that @xmath77 and that for @xmath78 , @xmath79 @xmath80 , @xmath81 , and @xmath82 see also figure [ fig : phi ] . and the orbit of @xmath83 ( dashed broken line ) for @xmath84 ( left ) and @xmath85 ( right ) . ]
thus we see that the behavior of the orbit of @xmath86 is very different in the even case compared to the odd case ; see also figure [ fig : situationschets ] , where the relevant terms of @xmath87 , @xmath88 , and @xmath89 are displayed for even @xmath40 .
direct verification yields the following lemma .
[ lemma1 ] for @xmath39 , @xmath90 , we have that @xmath91 with @xmath92 if and only if @xmath72 , and @xmath93 if and
only if @xmath94 . in @xcite , the sequence @xmath87 plays a crucial role in the construction of the natural extension of the rosen fractions . due to the fact that the orbits of both @xmath83 and @xmath95 would become constant @xmath96 after a finite number of steps ( depending on @xmath40 ) , the natural extension of the rosen fraction could be easily constructed . in this paper ,
the @xmath88 and @xmath89 play a role comparable to that of the sequence @xmath87 ( even though the @xmath73 s are frequently used as well ) .
let @xmath97 $ ] be such , that @xmath98 for @xmath99 .
then it follows from ( [ recurrence relations ] ) that the @xmath0-rosen convergents of @xmath26 satisfy @xmath100 as in @xcite , we define the auxiliary sequence @xmath101 by @xmath102 this yields for @xmath103 that @xmath104 , @xmath105 , and @xmath106 by ( [ xvstn ] ) .
it follows that @xmath107.\ ] ] for @xmath108=-[(-1:1)^n,\ldots\,]\,$ ] , we obtain similarly @xmath109 , @xmath105 , thus @xmath110 and @xmath111\,.\ ] ] it is easy to see that @xmath112 , hence @xmath113 is a periodic sequence with period length @xmath114 .
let @xmath115 , @xmath116 , @xmath117 .
essential in the construction of the natural extension is the following theorem .
[ thm : evenorder ] let @xmath115 , @xmath116 , @xmath117 , and let the sequences @xmath88 and @xmath89 be defined as before . if @xmath118 , then we have that @xmath119 @xmath120 and @xmath121 .
if @xmath72 , then we have that @xmath122 if @xmath94 , then we have that @xmath123 if @xmath27 , then @xmath124 and @xmath125 , hence ( [ evenorder 1/2 ] ) is an immediate consequence of ( [ evenphi s ] ) . in general , in view of lemma [ lemma1 ] and the fact that @xmath126 $ ] , we have the following situation : @xmath127 and @xmath128 for @xmath129 , cf . figure [ fig : situationschets ] .
this yields that @xmath130.$ ] the relevant terms of @xmath87 , @xmath88 , and @xmath131 for even @xmath40 . ] since @xmath132 , we obtain @xmath133 by ( [ ln ] ) , we have therefore that @xmath134 with @xmath135 if and only if @xmath94 . for @xmath94
, we clearly have that @xmath136 , thus @xmath137 and ( [ evenorder 1/lambda ] ) is proved .
if @xmath118 , then we have that @xmath138 , hence @xmath139 $ ] , with @xmath140 , and again due to ( [ ln ] ) we obtain @xmath141 similarly , we have that @xmath142 $ ] and , by ( [ rn ] ) , @xmath143 thus ( [ evenorder ] ) is proved . since @xmath144 it follows from the definition of @xmath56 that @xmath121 and @xmath145 . with @xmath146 ,
the theorem is proved .
the structure of the @xmath74 s and @xmath147 s allows us to determine all possible sequences of `` digits '' .
for example , the longest consecutive sequence of digits @xmath148 contains @xmath149 terms if @xmath150 since @xmath138 and @xmath151 . in case @xmath94
, we only have @xmath152 .
in particular in case @xmath153 , @xmath94 , the cylinder @xmath154 is empty .
on the other hand , @xmath155 is always followed by @xmath152 since @xmath156 , with @xmath157 in case @xmath94 .
now we construct the domain @xmath158 upon which @xmath159 is bijective .
[ thm : even case nat ext ] let @xmath115 with @xmath160
. then the system of relations @xmath161 admits the ( unique ) solution @xmath162 in particular @xmath163 .
let @xmath118 and @xmath164 $ ] with @xmath165 , @xmath166 for @xmath167 , and @xmath168 .
then the map @xmath169 given by ( [ natural extension map ] ) is bijective off of a set of lebesgue measure zero .
it is easily seen that the solution of this system of relations is unique and valid , and that @xmath170 is injective .
we thus concern ourselves with the surjectivity of @xmath159 ; see also figure [ fig : nat ext q=6 ] . by ( [ evenorder ] )
, we have @xmath171 for @xmath172 , thus @xmath173 ) = j_n\times\left[\frac1\lambda,\frac1{\lambda - h_{n-2}}\right ] = j_n\times[h_2,h_n]\,,\ ] ] where we have used ( @xmath174 ) and ( @xmath175 ) .
furthermore , @xmath176 gives @xmath177 ) = [ \ell_{p-1},r_0)\times\left[\frac1\lambda,\frac1{\lambda - h_{2p-3 } } \right ] = j_{2p-1}\times[h_2,h_{2p-1}]\,.\ ] ] for @xmath178 , we have that @xmath179)=[\ell_0 , r_0)\times\left[\frac1{n\lambda},\frac1{n\lambda - h_{2p-3}}\right].\ ] ] the remaining part of the rectangle @xmath180 $ ] is mapped to @xmath181)= [ \ell_0,r_p ) \times \left[\frac1{d_p(r_0)\lambda},\frac1{d_p(r_0)\lambda - h_{2p-3}}\right].\ ] ] now consider the image of @xmath182 $ ] .
if @xmath183 , then it is split into @xmath184\times[0,h_{2p-1 } ] = [ \ell_0,r_p ) \times \left[\frac1{d_p(\ell_0)\lambda+h_{2p-1 } } , \frac1{d_p(\ell_0)\lambda}\right ] , \\
\mathcal t_\alpha((-\delta_n,-\delta_{n-1}]\times[0,h_{2p-1 } ] ) = [ \ell_0,r_0)\times\left[\frac1{n\lambda+h_{2p-1}},\frac1{n\lambda } \right]\ \text { for } n=2,3,\ldots , d_p(\ell_0)-1 , \\ \mathcal t_\alpha((\delta_1,r_0)\times[0,h_{2p-1 } ] ) = ( r_1,r_0)\times \left[\frac1{\lambda+h_{2p-1}},\frac1\lambda\right ] = ( r_1,r_0)\times[h_1,h_2],\end{gathered}\ ] ] where we have used ( @xmath185 ) . since @xmath186 and @xmath120 ,
the different parts of @xmath187)$ ] and @xmath188\times[0,h_{2p-1}])$ ] `` layer one under the other '' and `` fill up like a jig - saw puzzle '' @xmath189\right ) \cup \left([\ell_0,r_0)\times\left[\frac1{d_p(r_0)\lambda},h_1\right]\right).\ ] ] in case @xmath190 , we simply have @xmath191)= [ \ell_0,r_p)\times[1/(2\lambda),h_1]\ , , \\ \mathcal t_\alpha((\ell_{p-1},r_0)\times[0,h_{2p-1 } ] = ( r_1,r_p)\times[h_1,h_2]\,.\end{gathered}\ ] ] finally , the image of the central rectangle @xmath192 $ ] is split into @xmath193 ) = [ r_p , r_0 ) \times \left[\frac1{d_p(r_0)\lambda},\frac1{d_p(r_0)\lambda - h_{2p-2}}\right ] , \\ \mathcal t_\alpha([-\delta_{n-1},-\delta_n)\times[0,h_{2p-2 } ] ) = [ \ell_0,r_0)\times\left[\frac1{n\lambda},\frac1{n\lambda - h_{2p-2 } } \right]\ \text { for } n > d_p(r_0 ) , \\
\mathcal t_\alpha((\delta_n,\delta_{n-1}]\times[0,h_{2p-2 } ] ) = [ \ell_0,r_0)\times\left[\frac1{n\lambda+h_{2p-2}},\frac1{n\lambda } \right]\ \text { for } n > d_p(\ell_0 ) , \\ \mathcal t_\alpha((\delta_{d_p(\ell_0)},\ell_{p-1}]\times[0,h_{2p-2 } ] ) = [ r_p , r_0 ) \times \left[\frac1{d_p(\ell_0)\lambda+h_{2p-2 } } , \frac1{d_p(\ell_0)\lambda}\right].\end{gathered}\ ] ] since @xmath194 and @xmath120 , the union of these images is @xmath195\right ) \cup \left([r_p , r_0)\times \left[\frac1{d_p(r_0)\lambda},\frac1{d_p(\ell_0)\lambda}\right]\right).\ ] ] therefore @xmath196 and @xmath158 differ only by a set of lebesgue measure zero . the natural extension domain @xmath158 ( left ) and its image under @xmath170 ( right ) of the @xmath0-rosen continued fraction ( @xmath197 ) ; here @xmath198 , @xmath199 , @xmath200 , @xmath201 . ] if @xmath72 , then the intervals @xmath202 are empty and @xmath203 .
the proof of theorem [ thm : even case nat ext ] remains valid , with @xmath204 ; see also @xcite . since @xmath205 for @xmath206
, we have @xmath207\right ) \cup \left(\left[0,\frac\lambda2\right)\times[0,1]\right).\ ] ] for @xmath94 , we just have the intervals @xmath202 , @xmath208 and add @xmath209 .
furthermore , we have that @xmath210 for @xmath167 and @xmath211 .
this provides the following theorem .
[ thm : even case nat ext 1/lambda ] let @xmath115 with @xmath160 and @xmath212.\ ] ] then @xmath213 is bijective off of a set of lebesgue measure zero . by ( [ evenorder 1/lambda ] )
, we have that @xmath214 , thus @xmath215\right ) \\ = \left[\frac{b_{n+1}-b_{n+2}}{b_{n+1}-b_n } , \frac{b_{n+2}-b_{n+3}}{b_{n+2}-b_{n+1}}\right)\times \left[\frac1\lambda,\frac{b_{n+1}}{b_{n+2}}\right ]
\quad\text{for } n=1,2,\ldots , p-2.\end{gathered}\ ] ] the different parts of @xmath216 are mapped to @xmath217\right ) & = [ 1-\lambda,1)\times\left[\frac1{n\lambda},\frac2{(2n-1)\lambda}\right]\ \text { for } n=2,3,\ldots \\
\mathcal t_{1/\lambda } \left((\delta_n,\delta_{n-1}]\times\left[0,\frac\lambda2\right]\right ) & = [ 1-\lambda,1)\times\left[\frac2{(2n+1)\lambda},\frac1{n\lambda}\right]\ \text { for } n=2,3,\ldots \\
\mathcal t_{1/\lambda } \left((\delta_1,1)\times\left[0,\frac\lambda2\right]\right ) & = [ 1-\lambda,1)\times\left[\frac2{3\lambda},\frac1\lambda\right]\end{aligned}\ ] ] and the union of these images is @xmath218 $ ] .
note that there is a simple relation between @xmath219 and @xmath220 , which will be useful in section [ mixing properties ] ; reflect @xmath219 in the line @xmath221 in case @xmath222 , and reflect @xmath219 in the line @xmath223 in case @xmath224 , to find @xmath220 ; see also figure [ fig : nat ext 1/2,1/lambda ] .
@xmath219 ( left ) and @xmath220 ( right ) ; here @xmath198 . ] as in @xcite , a jacobian calculation shows that @xmath159 preserves the probability measure @xmath225 with density @xmath226 where @xmath227 is a normalizing constant . for the calculation of this constant ,
we need the following lemma .
[ lem : b - identity ] if @xmath228 , then we have that @xmath229 with @xmath230 , we have that @xmath231 [ lem : norm const even ] for @xmath90 , the normalizing constant is @xmath232 similarly to @xcite , integration of the density over @xmath158 gives @xmath233 for @xmath118 , by theorem [ thm : even case nat ext ] . using ( [ ln ] ) , ( [ rn ] ) and lemma [ lem : b - identity ] , we find @xmath234 for @xmath167 , and @xmath235 putting everything together , we obtain @xmath236 for @xmath72 , we have the same constant by the remark following theorem [ thm : even case nat ext ] and by @xcite .
finally , the remark following theorem [ thm : even case nat ext 1/lambda ] shows that @xmath237 is the same constant as well .
let @xmath238 be the projection of @xmath239 on the first coordinate , let @xmath240 be the restriction of the two - dimensional @xmath241-algebra on @xmath242 , and @xmath243 be the lebesgue @xmath241-algebra on @xmath244 $ ] . in @xcite , rohlin introduced and studied the concept of _ natural extension _ of a dynamical system . in our setting , a natural extension of @xmath245
is an invertible dynamical system @xmath246 , which contains @xmath245 as a factor , such that the borel @xmath241-algebra @xmath247 of @xmath248 is the smallest @xmath249-invariant @xmath241-algebra that contains @xmath250 , where @xmath251 is the factor map .
a natural extension is unique up to isomorphism .
we have the following theorem .
[ thm : natural - extension - even - case ] let @xmath39 , @xmath115 , and let @xmath252 . then the dynamical system @xmath253 is the natural extension of the dynamical system @xmath254 . in case @xmath255 and @xmath256 the proof is a straightforward application of theorem 21.2.2 from @xcite ; see also examples 21.3.1 ( the case of the rcf ) and 21.3.2 ( the nicf ) in @xcite .
however , for @xmath257 , some extra work is needed . to be more precise , let us recall some terminology from @xcite .
let @xmath258 be a set , and @xmath259 be a map .
the pair @xmath260 is called a _ fibred system _ if the following three conditions are satisfied : * there is a finite or countable set @xmath261 ( called the digit set ) ; * there is a map @xmath262 . then the sets @xmath263 form a partition of @xmath258 ; * the restriction of @xmath258 to any @xmath264 is an injective map
; see @xcite , definition 1.1.1 .
clearly , in our setting @xmath265 $ ] , @xmath266 , @xmath267 , and @xmath268 .
the pair @xmath269 is called a _ dual fibred system _ ( or _ backward algorithm _ ) with respect to @xmath260 if the following condition holds : @xmath270 is an admissible block of digits for @xmath271 if and only if @xmath272 is admissible for @xmath273 ; see definition 21.1.1 in @xcite . furthermore , @xmath274 see definition 21.1.7 from @xcite .
the local inverse of the map @xmath275 is denoted by @xmath276 .
f. schweiger obtained the following theorem ; see @xcite , theorem 21.2.1 .
consider the following dynamical system @xmath277 , with @xmath278 , and where @xmath279 is defined by @xmath280 if @xmath281 is measurable , then @xmath277 equipped with the obvious product @xmath241-algebra is an invertible dynamical system with invariant density @xmath282 .
this dynamical system @xmath277 is the natural extension of @xmath260 . in our setting
we have by construction that @xmath283 .
so to apply schweiger s theorem , we need to find a backward algorithm @xmath269 , such that @xmath284 from our construction of @xmath242 , and in particular from the proof of theorem [ thm : even case nat ext ] , we see that a natural choice of @xmath285 seems . ] to be the interval @xmath286 $ ] , with partition elements @xmath287=\left [ \frac{1}{\lambda } , 1\right ] , \qquad \delta^{\ # } ( 1:1)=[h_1,h_2]=\left [ \frac{1}{\lambda + 1},\frac{1}{\lambda}\right ] , \ ] ] and for @xmath288 , @xmath289 , @xmath290 , \qquad \delta^{\ # } ( 1:d)=\left [ \frac{1}{d\lambda + h_{2p-1}},\frac{1}{d\lambda}\right ] .\ ] ] on these intervals @xmath291 we define the map @xmath273 as @xmath292 an easy calculation shows that @xmath293 satisfies ( [ vshap - map ] ) . for @xmath294 , the rectangle @xmath295
$ ] is partitioned as in figure [ blow - up ] ( see also theorem [ thm : evenorder ] , the proof of theorem [ thm : even case nat ext ] , and figure [ fig : nat ext q=6 ] , for the case @xmath198 , @xmath199 ) . for even @xmath40 , @xmath118 . ]
note that this case does not occur in case @xmath255 or @xmath296 ; see also examples 21.3 in @xcite .
recall , that @xmath163 , @xmath297 , and @xmath298 ; see theorem [ thm : even case nat ext ] . we are left to define @xmath273 on the interval @xmath299 $ ] .
clearly , on the interval @xmath300 $ ] we should define @xmath273 by @xmath301 while on @xmath302=\left [ \frac{1}{d\lambda + \lambda /2},\frac{1}{d\lambda}\right]$ ] , the map @xmath273 should be given by @xmath303 this leaves us how to define @xmath273 on @xmath304 $ ] .
the problem here is , that @xmath305 s in this interval have``two names '' ; see also figure [ blow - up ] . in order to overcome this problem , consider the following isomorphic copy @xmath306 of @xmath307 ; let @xmath308\times \left[\frac{1}{d\lambda + 1 } , \frac{1}{d\lambda + \lambda /2 } \right ] \right ) \cup \left ( [ \ell_p , r_0]\times \left [ \frac{-1}{d\lambda + \lambda /2 } , \frac{-1}{d\lambda + 1}\right]\right ) , \ ] ] let @xmath309 be defined by @xmath310 where @xmath311\times \left [ \frac{1}{d\lambda + 1 } , \frac{1}{d\lambda + \lambda /2}\right ] \right ) \\ ( x ,- y ) & \text{if } ( x , y)\in ( \ell_p , r_0]\times \left [ \frac{1}{d\lambda + 1 } , \frac{1}{d\lambda + \lambda /2 } \right ] , \end{cases}\ ] ] and set @xmath312 setting @xmath313\cup \left ( ( \ell_p , r_0]\times \left [ \frac{-1}{d\lambda + \lambda /2 } , \frac{-1}{d\lambda + 1 } \right]\right ) , \ ] ] and defining @xmath273 on @xmath314 $ ] by @xmath315 and @xmath273 on @xmath316 $ ] by @xmath317 \\ \frac{-1}{x}-(d+1)\lambda & \text{if } \frac{1}{x}+(d+1)\lambda
\in \left [ \frac{1}{d\lambda + 1 } , \frac{1}{d\lambda + \lambda /2}\right ] , \end{cases}\ ] ] we find that @xmath269 is the dual fibred system with respect to @xmath318 .
we already saw that the density @xmath282 of the invariant measure is given by @xmath319 see proposition [ lem : norm const even ] .
thus it follows from schweiger s theorem that @xmath320 is the natural extension of @xmath321 .
let @xmath78 , @xmath322 . the @xmath74 s and @xmath147 s
are ordered in the following way .
[ thm : oddorder ] let @xmath78 , @xmath322 , let the sequences @xmath323 and @xmath89 be defined as above , and let @xmath324 .
then we have the following cases : @xmath325 in @xcite , section 3.2 ( see also the introduction of this section ) , it was shown that @xmath326 @xmath80 , and @xmath327 in view of this and lemma [ lemma1 ] , we therefore have that @xmath328 .
an important question is to know , where @xmath329 is located . since @xmath330
, we have @xmath331 . for @xmath78
, we have @xmath332 , thus @xmath333 hence we obtain , by ( [ ln ] ) , @xmath334 the position of @xmath335 with respect to @xmath329 leads us to distinguish between the possible cases .
we have that @xmath336 if and only if @xmath337 , i.e. , @xmath338 .
note that @xmath339 assume first that @xmath340 .
then we have that @xmath341 , from which it immediately follows that @xmath342 if @xmath150 , then @xmath343 , hence @xmath344 and @xmath345 . in case @xmath94 , then @xmath346 .
hence the last two cases are proved .
it remains to consider @xmath347 .
now we have that @xmath348 , with @xmath349 if and only if @xmath350 .
consequently , we immediately find that @xmath351\,.\ ] ] to see that @xmath352 , note that this is equivalent to @xmath353 which holds because of the assumption @xmath354 .
this assumption also implies that @xmath355 , where again equality holds if and only if @xmath350 .
this proves the case @xmath350 . for @xmath356
, we have @xmath357\,,\ ] ] hence the convergents of @xmath358 satisfy @xmath359 , @xmath360 and @xmath361 the recurrence @xmath362 for @xmath363 , yields @xmath364 for the @xmath365 s , we have similarly that @xmath366 , @xmath367 , thus @xmath368 and @xmath369 by ( [ xvstn ] ) , we obtain , for @xmath370 , @xmath371 for the convergents of @xmath372 , only the sign of the @xmath373 is different and we get @xmath374 this yields that @xmath375 hence @xmath376 , @xmath377 , and the theorem is proved . for the construction of the natural extension , we have to distinguish between the different cases of the previous theorem .
consider first @xmath340 .
[ thm : nat ext odd1 ] let @xmath78 with @xmath378 .
then the system of relations @xmath379 admits the ( unique ) solution @xmath380 in particular @xmath381 .
let @xmath382 and @xmath383 $ ] with @xmath165 , @xmath166 for @xmath363 , @xmath384 and @xmath385 .
then the map @xmath386 given by ( [ natural extension map ] ) is bijective off of a set of lebesgue measure zero .
the case @xmath1 , @xmath387 , which is the case of nakada s @xmath0-expansions for @xmath388 , has been dealt with in @xcite ; see also @xcite , @xcite , and @xcite .
the proof of theorem [ thm : nat ext odd1 ] is very similar to that of theorem [ thm : even case nat ext ] and therefore omitted , see also figure [ fig : nat ext q=5 case1 ] . in case @xmath94 , the intervals @xmath389 are empty .
( left ) and its image under @xmath159 ( right ) of the @xmath0-rosen continued fraction ( @xmath197 ) ; here @xmath390 , @xmath391 , @xmath392 , @xmath393 . ] once more , a jacobian calculation shows that @xmath170 preserves the probability measure @xmath225 with density @xmath226 where @xmath227 is a normalizing constant given by the following proposition .
[ lem : norm const odd1 ] if @xmath78 and @xmath382 , then the normalizing constant is @xmath394 integration gives @xmath395 using ( [ ln ] ) and ( [ rn ] ) , we find @xmath396 and @xmath397 where we have used @xmath398 putting everything together , we obtain @xmath399 note that for @xmath1 this result confirms nakada s result from @xcite for @xmath0 between @xmath400 and @xmath59 ; in this case , the normalizing constant is indeed @xmath401 .
now consider @xmath356 .
[ thm : nat ext odd2 ] let @xmath78 with @xmath378 .
then the system of relations @xmath402 admits the ( unique ) solution @xmath403 in particular @xmath404 .
let @xmath405 , and @xmath406 $ ] with @xmath407 , @xmath408 , @xmath409 , @xmath410 for @xmath363 , @xmath411 , @xmath412 and @xmath413 .
then the map @xmath386 given by ( [ natural extension map ] ) is bijective off of a set of lebesgue measure zero .
the proof of the bijectivity runs along the same lines as the proof of theorem [ thm : even case nat ext ] and is therefore omitted , see also figure [ fig : nat ext q=5 case2 ] .
the @xmath414 s are determined by @xmath415 .
the @xmath416 s are determined by @xmath417 and @xmath418 . by @xmath419
, we obtain @xmath420 and @xmath421 thus @xmath422 , i.e. @xmath404 .
finally , the @xmath423 s are determined by @xmath424 .
for @xmath72 , the intervals @xmath425 and @xmath426 are empty .
( left ) and its image under @xmath159 ( right ) of the @xmath0-rosen continued fraction ( @xmath197 ) ; here @xmath390 , @xmath427 , @xmath428 , @xmath429 . ] again , @xmath170 preserves the probability measure @xmath225 with density @xmath430 , where @xmath227 is a normalizing constant given by the following proposition .
[ lem : norm const odd2 ] if @xmath431 and @xmath405 , then the normalizing constant is @xmath432 integration yields that @xmath227 is equal to @xmath433 using ( [ ln ] ) , ( [ rn ] ) , ( [ lnh ] ) , ( [ rnh ] ) and lemma [ lem : b - identity ] , we find @xmath434 and @xmath435 putting everything together , we obtain that @xmath436 this confirms again nakada s result for @xmath1 , i.e. , @xmath437 for @xmath438 .
the case @xmath350 is slightly different from both other cases ( similarly to @xmath94 for even @xmath40 ) .
let @xmath78 with @xmath378 , @xmath350 and @xmath439\right ) \cup \left([r_j,\ell_j)\times\left[0,\frac{b_j}{b_{j+1}}\right]\right ) \cup \left([\ell_h,\rho)\times\left[0,\frac\lambda2\right]\right).\ ] ] then @xmath440 is bijective off of a set of lebesgue measure zero .
the normalizing constant in this case is @xmath441 as above . as in the even case
, we set @xmath238 the projection of @xmath239 on the first coordinate , @xmath240 be the restriction of the two - dimensional @xmath241-algebra on @xmath242 , and @xmath243 be the lebesgue @xmath241-algebra on @xmath244 $ ] .
we have the following theorem , whose proof is similar to the proof of theorem refthm : natural - extension - even - case in the even case .
[ thm : natural - extension - odd - case ] let @xmath442 , @xmath443 , and let @xmath252 .
then the dynamical system @xmath253 is the natural extension of the dynamical system @xmath444 .
( left ) , @xmath445 ( middle ) and @xmath220 ( right ) ; here @xmath390 . ]
now we can prove easily that the @xmath0-rosen continued fractions converge .
if @xmath446 for some @xmath447 , then this is clear . therefore assume that @xmath448 for all @xmath447 . setting @xmath449
, it follows directly from the definition ( [ natural extension map ] ) of @xmath450 that @xmath451 $ ] .
furthermore , an immediate consequence of ( [ recurrence relations ] ) is that @xmath452 $ ] , i.e. , @xmath453 .
theorems [ thm : even case nat ext ] , [ thm : nat ext odd1 ] and [ thm : nat ext odd2 ] ( see also figures [ fig : nat ext q=6 ] , [ fig : nat ext q=5 case1 ] and [ fig : nat ext q=5 case2 ] ) show that @xmath454 , i.e. , @xmath455 , and that @xmath456 if and only if @xmath78 , @xmath457 , @xmath458 , @xmath459 for @xmath460 , which is possible only if @xmath340 .
furthermore we have that @xmath461 for some constant @xmath462 , i.e. , @xmath463 .
it follows from ( [ xvstn ] ) that @xmath464 hence the @xmath0-rosen convergents @xmath24 converge to @xmath26 as @xmath465 .
in case @xmath40 is even , and @xmath94 , we saw in the previous section that there is a simple relation between @xmath219 and @xmath220 ; see also figure [ fig : nat ext 1/2,1/lambda ] .
define in this case the map @xmath466 by @xmath467 clearly , @xmath466 is bijective and bi - measurable transformation , and @xmath468 , for every borel set @xmath469 . by comparing the partitions of @xmath470 ( on @xmath220 ) and that of @xmath471 ( on @xmath219 ) , we find that @xmath472 this implies that the dynamical systems @xmath473 and @xmath474 are isomorphic . in
@xcite it was shown that the dynamical system system @xmath475 is weakly bernoulli with respect to the natural partition , hence @xmath475 and @xmath473 are isomorphic . as a consequence
we find that the dynamical systems @xmath476 and @xmath474 are isomorphic . in this section
, we will show that this result also holds for all @xmath40 and all @xmath0 strictly between @xmath477 and @xmath478 , using a result by m. rychlik @xcite .
for completeness , we state explicitly the hypothesis needed for rychlik s result ( the reader is referred to @xcite for more details ) .
let @xmath479 be a totally ordered order - complete set .
open intervals constitute a base of a complete topology in @xmath479 , making @xmath479 into a topological space . if @xmath479 is separable , then @xmath479 is homeomorphic with a closed subset of an interval .
let @xmath480 be the borel @xmath241-algebra on @xmath479 , and @xmath481 a fixed regular , borel probability measure on @xmath479 ( in our case @xmath481 will be the normalized lebesgue measure restricted to @xmath479 ) .
let @xmath482 be an open dense subset of @xmath479 , such that @xmath483 .
let @xmath484 , clearly @xmath485 .
let @xmath486 be a continuous map , and @xmath487 a countable family of closed intervals with disjoint interiors , such that @xmath488 . furthermore , suppose that for any @xmath489 one has that @xmath490 consists only of endpoints of @xmath258 , and that @xmath271 restricted to @xmath491 admits an extension to a homeomorphism of @xmath258 with some interval in @xmath479 .
suppose that @xmath492 for @xmath493 , and let @xmath494 for @xmath495 , @xmath496 .
let @xmath497 be the perron - frobenius operator of @xmath271 , @xmath498 in @xcite , it was proved ( among many other things ) that , if @xmath499 and @xmath500 , then there exist functions @xmath501 of bounded variation , such that * @xmath502 ; * @xmath503 ; * there exists a measurable partition @xmath504 of @xmath479 with @xmath505 for @xmath506 ; * the dynamical system @xmath507 , where @xmath508 and @xmath509 are exact , and @xmath510 is the unique invariant measure for @xmath511 , absolutely continuous with respect to @xmath512 .
rychlik also showed that if @xmath513 , i.e. , if @xmath59 is the only eigenvalue of @xmath514 on the unit circle and if there exists only one @xmath515 with @xmath516 and @xmath517 , ( @xmath518 ) , then the natural extension of @xmath519 is isomorphic to a bernoulli shift .
returning to our map @xmath31 , defined on @xmath520 $ ] , and using the same notation as above , we let @xmath481 be normalized lebesgue measure on @xmath479 , @xmath521 and @xmath522 .
note that @xmath523 is continuous , and that the restriction of @xmath56 to each open interval is homeomorphic to an interval ( in fact to @xmath479 itself , except for the first and last interval ) .
we have that @xmath524 on @xmath525 , hence @xmath499 ( since @xmath526 ) , and @xmath500 .
it is easy , but tedious ( cf .
@xcite for a proof of the regular case ) , to see that @xmath271 is ergodic , hence @xmath513 , and we can apply rychlik s result to obtain the following theorem .
[ thmbernoulli ] the natural extension @xmath527 of @xmath528 is weakly bernoulli .
hence , the natural extension is isomorphic to any bernoulli shift with the same entropy .
an important reason to introduce and study the natural extension of the ergodic system underlying any continued fraction expansion , is that such a natural extension facilitates the study of the continued fraction expansion at hand ; see e.g. @xcite and @xcite , chapter 4 .
the following theorem is a consequence of this ; see @xcite , @xcite , or @xcite , chapter 4 .
[ distribution - orbit ] let @xmath442 , and let @xmath29 . for almost all @xmath529-irrational numbers @xmath26 , the two - dimensional sequence @xmath530 , @xmath16 , is distributed over @xmath242 according to the density function @xmath531 , given by @xmath532 for @xmath533 , and @xmath534 otherwise . here
@xmath227 is the normalizing constant of the @xmath535-invariant measure @xmath239 .
due to proposition [ distribution - orbit ] , it is possible to study the distribution of various sequences related to the @xmath0-rosen expansion of almost every @xmath536 .
classical examples of these are the frequency of digits , or the analogs of various classical results by lvy and khintchine
. however , these results can already be obtained from the projection @xmath537 of @xmath538 which is also ergodic and the ergodic theorem . for the distribution of the so - called _ approximation coefficients _
, the natural extension @xmath539 is necessary .
these approximation coefficients @xmath540 , are defined by @xmath541 where @xmath24 is the @xmath25th @xmath0-rosen convergent , which is obtained by truncating the @xmath0-rosen expansion . in view of ( [ regular - thetas2 ] ) and ( [ regular - thetas3 ] )
, we define the map @xmath546 it is now easy calculation , see e.g. @xcite , p. 1293
, that due to proposition [ distribution - orbit ] one has for almost all @xmath536 that the sequence @xmath547 is distributed on @xmath548 according to the density function @xmath549 . setting @xmath550 we have found the following theorem . [ twotheta s ]
let @xmath442 , @xmath551 , and define the functions @xmath552 and @xmath553 by @xmath554 and @xmath555 otherwise .
then the sequence @xmath556 lies in the interior of @xmath557 for all @xmath529-irrational numbers @xmath26 , and for almost all @xmath26 this sequence is distributed according to the density function @xmath558 , where @xmath559 by this last statement we mean that , for almost all @xmath26 and for all @xmath560 , the limit @xmath561 exists , and equals @xmath562 several corollaries can be drawn from theorem [ twotheta s ] , see e.g. [ k1 ] , where ( for @xmath1 ) for almost all @xmath26 the distributions of the sequences @xmath563 , @xmath564 , @xmath565 were determined . in view of the expression of @xmath569 in ( [ regular - thetas2 ] ) , we consider curves given by @xmath570 where @xmath571 is a constant , and @xmath572 $ ] .
note that these curves are monotonically increasing on @xmath573 $ ] , and that the curve given by @xmath574 lies `` above '' the curve given by @xmath575 , if and only if @xmath576 .
now let @xmath577 be defined as the positive largest @xmath578 for which the curve @xmath579 lies in @xmath242 for @xmath572 $ ] , i.e. , @xmath580\right\ } .\ ] ] it follows from theorem [ distribution - orbit ] that for all @xmath581 , and for almost all @xmath529-irrationals @xmath26 one has that @xmath582 where @xmath227 is the normalizing constant of the invariant measure ( which has density @xmath531 ) .
so we are left to show that @xmath583 . in case
@xmath118 , first note that the curve @xmath585 goes through @xmath586 if and only if @xmath587 .
since in this case @xmath588 and the curve @xmath585 is monotonically increasing on @xmath573 $ ] , we immediately find that this curve is in @xmath242 for @xmath589 $ ] , yielding that @xmath590 . from theorem [ thm : evenorder ]
we see that @xmath591 .
setting @xmath592 it follows from theorem [ thm : even case nat ext ] ( see also figure [ fig : nat ext q=6 ] for @xmath198 ) that @xmath593 , since @xmath594 . for @xmath40 ( and therefore @xmath595 )
fixed , and for @xmath596 $ ] , one easily shows that @xmath597 is a monotonically increasing function of @xmath0 , with @xmath598 , and @xmath599 , while @xmath600 is a line with slope @xmath601 . since @xmath602 , and @xmath603
, we find for @xmath604 that @xmath605 in case @xmath256 , the point @xmath606 yields @xmath607 .
since the curve @xmath579 is monotonically increasing on @xmath573 $ ] , and from the fact that for @xmath608 we have that @xmath609 , where @xmath478 is the `` smallest height '' of @xmath242 , we find that @xmath610 this proves the theorem . | the rosen fractions form an infinite family which generalizes the nearest - integer continued fractions . in this paper
we introduce a new class of continued fractions related to the rosen fractions , the @xmath0-rosen fractions .
the metrical properties of these @xmath0-rosen fractions are studied .
we find planar natural extensions for the associated interval maps , and show that their domains of definition are closely related to the domains of the ` classical ' rosen fractions .
this unifies and generalizes results of diophantine approximation from the literature . |
in a model based on supersymmetry ( susy ) , slepton mixing is a source of lepton flavor violation ( lfv ) .
two types of effective lfv couplings are induced at low energies ; i.e. , those mediated by the neutral gauge bosons and those by the neutral higgs bosons .
in contrast to the gauge boson mediation , the contributions from the higgs - mediation do not decouple even if the soft susy breaking scale is as large as @xmath6 tev @xcite . to study the higgs - mediated lfv couplings the tau - associated processes are useful because they are proportional to the mass of the relevant charged leptons .
the lfv couplings associated with a tau lepton have been measured at @xmath7 factories by searches for rare tau decays , such as @xmath8 , @xmath9 , @xmath10 , @xmath11 , etc . at future collider experiments ,
they are directly tested via the decays of the higgs bosons ( @xmath12 ) , @xmath13 @xcite . in this talk , we discuss a reaction at the deep inelastic scattering ( dis ) region with high - intensity and high - energy muon beams as an alternative approach of searching for the lfv couplings associated with a tau lepton . here , @xmath14 is a target nucleon , and @xmath15 represents all final state particles .
sher and turan have discussed this process in a model independent approach @xcite . instead , we here consider this process in the framework of susy @xcite .
when the scalar lfv coupling of @xmath16 is independent of the other types of couplings , its experimental constraint comes from the @xmath17 result . the total cross section of the process mediated by the scalar lfv coupling
could then be as large as 0.5 fb at muon energy ( @xmath0 ) of 50 gev .
for this case , with @xmath3 muons per year on a @xmath4 g/@xmath18 target mass , about @xmath19 of the events can be produced , or no observation of the signal would improve the limits by six orders of magnitude @xcite . dis process , assuming the effective scalar and pseudo - scalar couplings @xcite and the mssm higgs coupling constrained from the current data @xcite .
cteq6l is used for the pdf .
, width=241 ] in the minimal supersymmetric standard model ( mssm ) , the higgs boson couplings are related each other . in the decoupling region ( @xmath20 gev ) , where @xmath21 and @xmath22 , the scalar coupling @xmath23 is nearly equal to the pseudo - scalar coupling @xmath24 .
therefore , both couplings are determined by the more constrained one , namely the pseudo - scalar coupling @xcite .
it is constrained by the @xmath25 decay ( @xmath26 ) @xcite . in fig .
[ fig : tot ] , the total cross sections are shown for the cases of the effective scalar and pseudo - scalar couplings as well as the susy model with the maximal value for the couplings under the current data of the rare tau decays .
the largest values of @xmath27 and @xmath28 can be realized with @xmath29 tev and the higgsino mass @xmath30 tev .
it should be noted that in such a situation , the gauge boson mediated couplings are strongly suppressed . we evaluate the cross sections of the reaction in the dis region for the higgs - mediated interaction with the maximally allowed values of the effective couplings for each quark contribution : see fig.[fig : total - cross - section - vs - emu ] .
the cross section sharply increases above @xmath31 gev .
this enhancement comes from a consequence of the @xmath1-quark contribution in addition to the @xmath32 and @xmath33-quark contributions .
the cross section is enhanced by one order of magnitude when the muon energy changes from 50 gev to 100 gev .
typically , for @xmath34 gev and @xmath35 gev , the cross section is @xmath36 fb and @xmath37 fb , respectively .
dis process as a function of the muon energy for the higgs mediated interaction .
it is assumed that the initial muons are purely left - handed .
cteq6l is used for the pdf .
, width=226 ] next , we study cases where the gauge - boson mediation is dominant ; i.e. , @xmath38 gev . since br@xmath39 @xcite , the contribution from the tensor interaction is found to be smaller than that from the higgs boson mediation by about five orders of magnitude . on the other hand ,
the vector and axial - vector interactions are suppressed at the same level as the pseudo - scalar interaction @xcite .
therefore , their contributions can be as large as those for the higgs boson mediation , if @xmath40 is less than 50 gev @xcite . at higher energies , the cross section for the gauge boson mediation becomes much smaller than that for the higgs boson mediation because of no enhancement by the @xmath1-quark sub - process .
[ fig : angle - dependence ]
with the intensity of @xmath3 muons per year and the target mass of 100 g/@xmath18 , about @xmath41 ( @xmath42 ) events could be expected for @xmath43 fb , which corresponds to @xmath44 @xmath45 gev from fig . [ fig : total - cross - section - vs - emu ] .
this would provide good potential to improve the sensitivity by four ( two ) orders of magnitude from the present limit from @xmath46 decay , respectively .
such a muon intensity could be available at a future muon collider and a neutrino factory . in the higgs boson mediated interaction ,
the tau leptons are emitted at a relatively large angle with respect to the beam direction .
this is in contrast to the gauge mediated interaction where the tau leptons are forward - peaked . in fig .
[ fig : angle - dependence ] , the @xmath47 dependence in the differential cross section is shown for each @xmath48 at @xmath0 = 50 and 100 gev . to identify the tau lepton from the reaction , direct measurement of tau lepton tracks ( such as by emulsions )
might not be possible at such a high beam rate .
instead , the identification might be possible by tagging the tau decay products and observing their decay kinematics . among various decay modes
, one might consider leptonic decays of the tau leptons .
another candidate could be to detect a hadron from the two - body tau decays .
the branching ratios , such as @xmath49 , @xmath50 and @xmath51 , are about 0.3 in total . in particular , in susy models with left - handed slepton mixing , the @xmath52(@xmath53 ) produced through the higgs - mediated interaction is only right - handed ( left - handed ) for an incident left - handed @xmath54 beam ( right - handed @xmath55 beam ) .
the hadrons from the right - handed @xmath52 decay ( left - handed @xmath53 decay ) tend to be emitted in the direction of the parent tau lepton , and therefore be rather energetic .
therefore , the signature of the events could be a hard hadron at a relatively large angle from the beam direction ( namely a hadron with large transverse momentum @xmath56 ) and some missing energy .
those hadrons from the tau decay should be discriminated from the hadrons from the target nucleons which have mostly soft energies .
this process can be useful to search for the higgs - boson mediated lfv coupling , especially when @xmath40 is higher than 50 gev .
there , the contributions from the sea @xmath1-quarks become significant , and the cross section is drastically enhanced . | a deep inerastic scattering process is discussed to study lepton flavor violation between muons and tau leptons . in supersymmetric models , the higgs boson mediated diagrams could be important for this reaction .
we find that at a muon energy ( @xmath0 ) higher than 50 gev , the predicted cross section significantly increases due to the contribution from sea @xmath1-quarks .
the number of produced tau leptons can be @xmath2 at @xmath0= 300 gev from @xmath3 muons , whereas @xmath4 events are given at @xmath5 gev . |
the @xmath3-ray emission following the nova explosion is dominated by the @xmath4 energy line , coming from the annihilation of positrons produced by the decay of radioactive nuclei . among them
, the @xmath0f is especially important because of its expected abundance in the nova environment and because of its lifetime , that well matches the timescale for nova ejecta to become transparent to @xmath3-ray emission . to understand the nova explosion phenomena , it is then important to know the rate of nuclear reactions producing and destroying @xmath0f . at relevant temperatures ,
the @xmath0f(p,@xmath1)@xmath2o reaction is expected to dominate by roughly a factor of 1000 @xcite and it is the uncertainty in this reaction rate that gives the main nuclear contribution to the overall uncertainty in the final abundance of @xmath0f .
the cross section of the reaction is dominated by the presence of several levels in the @xmath5ne compound nucleus around the @xmath0fp threshold .
moreover , the tails of some subthreshold levels are expected to contribute to the resonance cross section .
especially important is not only the width of these resonance states but also their interference effects .
it is then important to explore the behavior of the cross section in the center of mass energy range from @xmath6 up to @xmath7 to get information about the reaction rate at nova temperatures . in spite of several and long lasting attempts to measure the cross section of @xmath0f(p,@xmath1)@xmath2o reaction ( see e.g. @xcite )
the situation is still not satisfactory . in order to get new complementary pieces of information on this process , we performed two experiments aiming at the measurement of the @xmath0f(p,@xmath1)@xmath2o using two different techniques : the direct thick target method and the indirect trojan horse method ( thm ) .
we report here only on the latter one .
the thm was applied to a reaction induced by a radioactive ion beam ( rib ) for the first time in the experiment described here .
namely , the three body reaction @xmath0f(d,@xmath1 @xmath2o)n was used to infer information on the process of interest , @xmath0f(p,@xmath1)@xmath2o , in the low energy region important for astrophysics .
the trojan horse method ( thm ) @xcite has been proposed and exploited to study the cross sections of nuclear reactions between charged particles at astrophysical energies . briefly , a reaction of the type + @xmath8 can be studied by using a three - body reaction of the type + @xmath9 where @xmath10 is a nucleus with a strong cluster structure of the type @xmath11 and it is often called the _ trojan horse _ nucleus . the measurement of the reaction ( 2 ) is performed at energy higher than both the coulomb barrier and the energy required to breakup particle @xmath10 into its components @xmath12 and @xmath13 . if reaction ( 2 ) proceeds through a quasi - free reaction mechanism , then it can be described by the diagram shown in figure [ thm ] and the cross section of the three - body reaction can be factorized in three parts : a kinematical term , a term describing the virtual decay of @xmath10 into @xmath12 and @xmath13 and one giving the cross section of the reaction ( 1 ) . using a plane wave impulse approximation approach to describe the diagram in figure [ thm ] ,
this can be written as @xcite + @xmath14 where @xmath15 is the kinematical factor , @xmath16 describes the impulse distribution of the spectator @xmath13 inside @xmath10 and @xmath17 is the momentum of the spectator .
the symbol @xmath18 for the cross section of the reaction ( 1 ) indicates that it is a pure nuclear cross section , whitout effects coming from coulomb interactions .
hence , by measuring the cross section of the three body process ( 2 ) and the impulse distribution of @xmath13 inside @xmath10 ( represented by @xmath19 ) , and by calculating the kinematical factor ( kf ) , one can derive the cross section for the process of astrophysical interest ( 1 ) .
interaction energy of the reaction ( 1 ) is given by the relation @xmath20 , where @xmath21 is the relative energy of the outgoing particles @xmath22 and @xmath23 and @xmath24 is the q - value of the two body process ( 1 ) . the calculated two body cross section is a purely nuclear one , the coulomb barrier having been overcome by the th mechanism @xcite .
hence , in order to compare the thm cross sections with the directly measured ones , the barrier suppression effects have to be taken into account by correcting the thm cross section for the probability of penetration under barrier . the thm has been used a number of times @xcite to measure cross sections of astrophysical interest and it is presently regarded as a very powerful tool by the nuclear astrophysics community . moreover the thm can be also used to study neutron induced reaction @xcite .
the simple theoretical approach discussed above has been improved along the years .
more sophisticated techniques have been introduced to treat more complicated situations as in the cases where the cross section is dominated by the presence of narrow , close and/or interfering resonances @xcite .
the @xmath0f beam was produced at the crib separator of the center for nuclear study ( cns ) of the university of tokyo installed at riken campus in wako , japan , by using the @xmath0o(p , n)@xmath0f reaction .
the @xmath0o primary beam ( @xmath25 , e=4.5 - 5 @xmath26 ) was provided by a avf cyclotron .
it impinged on a primary gas target of h@xmath27 .
the reaction products were then separated by the crib doubly achromatic spectrometer , described in detail in ref .
@xcite .
two beam production tests were performed before the experiment .
the characteristics of the @xmath0f beam used during the experiment were the following : beam energy peaked at @xmath28 ( fwhm @xmath29 ) with maximum beam intensity of 2@xmath3010@xmath31 pps and purity better than 98% .
the intensity was higher than 5@xmath3010@xmath32 pps throughout the measurement .
this beam was used to bombard a cd@xmath27 secondary target .
the experimental set - up , schematically shown in figure ( [ astrho ] ) , was very different from the typical ones used for thm experiments .
indeed , it consisted essentially of two parts : a beam tracker placed before the secondary target and two planes of position sensitive silicon detectors to detect the @xmath2o and @xmath1 ejectiles .
the tracks of the beam particles were reconstructed , event by event , using two parallel plate avalanche chambers ( ppac ) with a position resolution ( @xmath12-@xmath33 ) of roughly 1 @xmath34 and a distance between the two planes of 500 @xmath34 . during the experiment the second ppac broke down and it was replaced by a microchannel plate detector ( mcp ) having the same position resolution .
the mcp detected the electrons coming from the interaction of the beam with a carbon foil placed in the same position of the broken ppac .
the precise particle tracking allowed for the determination of the intersection point of the projectiles on the secondary target .
this is a key point for the application of the thm , as discussed in @xcite .
the secondary target was made of a thin ( roughly 150 @xmath35g/@xmath36 ) cd@xmath27 foil of 30 mm diameter .
the detector setup was based on the modular system built at the infn laboratori nazionali del sud called astrho ( a silicon array for trojan horse ) . in the configuration used for this experiment , astrho consists of a mechanical frame that hosts two planes of position sensitive detectors .
the first plane consisted of 8 bidimentional position sensitive silicon detectors manufactured by hamamatsu .
each detector is 45@xmath3045 mm@xmath37 and measures the energy and the position on @xmath12 and @xmath33 directions of the impinging particle .
the spatial resolution is of 1 @xmath34 in both directions .
the second plane was made of a pair of double sided silicon multistrip detectors ( dsssd ) , manufactured by micron .
each detector ( 59@xmath3050 mm@xmath37 ) has 16 @xmath3016 strips in @xmath12 and @xmath33 direction respectively , with a spatial resolution of each strip of 1.5 @xmath34 .
the detector holders were especially designed for this new apparatus , with the aim of having a well defined geometry by construction and a support that can be adapted for different experiments by changing the distances among the different detector planes and the target . in the present case
the first plane of bidimensional position sensitive detectors was optimized to detect the @xmath1 particles , while @xmath2o is detected at most forward angles by dsssd .
the angular ranges covered by the experimental set - up were @xmath38 for @xmath1 particles and @xmath39 for @xmath2o ions .
for each detector the @xmath12 and @xmath33 position and the energy of the impinging particles were recorded together with the time of flight .
in order to apply the thm to the study of @xmath0f(p,@xmath1)@xmath2o , we selected the @xmath0f(d,@xmath1 @xmath2o)n three body reaction
. given the low intensity of a rib with respect to a stable beam , the use of a large solid angle setup like astrho was necessary to increase the global detection efficiency .
moreover the beam track reconstruction event by event was mandatory to reconstruct the angles of the ejected particles with respect to the actual incident particle trajectories , improving the angular resolution of the experiment . in figure [ beam ]
the beam image reconstructed on the two detectors planes and on the target are reported .
it is clearly seen that even if the beam spot on the target position is small , the beam divergence is not negligible .
the calibration in energy of the detectors was performed by using the elastic scattering peaks measured at different beam energies for the reaction @xmath0o@xmath40au . moreover
, a triple alpha source was used to calibrate at low energies .
the detectors made by hamamatsu were calibrated in position by using grids placed in front of each detector during the calibration runs .
the energy lost on the target by the beam , the detected particle energy loss on the target and on the dead layer of the detectors were also taken into account for the calibration in position and energy .
after calibration , the data analysis of a thm experiment mainly proceeds through the following steps : identification of the process of interest , selection of the events corresponding to the quasi - free reaction channel , extraction of the two body cross section of astrophysical interest . in the present experiment , due to
the limitation in mass and z identification of the outgoing particles , other three reactions with three bodies in the final state can give contribution in the phase space region where one expects events coming from @xmath0f(d,@xmath1 @xmath2o)n .
these reactions are @xmath0f(d,@xmath1 @xmath2n)p , @xmath0f(d , p @xmath0o)p and @xmath0f(d , p @xmath0f)n .
we were able to select the channel of interest by applying convenient cuts in the phase space .
figure 4 shows the q - value spectra that was obtained by applying the event selection procedure .
the position and width of the peak are in very good agreement with the experimental conditions .
the expected value for events coming from @xmath0f(p,@xmath1)@xmath2o is 0.658 mev to be compared with a measured one of 0.647 mev .
also , the measured width of the q - value peak of fwhm 2.3 mev is well accounted for by the intrinsic beam energy spread ( 1.9 mev fwhm ) plus the kinematical enlargement .
this gives us confidence of the correct identification of the events coming from the channel of interest .
the second step mentioned above in a thm analysis is the selection of events that come from the quasi - free channel in the @xmath0f(d,@xmath1 @xmath2o)n reaction , among other processes falling the same particles in the final state .
the strongest evidence of the predominance of the quasi - free mechanism is given by the shape of the momentum distribution for the @xmath41-@xmath42 intercluster motion in deuteron .
this is shown in figure [ ps ] by black dots as a function of the momentum of the spectator particle , @xmath17 .
the solid line in figure [ ps ] represents the hulthn function in momentum space with the standard parameters values @xcite .
this is expected for deuteron as _ trojan horse _
nucleus if the plane - wave impulse approximation is used to describe the reaction mechanism .
distortion effects influence the behavior of the distribution at high values of momentum , so the rest of the analysis was done by imposing a cut on the momentum of the spectator particle . namely , only events where the spectator momentum was lower than 50 mev / c were accepted . assuming the events selected according to the previous procedure are actually coming from the quasi - free contribution to the reaction yield
, then one can apply equation ( 3 ) to obtain the cross section of interest for the @xmath0f(p,@xmath1)@xmath2o process at astrophysical energies .
the result is shown in figure 6 , that represents the nuclear cross section for @xmath0f(p,@xmath1)@xmath2o measured by thm down to zero ( and even negative ) energies without any suppression due to the coulomb barrier . in order to compare the present result with the direct data ,
it is necessary to correct the obtained nuclear cross section for the probability of penetration of the coulomb and , possibly , centrifugal barrier .
data analysis is still in progress .
the thm was applied for the first time to study a reaction induced by a radioactive ion beam .
even with the use of indirect methods like thm , the measurement of cross sections of interest for nuclear astrophysics remains one of the most difficult tasks in nuclear physics , especially if one has to use ribs when the low beam intensity add on the top of the low cross sections typical of astrophysical nuclear processes .
the use of a new large solid angle detector setup especially designed for this experiment was hence necessary in order to recover , at least in part , the possibility of accumulating a statistics high enough to study the processes of interest .
the preliminary data analysis has demonstrated the feasibility of the experiment and gives us confidence that the results coming from the final analysis of the data will be helpful in understanding the problem of @xmath0f destruction in novae .
we are indebted to the avf and crib technical staff for their invaluable contribution to the experiment .
also we would like to thank the infn - lns workshop crew headed by mr.b .
trovato for the perfect construction of the astrho mechanical frame and mr . c. marchetta from the target laboratory at infn - lns for providing us with , among others , high quality and high purity cd@xmath27 targets .
the work was supported in part by the italian ministry of university and research under grant no .
rbfr082838 ( firb2008 ) .
+ + 99 a. coc et al . , a.&a . , 357 ( 2000 ) 561 and references therein
. m. wiescher and k.u .
kettner , astrophys .
j. 263 ( 1982 ) 891 .
rehm , m. paul , a.d .
roberts , et al . , phys .
c52 ( 1995 ) r460 .
rehm , m. paul , a.d .
roberts , et al .
c53 ( 1996 ) 1950 .
rehm , c.l .
jiang , m. paul , et al .
c55 ( 1997 ) r566 .
graulich , f. binon , w. bradfield - smith , et al .
a626 ( 1997 ) 751 .
r. coszach , m. cogneau , c.r .
bain , et al .
lett . b353
( 1998 ) 184 .
s. utku , j.g .
ross , n.p.t .
bateman , et al .
c57 ( 1998 ) 2731 and c58 ( 1998 ) 1354 .
y.m . butt , j.w .
hammer , m. jaeger , et al .
c58 ( 1998 ) r10 .
graulich , s. cherubini , r. coszach , et al.,phys .
c63 ( 2001 ) 011302 and references therein .
bardayan , j.c .
blackmon , w. bradfield - smith , et al .
, phys . rev .
c63 ( 2001 ) 065802 .
bardayan , j.c .
batchelder , j.c .
blackmon , et al .
( 2002 ) 262501 .
n. de sreville , a. coc , c. angulo , et al .
c67 ( 2003 ) 052801 ( r ) and references therein .
kozub , d.w .
bardayan , j.c . batchelder , et al .
c71 ( 2005 ) 032801 ( r ) .
kozub , d.w .
bardayan , j.c .
batchelder , et al .
a 758 ( 2005 ) 753c-756c . c. e. beer et al .
c 83 , 042801 ( 2011 ) a. s. adekola , d. w. bardayan , j.c .
blackmon , et al .
, phys . rev . * c83 * , 052801 ( r ) , ( 2011 ) . c. spitaleri . _ in proceedings of the fifth hadronic physics winter seminar , folgaria - tn , italy _
world scientific , singapore , 1990 .
s. cherubini , v. n. kondratyev , m. lattuada , c. spitaleri , d. miljanic , m. zadro , and g. baur .
_ apj _ , * 457 * , 855 , ( 1996 ) g. baur and s. typel .
_ , * 154 * , 333 , ( 2004 ) s. typel and g. baur . , * 305 * , 228 , ( 2003 ) a. m. mukhamedzhanov
_ et al . _ , * a787 * , 321c , 2007 .
c. spitaleri , m. aliotta , s. cherubini , m. lattuada , dj .
miljani , s. romano , n. soi , m. zadro , and r. a. zappal , , 055802 , 1999 .
m. gulino _
et al._,_j . phys .
g : nucl.part.phys._ * 37 * , 125105 ( 2010 ) y. yanagisawa , s. kubono , t. teranishi , et al .
instr . meth . a 539 ( 2005 )
74 s. cherubini
_ et al._,_aip conf .
_ proceedings of omeg11 , in press ( 2012 ) m. zadro _
_ , _ phys.rev .
_ * c40 * 181 , 1989 | the trojan horse method was applied for the first time to a radioactive ion beam induced reaction to study the reaction @xmath0f(p,@xmath1)@xmath2o via the three body reaction @xmath0f(d,@xmath1 @xmath2o)n at the low energies relevant for astrophysics .
+ the abundance of @xmath0f in nova explosions is an important issue for the understanding of this astrophysical phenomenon .
for this reason it is necessary to study the nuclear reactions that produce or destroy @xmath0f in novae .
@xmath0f(p,@xmath1)@xmath2o is one of the main @xmath0f destruction channels .
+ preliminary results are presented in this paper . |
thanks to the out - standing developments in recent neutrino physics a rather clear picture of neutrino oscillation parameters is emerging .
we know that there are two large angles in the lepton mixing matrix , @xmath10 and @xmath0 , and one small angle @xmath11 , and the mass - squared differences @xmath12 and @xmath13 are well determined by the global data from solar @xcite atmospheric @xcite , reactor @xcite , and accelerator @xcite neutrino experiments . a recent global analysis of world neutrino oscillation data can be found _
e.g. _ in ref .
despite these enormous achievements there are still important open questions related to neutrino physics , which have to be addressed by future neutrino oscillation experiments : 1 . what is the value of the mixing angle @xmath11 ?
2 . what is the value of the complex phase @xmath6 in the mixing matrix ?
3 . is the neutrino mass ordering normal ( @xmath14 ) or inverted ( @xmath15 ) , _ i.e. _ , what is the sign of @xmath2 ?
in contrast to the `` discovery phase '' , which was dominated by natural neutrino sources such as the sun or the atmosphere , the subsequent generation of oscillation experiments will be mainly based on man - made neutrinos , where the neutrino source is well under control . among the future projects are conventional beam experiments @xcite , new reactor experiments @xcite , superbeam experiments @xcite , and eventually experiments based on a neutrino factory or a beta - beam @xcite .
a characteristic feature in the analysis of future long - baseline ( lbl ) experiments is the presence of so - called _
parameter degeneracies _ , see _ e.g. _ , refs . @xcite . due to the inherent three - flavor structure of the oscillation probabilities ,
for a given experiment in general several disconnected regions in the multi - dimensional space of oscillation parameters will be present .
traditionally these degeneracies are referred to in the following way : * the _ intrinsic _ or ( @xmath16)-degeneracy @xcite : for a measurement based on the @xmath17 oscillation probability for neutrinos and anti - neutrinos two disconnected solutions appear in the ( @xmath16 ) plane . * the _ hierarchy _ or sign(@xmath2)-degeneracy @xcite : the two solutions corresponding to the two signs of @xmath2 appear in general at different values of @xmath6 and @xmath11 . * the _ octant _ or @xmath0-degeneracy @xcite : since lbl experiments are sensitive mainly to @xmath18 it is difficult to distinguish the two octants @xmath19 and @xmath20 .
again , the solutions corresponding to @xmath0 and @xmath21 appear in general at different values of @xmath6 and @xmath11 .
this leads to an eight - fold ambiguity in the determination of @xmath11 and @xmath6 @xcite , and hence degeneracies provide a serious limitation in the ability to answer the questions 1 and 2 above .
moreover , the fact that one speaks of a `` sign(@xmath2)-degeneracy '' illustrates that answering question 3 is difficult : determining the neutrino mass ordering is equivalent to resolving the sign(@xmath2)-degeneracy .
several methods to resolve these degeneracies have been proposed , among them the combination of experiments at various baselines and/or @xmath22-values @xcite , the use of spectral information @xcite , the combination of @xmath23 and @xmath24 oscillation channels @xcite , and the combination of lbl and reactor experiments @xcite . in the present work
we discuss a new possibility to resolve the lbl parameter degeneracies , based on the data from atmospheric neutrinos .
it is known that atmospheric neutrinos are in principle sensitive to @xmath11 and the neutrino mass hierarchy due to earth matter effects @xcite in the @xmath1-like events @xcite .
these effects are most pronounced in the multi - gev energy range and for large zenith angles , corresponding to neutrino trajectories crossing the earth mantle or the mantle and the core .
in addition effects from the solar parameters @xmath10 and @xmath12 on @xmath1-like events in the sub - gev energy range provide sensitivity to the octant of @xmath0 @xcite , and in principle even on @xmath6 @xcite .
for a recent discussion of sub - leading effects in atmospheric neutrino oscillations see ref .
it turns out , however , that atmospheric ( atm ) data on its own can never compete with lbl in many respects , such as the determination of @xmath25 , or @xmath18 .
one reason is that they are limited by systematical uncertainties , _
e.g. _ from the neutrino fluxes .
however , as we will show in the following , due to the effects mentioned above atm data can break the sign(@xmath2)- and @xmath0-degeneracies in lbl data , and hence the combined analysis of lbl and atm leads to significant synergies .
an important part of the future neutrino program are gigantic water cherenkov detectors at the megaton scale .
apart from serving as detector for lbl experiments such a facility will provide unprecedented opportunities for proton decay , solar and atmospheric neutrino experiments , as well as for the detection of supernova and other astrophysical neutrinos .
the projects under discussion are uno @xcite in the us , a megaton detector in the frejus laboratory @xcite in europe , and the hyperkamiokande project @xcite in japan .
if a lbl experiment with such a detector will be built atmospheric neutrino data come for free .
therefore , our method provides a very efficient possibility to resolve the parameter degeneracies from lbl data , in contrast to the previously discussed methods , which in general are based on the combination of two or more expensive experiments . in the following
we will illustrate how the lbl+atm combination works by considering the phase ii of the t2k experiment ( ) @xcite , assuming a 4 mw superbeam produced at the j - parc accelerator , and the 1 mt hyperkamiokande ( ) detector serving as the far detector for the lbl experiment as well as providing the high statistics atmospheric neutrino data .
let us note that similar results are expected also for the other megaton detector proposals .
furthermore , if huge magnetized iron detectors are available one expects similar synergies between lbl and atm data . for these type of detectors the ability to distinguish between neutrinos and anti - neutrinos will increase the potential of atm data even further @xcite . in the present work we stick to the water cherenkov detector
, the potential of magnetized detectors will be considered elsewhere .
the outline of the paper is as follows . in section [ sec : technicalities ]
we give some technical details of our simulation of the experiment ( section [ sec : t2k ] ) and of the atmospheric neutrino analysis ( section [ sec : atm ] ) , and we describe the statistical methods used in the following ( section [ sec : statistics ] ) . in section [ sec : degeneracies ] we illustrate the lbl / atm complementarity by discussing the effects of parameter degeneracies for the lbl data ( section [ sec : qual - lbl ] ) , whereas in sections [ sec : qual - atm - th23 ] and [ sec : qual - atm - hier ] we consider the effects in atm data which allow to resolve the octant and the sign(@xmath2 ) degeneracies , respectively . in section
[ sec : resolve ] we investigate in detail the ability to exclude degenerate solutions with lbl+atm data by performing a systematic scan of the parameter space of @xmath11 , @xmath0 and @xmath6 . in section
[ sec : hierarchy ] we discuss the potential to identify the neutrino mass ordering , for the special case @xmath26 ( section [ sec : hier - max ] ) as well as for the general case ( section [ sec : hier - nonmax ] ) . in section [ sec : octant - th13 ]
we show how the sensitivity to @xmath11 is improved by resolving the @xmath0-degeneracy by atm data , and we summarize our results in section [ sec : conclusions ] .
in the following the label `` lbl '' refers to the phase ii of the t2k experiment ( ) @xcite .
we are assuming a high luminosity superbeam with mean neutrino energy of 0.76 gev , produced with a target power of 4 mw at the j - parc accelerator .
the neutrinos are detected at a 1 mt water cherenkov detector , hyperkamiokande ( ) , at a baseline of 295 km and an off - axis angle of @xmath27 .
we consider 2 years running time with a neutrino beam , and 6 years with anti - neutrinos , such that comparable event numbers are obtained for neutrinos and anti - neutrinos .
our simulation of the experiment is performed by using the globes software package @xcite .
we take into account realistic neutrino fluxes , detection cross sections , energy resolution , and efficiencies @xcite .
this experimental information is folded with the three - flavor oscillation probability , fully taking into account the earth matter effect .
we consider all available information , from @xmath28 appearance , as well as @xmath29 disappearance channels .
the signal events are given by @xmath30 and @xmath29 charged current ( cc ) interactions , respectively .
we divide the signal into the total event rate , where the full cc cross section is used , and into the energy spectrum with free normalization , where only events from quasi - elastic scattering are used , since non - quasi - elastic events do not allow to reconstruct the neutrino energy .
various backgrounds such as a @xmath30 contamination of the beam , @xmath29 neutral current events , and misidentification of muon neutrinos as electron neutrinos are taken into account , using information given in ref .
we list the numbers of signal and background events expected in for typical oscillation parameters in table [ tab : eventnumbers ] .
more details of our analysis can be found in ref .
@xcite .
.number of events in the lbl and atm experiments considered in our work for the oscillation parameters @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 ev@xmath36 , and @xmath37 ev@xmath36 . [ cols= " < , > , > , <
, > , > " , ] [ tab : eventnumbers ] to simulate atmospheric neutrino data in the detector we follow closely ref .
. a crucial element of the analysis is to take into account the full three - flavor oscillation probability , including earth matter effects , as well as oscillations induced by the `` solar '' mass splitting @xmath12 ( for other three - flavor analyses see refs .
@xcite ) . in our analysis
we consider charged current data , divided into sub - gev and multi - gev @xmath1-like and @xmath38-like contained event samples ( each grouped into 10 bins in zenith angle ) , as well as stopping ( 5 angular bins ) and through - going ( 10 angular bins ) up - going muon events .
details of our statistical analysis can be found in the appendix of ref .
@xcite . together with the statistical errors , we consider theoretical as well as systematical uncertainties , where theoretical uncertainties are uncertainties in the original atmospheric neutrino fluxes and in the cross - sections .
we are using the atmospheric neutrino fluxes from ref .
@xcite , and flux uncertainties include total normalization errors ( 20% ) allowing for an independent fluctuation of neutrino and anti - neutrino fluxes ( 5% ) as well as @xmath29 and @xmath30 fluxes ( 5% ) , spectral uncertainty of the fluxes ( `` tilt '' factor ) , and an uncertainty on the zenith angle dependence which induces an error in the up / down asymmetry of events ( 5% ) .
we also include independent normalization errors for the different contributions to the interaction cross section : quasi - elastic scattering ( 15% ) , single pion production ( 15% ) , and deep inelastic scattering ( 15% for contained events and 10% for upward - going muons ) .
moreover , we include as systematical errors experimental uncertainties associated with the simulation of the hadronic interactions , the particle identification procedure , the ring - counting procedure , the fiducial volume determination , the energy calibration , the relative normalization between partially - contained and fully - contained events , the track reconstruction of up - going muons , the detection efficiency of up - going muons , and the stopping / through - going separation .
the current atmospheric neutrino data sample from superkamiokande ( sk - i ) consists of 1489 days of data ( contained events ) with a detector mass of 22.5 kt , which gives roughly @xmath39 of data . in this work we are considering a data taking period for the lbl experiment of 8 years .
we assume that the 1 mt detector will be finished one year before the beam , and hence @xmath40 of atmospheric neutrino data will be available . for the atmospheric data sample used in our analysis , which we will denote in the following by the label `` atm '' , we scale the present sk - i sample ( 1489 days contained events , 1657 days of stopping , and 1678 days of through - going muons ) by a factor 100 .
event numbers for various atm data samples for typical oscillation parameters are given in table [ tab : eventnumbers ] . in order to investigate the potential of the experiments described in the previous sub - sections we adopt the standard method for analyzing future experiments .
first , artificial `` data '' is simulated by calculating event numbers for lbl and atm for some `` true values '' @xmath41 for the oscillation parameters @xmath42 .
then a @xmath43-analysis of these `` data '' is performed to extract allowed regions for the oscillation parameters .
it is important to note that in general the results will depend on the values adopted for @xmath41 .
for all calculations we will use for @xmath44 and @xmath45 the best fit values obtained in ref .
@xcite , @xmath46 , @xmath47 , and sign(@xmath2 ) , and show our results as a function of these parameters .
furthermore , since neither lbl nor atm data allow an accurate determination of @xmath10 and @xmath12 we assume that these two parameters are known with an uncertainty of 10% from solar and reactor neutrino experiments . if one is interested in the allowed range for a certain parameter @xmath48 for a given choice of @xmath41 , the function @xmath49 has to be minimized with respect to all other parameters @xmath50 except @xmath51 to take into account the correlations and degeneracies between parameters .
this minimization is performed by using the globes software package @xcite , which has been generalized in order to include the atmospheric neutrino code .
let us stress that for both data samples , lbl as well as atm , a full three - flavor analysis including matter effects is performed .
the only approximation is to neglect the ( very weak ) dependence on @xmath10 .
since lbl depends in leading order only on the product @xmath52 fixing @xmath10 does not introduce any error , as long as the dependence on @xmath12 is properly taken into account , and for atm varying @xmath10 in the allowed range is expected to have a very small impact .
apart from this simplification the full parameter dependence of both data sets has been taken into account .
however , we assume that lbl and atm data are statistically independent .
this might not be completely correct since both data sets are based on the same detector , and hence , uncertainties related to the detection process can introduce correlations between lbl and atm data .
although such effects will have to be included eventually in the analysis of _ real _ data , we expect that such correlations introduce only very minor corrections to our present results .
in this section we discuss the problem of parameter degeneracies for the experiment .
first we note that for this particular experiment the @xmath53-degeneracy does not occur .
it is known that for experiments operating at the oscillation maximum of @xmath2 the second solution in the @xmath54 plane can be disfavored efficiently @xcite . moreover , spectral information is important , since @xmath6-dependent and @xmath6-independent terms in the oscillation probability show a different energy dependence ( compare eq .
( [ eq : emuprob ] ) later in this section ) .
thus it is difficult to leave the spectrum unchanged when @xmath6 is varied , and the @xmath55-degenerate solution , which implies a different value of @xmath6 , is disfavored .
an illustration of the relevance of spectral information for the @xmath55-degeneracy in the experiment can be found in figure 5 of ref .
@xcite .
therefore , in the case of we are confronted only with the sign(@xmath2)- and @xmath0-degeneracies .
apart from the incapacity to determine the neutrino mass ordering and the octant of @xmath0 this leads to a four - fold ambiguity in the determination of @xmath11 and @xmath6 . the impact of the degeneracies can be appreciated in figure [ fig : d - th - nonmax-2 ] , where the solid curves show the allowed regions from lbl data in the @xmath56 plane for an example - point with the true values @xmath57 , @xmath58 , and non - maximal values of @xmath59 .
apart from the true solution , three degenerate regions are present , corresponding to the wrong octant of @xmath0 , the wrong sign of @xmath2 , and the wrong octant as well as the wrong hierarchy .
let us introduce the following abbreviations to denote these four solutions : @xmath60 plane at 2@xmath61 , 99 @xmath62 , @xmath63 , and @xmath64 ( left ) and @xmath65 ( right ) .
the solid curves correspond to lbl data only , and the shaded regions correspond to lbl+atm data .
the true best fit point is marked with a star , the best fit points of the degenerate solutions are marked with dots , and the corresponding @xmath66-values of lbl+atm data are given in the figure .
the true mass ordering is the normal hierarchy.,scaledwidth=95.0% ] [ fig : d - th - nonmax-2 ] a qualitative understanding of the degenerate solutions can be obtained from the approximate formula for the @xmath67 appearance probability in vacuum @xcite @xmath68 with @xmath69 and @xmath70 .
the sign in the second term is ` @xmath71 ' for neutrinos and ` @xmath72 ' for anti - neutrinos . since in the experiment matter
effects are small this expression for the probability suffices to obtain a qualitative understanding of most of the effects in lbl data presented throughout this work .
following ref .
@xcite the location of the @xmath0-degenerate solution @xmath73 can be estimated from eq .
( [ eq : emuprob ] ) in the following way .
since the experiment operates close to the oscillation maximum it is a good approximation to use @xmath74 .
furthermore , for values of @xmath75 the @xmath76 term can be neglected . under these approximation solving the two equations
@xmath77 leads to @xmath78 for the parameter values of the wrong-@xmath0 solution .
for the example shown in figure [ fig : d - th - nonmax-2 ] eq . ( [ eq : fake - th13-th23 ] ) gives @xmath79 for @xmath64 , and @xmath80 for @xmath65 .
furthermore , eq . ( [ eq : fake - delta - th23 ] ) gives @xmath81 for @xmath64 , and @xmath82 for @xmath65 .
these numbers are in good agreement with the actual fit shown in figure [ fig : d - th - nonmax-2 ] .
one observes that the @xmath0-degeneracy has a strong impact on the measurement of @xmath5 , however , it hardly affects the determination of @xmath6 @xcite ; the @xmath73-solution occurs practically at the same value of @xmath6 as the true one .
this is in contrast to the sign(@xmath2)-degeneracy , which in general leads to a severe ambiguity for @xmath6 , whereas the @xmath5 measurement is essentially unaffected .
following ref .
@xcite we observe from eq .
( [ eq : emuprob ] ) that only the term in the second line is affected by changing the sign of @xmath2 .
this term transforms as @xmath83 under @xmath84 .
therefore , the full probability stays invariant under this transformation if @xmath6 is adjusted such that @xmath85 , which can be achieved for @xmath86 according to this reasoning we would obtain in the example plotted in figure [ fig : d - th - nonmax-2 ] the value @xmath87 .
although in this case the accuracy is not excellent , eq .
( [ eq : fake - delta - h ] ) still provides a rough method to estimate the location of the @xmath88-solution in @xmath6 ( for more refined methods see _ e.g. _ refs .
@xcite ) . finally , the values of @xmath5 and @xmath6 corresponding to the combined @xmath0 and sign(@xmath2)-degeneracy @xmath89 can be estimated by applying simultaneously eqs .
( [ eq : fake - th13-th23 ] ) and ( [ eq : fake - delta - h ] ) .
this four - fold degeneracy can be lifted to large extent if lbl data is combined with data from atmospheric neutrinos .
we observe from figure [ fig : d - th - nonmax-2 ] that the degenerate solutions corresponding to the wrong octant of @xmath0 are highly disfavored by the inclusion of atm data , at the level of @xmath90 .
furthermore , also the solution with the wrong mass ordering gets disfavored in the combined analysis , although in this case the ability to resolve the degeneracy is more subtle . in the following subsections we discuss the relevant features of atm data to resolve the degeneracies in lbl data .
the experiment will provide a very precise determination of @xmath18 thanks to the large statistics data from the @xmath29 disappearance channel ( compare table [ tab : eventnumbers ] ) .
the relative accuracy at 2@xmath61 is expected to be better than 1% . despite this impressive performance on @xmath18
there are some subtleties related to the measurement of @xmath91 ( see , _
e.g. _ , refs .
especially , if @xmath0 deviates from @xmath92 it will be impossible to distinguish between the two solutions at @xmath0 and @xmath21 . from the maximal value
e.g. _ refs .
@xcite and references therein .
] although the determination of @xmath93 from atmospheric data is significantly less precise than from lbl they provide the very interesting ability to distinguish between the two octants of @xmath0 .
the sensitivity of atmospheric data to the deviation of @xmath0 from @xmath92 follows mainly from effects of the solar mass splitting @xmath12 for @xmath1-like events in the sub - gev region , see _ e.g._refs .
the excess of @xmath1-like events can be written as @xmath94 here @xmath95 is the number of @xmath1-like events with ( without ) oscillations , @xmath96 is the averaged two neutrino oscillation probability given by the solar parameters @xmath97 and @xmath10 including the ( weighted ) contributions from neutrinos and anti - neutrinos , and @xmath98 is the ratio of the initial muon and electron neutrino fluxes . since for sub - gev energies @xmath99 the effect is suppressed for @xmath100 , however it provides a sensitive measure for deviations from maximal mixing .
given the present lma parameters , for @xmath101 one expects @xmath102 values of a few percent ( see _ e.g. _ fig .
8 of ref .
in contrast to lbl data , which is essentially sensitive only to @xmath18 , this effect depends on @xmath103 , and therefore the discrimination between @xmath104 and @xmath19 becomes possible .
of the solution in the wrong octant of @xmath0 as a function of the true value of @xmath91 for lbl data only , for atm data only , and for the lbl+atm combination . furthermore , we take @xmath105.,scaledwidth=50.0% ] [ fig : wrong - th23 ] building upon the results of ref .
@xcite we show in figure [ fig : wrong - th23 ] the difference in @xmath43 between the true solution ( with @xmath106 ) and the @xmath43-minimum in the wrong octant of @xmath0 ( @xmath73-solution ) .
it is clear form this figure that lbl data alone have no sensitivity at all to the octant of @xmath0 , whereas atm can reject the fake solution efficiently .
taking the combination of lbl+atm data improves the sensitivity slightly , due to the more precise determination of other oscillation parameters by the lbl data . using the combined lbl+atm data the wrong octant can be rejected at 3@xmath61 if @xmath107 .
let us note that the results shown in figure [ fig : wrong - th23 ] do hardly depend on our choice for the true neutrino mass ordering , thanks to @xmath105 .
non - zero values for @xmath108 will be considered in section [ sec : resolve ] .
the determination of the ordering of the neutrino mass states is one of the most challenging tasks of future neutrino experiments . in long - baseline experiments the effects of @xmath11 , @xmath6 and the sign of @xmath2
are highly entangled , and the determination of sign(@xmath2 ) is probably only possible through the matter effect induced in an experiment with a very long baseline plus additional information from other lbl or reactor experiments @xcite .
therefore , the combined analysis of lbl and atm data as offered from an oscillation experiment with a mt water cherenkov detector provides a very interesting possibility to answer the question of the neutrino mass hierarchy .
the sensitivity of atmospheric neutrino data to the neutrino mass hierarchy comes mainly from the modification of @xmath1-like multi - gev events by earth matter effects for not too small values of @xmath5 @xcite .
similar as in eq .
( [ eq : excess - sub ] ) , one finds for the @xmath11 induced excess of @xmath1-like multi - gev events @xmath109 now @xmath110 is an effective two - flavor probability governed by @xmath2 and @xmath11 , appropriately averaged and including the ( weighted ) contributions from neutrinos and anti - neutrinos .
the effect is most pronounced for zenith angles corresponding to neutrino trajectories crossing the earth mantle , or earth mantle and core , where @xmath5-effects can be resonantly enhanced due to matter effects @xcite . in the relevant zenith angle bins
@xmath111 can reach values of the order of 10% ( see _ e.g. _ fig .
5 of ref .
qualitatively @xmath111 shows the following behavior @xcite : * @xmath111 vanishes for @xmath112 and increases monotonically with @xmath5 .
* for the normal hierarchy the resonant matter enhancement occurs for neutrinos , whereas for the inverted hierarchy it occurs for anti - neutrinos .
since the event numbers in water cherenkov detectors are dominated by neutrinos because of larger cross sections ( see table [ tab : eventnumbers ] ) , @xmath111 is larger by a factor of @xmath113 for the normal hierarchy than for the inverted one . * since in the multi - gev range we have @xmath114 , the factor @xmath115 in eq .
( [ eq : excess - multi ] ) suppresses the excess of @xmath1-like events for small values @xmath116 , whereas larger values of @xmath91 increase @xmath111 .
in particular , the effect is enhanced for @xmath20 .
therefore , if the true hierarchy is normal and @xmath19 there is only a small excess of @xmath1-like multi - gev events , which can be accommodated to some extent with the inverted hierarchy .
for the example point with @xmath64 chosen in the left panel of figure [ fig : d - th - nonmax-2 ] the wrong hierarchy can be disfavored only with a @xmath117 .
in contrast , for @xmath20 the excess of @xmath1-like events is enhanced by the flux - factor @xmath118 .
therefore , a true normal hierarchy ( resonant enhancement for neutrinos ) plus the flux - factor enhancement leads to large values of @xmath111 , which can not be fitted with the inverted hierarchy ( resonant enhancement for the smaller anti - neutrino sample ) .
this explains the strong rejection of the wrong hierarchy solution in the right panel of figure [ fig : d - th - nonmax-2 ] , with a @xmath119 .
we will discuss the mechanisms relevant for the rejection of the wrong hierarchy in more detail in sections [ sec : resolve ] and [ sec : hierarchy ] .
finally we note that in addition to @xmath12-effects of eq .
( [ eq : excess - sub ] ) and @xmath11-effects of eq .
( [ eq : excess - multi ] ) also an interference term between the two contributions is present @xcite ( see also ref .
it is proportional to @xmath120 and depends on the cp - phase @xmath6 . because of the different dependence on the flux ratio @xmath121 the interference term may become important in cases where the effects governed by eqs .
( [ eq : excess - sub ] ) and ( [ eq : excess - multi ] ) are suppressed .
and @xmath91 for various @xmath47 if the true hierarchy is normal ( left ) or inverted ( right ) .
solid curves correspond to the solution with the wrong octant and the right hierarchy @xmath73 , dashed curves to the right octant and the wrong hierarchy @xmath88 , and shaded regions to the wrong octant and the wrong hierarchy @xmath89 .
we show the contours of @xmath122 between the fake and the true solution , corresponding to a rejection of the fake solution at the @xmath123 , @xmath3 , and @xmath124 cl ( from light to dark shading ) for 1 dof .
, title="fig:",scaledwidth=58.0% ] and @xmath91 for various @xmath47 if the true hierarchy is normal ( left ) or inverted ( right ) .
solid curves correspond to the solution with the wrong octant and the right hierarchy @xmath73 , dashed curves to the right octant and the wrong hierarchy @xmath88 , and shaded regions to the wrong octant and the wrong hierarchy @xmath89 .
we show the contours of @xmath122 between the fake and the true solution , corresponding to a rejection of the fake solution at the @xmath123 , @xmath3 , and @xmath124 cl ( from light to dark shading ) for 1 dof .
, title="fig:",scaledwidth=58.0% ] and @xmath91 for various @xmath47 if the true hierarchy is normal ( left ) or inverted ( right ) .
solid curves correspond to the solution with the wrong octant and the right hierarchy @xmath73 , dashed curves to the right octant and the wrong hierarchy @xmath88 , and shaded regions to the wrong octant and the wrong hierarchy @xmath89 .
we show the contours of @xmath122 between the fake and the true solution , corresponding to a rejection of the fake solution at the @xmath123 , @xmath3 , and @xmath124 cl ( from light to dark shading ) for 1 dof .
, title="fig:",scaledwidth=58.0% ] and @xmath91 for various @xmath47 if the true hierarchy is normal ( left ) or inverted ( right ) .
solid curves correspond to the solution with the wrong octant and the right hierarchy @xmath73 , dashed curves to the right octant and the wrong hierarchy @xmath88 , and shaded regions to the wrong octant and the wrong hierarchy @xmath89 .
we show the contours of @xmath122 between the fake and the true solution , corresponding to a rejection of the fake solution at the @xmath123 , @xmath3 , and @xmath124 cl ( from light to dark shading ) for 1 dof . , title="fig:",scaledwidth=58.0% ] [ fig : resolve ] the examples shown in figure [ fig : d - th - nonmax-2 ] suggest that the lbl+atm combination offers an efficient method to reject the degenerate solutions . to investigate this more
systematically we have performed a scan over the true values for @xmath11 and @xmath0 and various values for @xmath47 . for a given point in the space of @xmath125 , @xmath126 and @xmath47 we test the ability to rule out each of the three degenerate solutions @xmath73 , @xmath88 , and @xmath89 .
this is done by minimizing @xmath127 with respect to all fit parameters , constraining the octant of @xmath0 and the sign of @xmath2 corresponding to the fake solution which we want to test .
the results of this analysis are shown in figure [ fig : resolve ] , which represents one of the main results of this work .
the solid curves in figure [ fig : resolve ] show that the @xmath73-solution corresponding to the @xmath0-degeneracy can be excluded at high confidence level if @xmath0 is far enough from @xmath92 .
( note that if @xmath0 is close to @xmath92 this degeneracy disappears anyway . ) for small values of @xmath125 this follows mainly from the atmospheric sub - gev @xmath1-like events , as discussed in section [ sec : qual - atm - th23 ] .
moreover , if the true hierarchy is normal ( left panels of figure [ fig : resolve ] ) we find an improvement of the octant sensitivity for @xmath128 .
this effect comes from the multi - gev @xmath1-like events , where resonant enhancement occurs for sufficiently large @xmath5 , and therefore an additional dependence on @xmath91 appears according to eq .
( [ eq : excess - multi ] ) . this can be seen from figure [ fig : resolve - atm - samples ] , where we show the ability to reject the fake solutions using only sub - gev ( left ) and multi - gev ( right ) atm data .
the improvement for large @xmath5 is also visible for the inverted mass ordering ( right panels of figure [ fig : resolve ] ) , however , in that case the effect is much smaller , since the resonance occurs for anti - neutrinos in the inverted hierarchy , which contribute less to the total atm data , and hence the significance of the effect is smaller than in the normal hierarchy , where the resonance occurs for neutrinos . the wrong hierarchy solution @xmath88 can be excluded to the right of the dashed curves in figure [ fig : resolve ] .
we observe that the ability to reject this solution increases with @xmath91 . as discussed in section [ sec : qual - atm - hier ] this follows from the fact that the excess of @xmath1-like multi - gev events is enhanced for large @xmath91 .
in particular , for @xmath129 the sensitivity is completely dominated by multi - gev data ( compare figure [ fig : resolve - atm - samples ] ) , and we observe a very small dependence on the true value of @xmath6 and on the true hierarchy .
for @xmath130 the situation becomes more complicated , and the sensitivity depends on the true value of @xmath6 . in that region also sub - gev data can contribute significantly , as visible in figure [ fig : resolve - atm - samples ] . for low values of @xmath91 the effect of multi - gev data becomes suppressed by the flux factor @xmath131 in eq .
( [ eq : excess - multi ] ) , and the contribution of sub - gev data from eq .
( [ eq : excess - sub ] ) can become of comparable size .
moreover , as discussed in ref .
@xcite for certain parameter values an interference term between the two contributions may become relevant , which depends on @xmath6 .
therefore , the final sensitivity emerges from a rather involved interplay of various effects and atm data samples . -like and @xmath38-like atm data as a function of the true values of @xmath5 and @xmath91 for @xmath132 ( upper ) and @xmath133 ( lower ) and true normal hierarchy .
solid curves correspond to the solution with the wrong octant and the right hierarchy @xmath73 , dashed curves to the right octant and the wrong hierarchy @xmath88 , and shaded regions to the wrong octant and the wrong hierarchy @xmath89 .
we show the contours of @xmath134 between the fake and the true solution , corresponding to a rejection of the fake solution at the @xmath123 , @xmath3 , and @xmath124 cl ( from light to dark shading ) for 1 dof.,title="fig:",scaledwidth=70.0% ] -like and @xmath38-like atm data as a function of the true values of @xmath5 and @xmath91 for @xmath132 ( upper ) and @xmath133 ( lower ) and true normal hierarchy .
solid curves correspond to the solution with the wrong octant and the right hierarchy @xmath73 , dashed curves to the right octant and the wrong hierarchy @xmath88 , and shaded regions to the wrong octant and the wrong hierarchy @xmath89 .
we show the contours of @xmath134 between the fake and the true solution , corresponding to a rejection of the fake solution at the @xmath123 , @xmath3 , and @xmath124 cl ( from light to dark shading ) for 1 dof.,title="fig:",scaledwidth=70.0% ] [ fig : resolve - atm - samples ] finally , moving to the @xmath89-solution , we find that in most cases this `` double '' degenerate solution can be excluded if either the @xmath73- or the @xmath88-solution is ruled out .
however , for certain parameter configurations an interesting degeneracy between the octant and the mass ordering appears , such that the @xmath89-solution can not be excluded , although the @xmath73- and the @xmath88-solutions are not allowed .
we find from figure [ fig : resolve ] that it is difficult to distinguish a normal mass hierarchy and @xmath135 , @xmath136 from an inverted hierarchy and @xmath137 , @xmath138 .
we observe from the upper panels of figure [ fig : resolve - atm - samples ] that for these particular parameters sub - gev as well as multi - gev data can not exclude the @xmath89-solution .
however , the appearance of this remaining degeneracy strongly depends on the true value of @xmath6 .
for example , from the lower panels of figure [ fig : resolve - atm - samples ] it becomes clear that it does not occur for @xmath139 .
the dependence on @xmath6 indicates that again the interference term between @xmath12- and @xmath11-effects is important .
in this section we discuss in detail the possibility to identify the type of the neutrino mass ordering from a combined lbl+atm analysis . to this aim
we simulate data for the lbl experiment as well as for atmospheric neutrino data assuming a `` true '' neutrino mass ordering
. then we fit these data with the `` wrong '' hierarchy and search for the minimum @xmath43-value in the full 6-dimensional space of oscillation parameters , taking into account both @xmath0-degenerate solutions @xmath88 and @xmath89 . if this minimum is larger than a certain value the wrong hierarchy can be excluded at the corresponding cl . before considering the general case we will first adopt the choice @xmath140 , such that the @xmath0-degeneracy is absent , and we are left only with the two - fold ambiguity related to sign(@xmath2 ) .
we show the sensitivity to the hierarchy in figure [ fig : hierarchy ] as a function of the true values of @xmath5 and @xmath6 .
first we observe from this figure that the sensitivity of lbl data alone strongly depends on the true value of @xmath6 .
for certain values of @xmath6 the ability to distinguish normal and inverted hierarchy is even lost for @xmath141 .
the main reason for the difficulties of lbl alone to determine the mass hierarchy , is that for the matter effect is very small .
atm data on their own allow to identify the normal mass hierarchy at @xmath3 cl for @xmath142 ( see left panel of figure [ fig : hierarchy ] ) , whereas there is rather poor sensitivity to the inverted hierarchy ( right panel of figure [ fig : hierarchy ] ) .
however , the sensitivity is significantly increased by combining the two data sets : for lbl+atm data the wrong hierarchy can be excluded at @xmath3 cl for @xmath143 , with a rather small dependence on the true value of @xmath6 or the true type of the hierarchy . and @xmath6 for @xmath140 , if the true hierarchy is normal ( left ) or inverted ( right ) .
we show the contours of @xmath122 between the wrong and the true hierarchy , corresponding to a rejection of the wrong hierarchy at the @xmath123 , @xmath3 , and @xmath124 cl ( from light to dark shading ) for 1 dof .
the shaded regions correspond to lbl+atm data combined , solid curves correspond to lbl - only , and dashed curves to atm - only.,scaledwidth=95.0% ] [ fig : hierarchy ] these results can be understood qualitatively from the discussion given in section [ sec : qual - atm - hier ] . to achieve the same value of @xmath111 as implied by the ( true ) normal hierarchy and a given @xmath125 with the inverted hierarchy
, one has to adopt values of @xmath5 larger than @xmath125 to increase the effect for the inverted hierarchy .
conversely , if the true hierarchy is inverted and one wants to fit @xmath111 with the normal ordering , @xmath5 has to be smaller than @xmath125 .
this expectation is confirmed in figure [ fig : delta - theta ] , where the allowed regions in the ( @xmath144 ) plane from atm data are shown as dashed curves for the true values @xmath57 and @xmath58 .
the left panel shows that the normal hierarchy can be fitted with the inverted one for rather large values of @xmath5 , whereas the opposite situation is visible for the inverted hierarchy in the right panel . )
plane assuming for the true mass spectrum a normal ( left ) and an inverted ( right ) hierarchy , and @xmath140 .
the true values @xmath62 and @xmath63 are marked by a star .
the shaded areas and the solid curves correspond to the 1 , 2 , and 3 @xmath61 allowed regions ( 2 dof ) for the true and the sign(@xmath145)-degenerate solution for lbl+atm and lbl data , respectively .
the dashed curves correspond to the @xmath66 contours of atm data for the wrong sign of @xmath145 .
the @xmath66 of the best fit point with the wrong hierarchy for lbl+atm data is given in the figure and its location is marked by a dot .
, scaledwidth=95.0% ] [ fig : delta - theta ] this behavior explains also the atm - only results shown in figure [ fig : hierarchy ] .
the reasonable sensitivity of atm data to the normal hierarchy visible in the left panel is given by the fact , that a fit with the inverted hierarchy would be only possible for very large values of @xmath5 .
this is disfavored by data , first because the values of @xmath5 required to fit the data start to get in conflict with the chooz bound , and hence are disfavored , and second because such large values of @xmath5 start to be excluded from atm data on their own .
on the other hand the reason for the poor sensitivity to the inverted hierarchy shown in the right panel of figure [ fig : hierarchy ] is that in general small values of @xmath111 are expected , which can be easily fitted with the normal hierarchy but smaller @xmath5 ( compare right panel of figure [ fig : delta - theta ] ) , well below the sensitivity of atm data .
let us now discuss the effect of the sign(@xmath2)-degeneracy for lbl data .
in agreement with the discussion of section [ sec : qual - lbl ] we observe from figure [ fig : delta - theta ] that the sign(@xmath145 ) degeneracy affects mainly the determination of @xmath6 , whereas the value of @xmath5 of the wrong - sign solution practically coincides with the true one . according to eq .
( [ eq : fake - delta - h ] ) we expect for the fake solution the value @xmath146 , which is in good agreement with the values obtained in the actual fit : @xmath147 for true hierarchy normal ( inverted ) . the deviations from the value of @xmath148 given by eq .
( [ eq : fake - delta - h ] ) are due to the matter effect , which is not included in the probability eq .
( [ eq : emuprob ] ) and to some extent also to spectral information , which in general is difficult to include in the analytical discussion .
note that because of the choice @xmath140 the @xmath88- and @xmath89-solutions coincide , and we are left only with the two - fold ambiguity implied by the sign(@xmath2)-degeneracy , as visible in figure [ fig : delta - theta ] . from figure
[ fig : delta - theta ] also the reason for the significant improvement of the sensitivity to the hierarchy for the lbl+atm combination becomes clear .
although for lbl data alone the degenerate solution is present at a wrong value of @xmath6 even at @xmath123 cl , the value of @xmath5 has to be very similar to the true one .
this makes it impossible to accommodate atm data with the wrong hierarchy , since either values @xmath5 significantly larger or smaller than the true one are necessary to obtain the correct value for @xmath111 with the wrong hierarchy .
the location in the ( @xmath144 ) plane required by the degenerate solution of lbl gives for atm data a @xmath43 corresponding to @xmath149 cl .
therefore the combined data lead to the rather good sensitivity implied by the @xmath66-values given in figure [ fig : delta - theta ] . to summarize ,
the example shown in figure [ fig : delta - theta ] demonstrates how the combination of lbl and atm data leads to a sensitivity to the mass hierarchy at the @xmath124 level , although each data set on its own can not distinguish normal from inverted mass ordering for this particular choice of parameter values .
the sensitivity of atm data alone suffers from the fact that the wrong hierarchy can be accommodated by adjusting @xmath11 , but also the parameters @xmath0 , @xmath2 , and @xmath6 are important .
to benefit from the earth matter effects in @xmath1-like events from atmospheric neutrinos for the hierarchy determination the rather precise measurement of the oscillation parameters from lbl data is mandatory . in particular , it is necessary to know @xmath0 and @xmath2 at the 1% level , and a reasonable constraint on @xmath5 is required .
finally , the determination of the correct mass hierarchy through the lbl+atm combination allows to resolve the ambiguity in @xmath6 from the sign(@xmath2)-degeneracy , and hence the potential to measure @xmath6 is significantly increased .
for the example shown in figure [ fig : delta - theta ] the correct value of @xmath6 can be identified at @xmath3 cl . and
@xmath91 for @xmath150 and a true normal ( left panel ) and inverted ( right panel ) hierarchy .
we show the contours of @xmath122 between the wrong and the true hierarchy , corresponding to a rejection of the wrong hierarchy at the @xmath123 , @xmath3 , and @xmath124 cl ( from light to dark shading ) for 1 dof .
the shaded regions correspond to lbl+atm data combined , solid curves correspond to lbl - only , and dashed curves to atm - only.,scaledwidth=95.0% ] [ fig : h.sq23 ] now we generalize the discussion of the previous subsection to non - maximal values of @xmath59 , such that in general all four solutions of eq .
( [ eq : solutions ] ) are present . to identify the correct mass hierarchy the @xmath88- as well as the @xmath89-solution has to be excluded . in figure
[ fig : h.sq23 ] we show the sensitivity to the mass hierarchy as a function of the true values of @xmath5 and @xmath91 for lbl , atm , and lbl+atm data .
first we note that the results for lbl data - only strongly depend on the true value of @xmath6 , as observed already in figure [ fig : hierarchy ] .
atm data on their own provide only a reasonable sensitivity to the mass hierarchy if the true hierarchy is normal and @xmath151 .
this is the region of large excess of @xmath1-like multi - gev events ( see discussion in section [ sec : qual - atm - hier ] ) , which can not be achieved for any configuration in the inverted hierarchy .
the sensitivity to the hierarchy for the lbl+atm combination can be inferred from figure [ fig : resolve ] by considering the intersection of the shaded region ( @xmath89-solution ) , with the region to the right of the dashed curves ( @xmath88-solution ) .
the shaded regions in figure [ fig : h.sq23 ] give an explicit example for @xmath132 .
for the true inverted hierarchy ( right panel ) the sensitivity to determine the mass hierarchy is given by the ability to exclude the @xmath88-solution .
as discussed in section [ sec : degeneracies ] the worsening of the sensitivity for low values of @xmath91 can be understood from the flux factor @xmath131 in the multi - gev @xmath1-like event excess . if the true hierarchy is normal ( left panel ) the @xmath88- as well as the @xmath89-solutions have to be taken into account .
the `` spike ''
visible in figure [ fig : h.sq23 ] at @xmath152 comes from the fact that the double degenerate solution @xmath89 can not be excluded . of the wrong hierarchy , panel ( b ) and ( c ) show the values of @xmath5 and @xmath91 at the best fit point of the wrong hierarchy as a function of the true value of @xmath91 .
the true hierarchy is normal and @xmath153 , @xmath132 .
, scaledwidth=60.0% ] [ fig : h - bfp ] we illustrate this behavior in figure [ fig : h - bfp ] , where we show the location of the best fit point for the wrong hierarchy as a function of the true value of @xmath91 .
we observe from panel ( c ) that for @xmath151 the minimum for lbl+atm follows the true value of @xmath91 , _ i.e. _ , it corresponds to the @xmath88-solution . as discussed in section [ sec : degeneracies ] this implies in turn that @xmath5 has to be close to the true value , as can be seen from panel ( b ) .
however , in that case the large values of @xmath111 from the normal hierarchy can not be obtained in the inverted hierarchy , which leads to the large @xmath43 values , _
i.e. _ good sensitivity , in that region . for @xmath154
one observes from figure [ fig : h - bfp ] that the best fit moves to the @xmath89-solution , characterized by the wrong @xmath0 , see panel ( c ) , and a value of @xmath5 given by eq .
( [ eq : fake - th13-th23 ] ) , see panel ( b ) . as mentioned in section [ sec : resolve ] the occurrence of this degeneracy strongly depends on the true value of @xmath6 , and follows from a delicate interplay of effects in sub- and multi - gev data . in the region of @xmath155 the best fit returns again to the @xmath88-solution , characterized by the true @xmath0 and a value of @xmath5 close to the true one . from figure
[ fig : h - bfp ] we find that the fit of lbl - only , as well as atm - only chooses best fit values of @xmath0 and @xmath5 close to the true ones .
this suggests that in that region the sensitivity comes from effects related to @xmath156 , and @xmath6 . as mentioned in section [ sec : resolve ] here the main sensitivity comes from the atm sub - gev data sample ( compare figure [ fig : resolve - atm - samples ] ) and the interference term between @xmath12- and @xmath2-effects @xcite can become important , which introduces the dependence on @xmath6 . and @xmath6 for @xmath157 ( left ) and @xmath158 ( right ) .
the true hierarchy is normal .
we show the contours of @xmath122 between the wrong and the true hierarchy , corresponding to a rejection of the wrong hierarchy at the @xmath123 , @xmath3 , and @xmath124 cl ( from light to dark shading ) .
the shaded regions correspond to lbl+atm data combined , solid curves correspond to lbl - only , and dashed curves to atm - only.,scaledwidth=95.0% ] [ fig : h - nonmax ] the strong dependence of the sensitivity on @xmath6 for small @xmath91 is visible in the left panel of figure [ fig : h - nonmax ] , where we show the sensitivity to the hierarchy as a function of @xmath125 and @xmath47 for @xmath159 .
the rich structure visible for the lbl+atm combination indicates a complicated interplay and/or cancellations of various effects .
on the other hand , from the right panel we observe that for large values of @xmath126 the sensitivity to the hierarchy becomes practically independent of the true value of @xmath6 . in this
range multi - gev @xmath2-effects are enhanced by the flux factor and dominate over @xmath12-effects and the interference term .
also atm data alone provide quite a good sensitivity , thanks to the big effect in multi - gev @xmath1-like events .
as discussed in section [ sec : degeneracies ] , the presence of the octant degeneracy leads to severe complications in the determination of @xmath5 . in this section
we consider the specific case @xmath161 , and we investigate how the upper bound on @xmath5 ( `` sensitivity to @xmath5 '' ) can be improved by the combined lbl+atm analysis .
the fact that the @xmath0-degeneracy affects the @xmath5-sensitivity is well - known , see _
e.g. _ , refs .
the choice @xmath105 essentially eliminates the differences between normal and inverted hierarchies , _
i.e. _ the degeneracy is perfect , and the fact that it can not be resolved introduces only very small ambiguities for other parameters . hence the @xmath73- and @xmath89-solutions coincide within good accuracy .
moreover , the analysis is independent of the true value of @xmath6 , since the phase becomes unphysical for @xmath162 .
we define the sensitivity to @xmath5 as the largest value of @xmath5 which fits the data generated for a true value @xmath163 at a given cl ( see appendix c of ref .
@xcite for a detailed discussion ) . as a function of the true value of @xmath91 for lbl data only ( dashed ) , and the combination lbl+atm ( solid ) . in panels ( a ) and ( b )
we restrict the fit of @xmath0 to the octant corresponding to @xmath59 and @xmath164 , respectively , whereas panel ( c ) shows the overall sensitivity taking into account both octants.,scaledwidth=95.0% ] [ fig : th13sens ] let us first assume that the true octant of @xmath0 was known .
in that case one expects from eq .
( [ eq : emuprob ] ) that the sensitivity for @xmath5 becomes better for increasing @xmath91 , since for large @xmath91 the first term in eq .
( [ eq : emuprob ] ) , which is proportional to @xmath5 , is enhanced .
moreover , the @xmath76 term gets suppressed for small @xmath103 , which reduces the effect of multi - parameter correlations induced by @xmath12 .
this expectation is confirmed in panel ( a ) of figure [ fig : th13sens ] , where the sensitivity to @xmath5 is shown as a function of the true value of @xmath91 , assuming that the octant is known .
however , if the octant of @xmath0 is not known , the degenerate solution prevents the smaller @xmath5-limits implied by @xmath151 . in that case
the data can be fitted by relatively large values of @xmath5 , since exchanging @xmath91 and @xmath103 reduces the effect of @xmath5 and increases the @xmath76 term in eq .
( [ eq : emuprob ] ) .
this effect is shown by the curves corresponding to lbl data in panel ( c ) of figure [ fig : th13sens ] .
we observe that the sensitivity to @xmath5 gets worse for non - maximal values of @xmath59 in both octants because of the presence of the degenerate solution .
since the octant - degeneracy can be efficiently resolved by atmospheric data one expects a significant improvement of the @xmath5-sensitivity for the combined lbl+atm analysis . in panel
( b ) of figure [ fig : th13sens ] we show the @xmath5-limit constraining the fit to the _ wrong _ octant of @xmath0 . as expected from figure [ fig : wrong - th23 ] there are no solutions for lbl+atm data if @xmath0 is sufficiently far from maximal , since the wrong solution is disfavored by atm data .
consequently , we find that the final result shown in panel ( c ) for lbl+atm is very close to the situation in panel ( a ) , where only the true solution for @xmath0 has been used .
hence , the @xmath5-sensitivity is significantly increased for @xmath104 through the exclusion of the wrong octant solution by atm data .
there is no relevant improvement for @xmath19 , since in that case the sensitivity to @xmath5 suffers for the true @xmath0 solution because of the @xmath5-suppression implied by the small values of @xmath91 .
in this work we have performed a combined analysis of future long - baseline ( lbl ) and atmospheric ( atm ) neutrino data . as a specific example we have considered the phase ii of the t2k experiment , consisting of a 4 mw superbeam produced at the j - parc facility .
the 1 mt hyperkamiokande water cherenkov detector will serve as detector for the lbl experiment and simultaneously provide high statistics atm neutrino data .
we have shown that the combined lbl+atm analysis offers a very appealing possibility to resolve parameter degeneracies in the lbl data .
in particular , the ambiguities implied by the sign(@xmath2)- and the @xmath0-degeneracies can be lifted to a large extent .
this becomes possible through three - flavor effects in atm data related to @xmath11 and @xmath12 .
a systematic scan of the parameter space has been performed to investigate the ability to determine the type of the neutrino mass ordering .
let us summarize our main findings : * for true values of @xmath129 the correct mass hierarchy can be identified at @xmath3 cl if @xmath165 , rather independent of the true value of @xmath6 or the type of the true hierarchy . in this region
the sensitivity is dominated by multi - gev @xmath1-like events in atm data . * for true values of @xmath130 the correct mass hierarchy
can be identified at @xmath3 cl if @xmath166 , where here the actual sensitivity depends on the true value of @xmath6 and the true hierarchy . the final sensitivity emerges from an interplay of various effects in the different atm data samples . *
the solution with the wrong octant of @xmath0 can be excluded at @xmath124 cl if @xmath167 , independent of the true values of @xmath11 , @xmath6 , and the hierarchy .
this follows mainly from @xmath12-effects in @xmath1-like sub - gev atm data . if @xmath168 and the true hierarchy is normal the octant sensitivity at @xmath124 cl improves to @xmath169 due to @xmath11-effects in the multi - gev atm data .
* the lifting of the degeneracies by atm data significantly increases the performance of the lbl experiment for the measurement of @xmath5 and @xmath6 , since fake solutions implied by the degeneracies can be ruled out .
generically , the determination of the correct octant of @xmath0 removes an ambiguity in the measurement of @xmath5 , whereas lifting the sign(@xmath2)-degeneracy allows the identification of the correct value of @xmath6 . * if no finite value for @xmath11 is found the upper limit on @xmath5 is significantly improved by resolving the @xmath0-degeneracy by atm data if @xmath20 .
let us stress that these results follow from the complementarity of the two data sets ; neither lbl data alone nor atm data alone can provide a comparable physics reach .
the lbl data allow a very precise determination of @xmath13 and @xmath18 , and although @xmath11 and @xmath6 suffer from ambiguities related to the degeneracies they are constrained to rather specific values .
however , there is very poor sensitivity to the mass hierarchy and to the octant of @xmath0 for lbl data alone . for atm data alone only for @xmath170 and a true normal hierarchy a reasonable sensitivity exists for the mass ordering , since in this case
a big excess of @xmath1-like multi - gev events is predicted , which can not be achieved by any configuration within the inverted hierarchy . for all other regions in the parameter space
the sensitivity of atm - only is rather poor , since the data can be fitted with the wrong hierarchy by adjusting the oscillation parameters . to benefit from the hierarchy sensitivity offered by @xmath11-earth matter effects in atm data the precise measurement of the oscillation parameters from lbl
data is mandatory . in particular
, it is necessary to know @xmath0 and @xmath13 at the 1% level , and a reasonable constraint on @xmath5 is required . hence , only the _ combination _ of lbl and atm data makes it possible to obtain a good sensitivity to the neutrino mass hierarchy .
let us add that to benefit from the lbl+atm combination indeed a few @xmath171 of atm data are necessary .
in particular we have checked that for exposures below @xmath172 the sensitivity to the hierarchy is essentially lost .
finally , let us remark that the methods to resolve parameter degeneracies discussed previously in general involve the combination of two or more ( expensive ) experiments , _
e.g. _ at different baselines or using different oscillation channels .
in contrast , once a lbl experiment with a megaton cherenkov detector is built , atm data come for free .
therefore , in addition to the interesting physics , the synergies between lbl and atm data offer a relatively economical way of resolving parameter degeneracies . | in this work we show that the physics reach of a long - baseline ( lbl ) neutrino oscillation experiment based on a superbeam and a megaton water cherenkov detector can be significantly increased if the lbl data are combined with data from atmospheric neutrinos ( atm ) provided by the same detector .
atm data are sensitive to the octant of @xmath0 and to the type of the neutrino mass hierarchy , mainly through three - flavor effects in @xmath1-like events .
this allows to resolve the so - called @xmath0- and sign(@xmath2)-parameter degeneracies in lbl data . as a consequence it becomes possible to distinguish the normal from the inverted neutrino mass ordering at @xmath3 cl from a combined lbl+atm analysis if @xmath4 . the potential to identify the true values of @xmath5 and the cp - phase @xmath6
is significantly increased through the lifting of the degeneracies .
these claims are supported by a detailed simulation of the t2k ( phase ii ) lbl experiment combined with a full three - flavor analysis of atm data in the hyperkamiokande detector .
madph-05 - 1412 + yitp - sb-04 - 69 + sissa 3/2005/ep + * resolving parameter degeneracies in long - baseline experiments by atmospheric neutrino data * * patrick huber , michele maltoni and thomas schwetz * _ _
@xmath7department of physics , university of wisconsin + 1150 university avenue , madison , wi 53706 , usa + @xmath8c.n .
yang institute for theoretical physics , suny at stony brook + stony brook , ny 11794 - 3840 , usa + @xmath9scuola internazionale superiore di studi avanzati + via beirut 24 , i34014 trieste , italy |
the oxygen abundance is one of the fundamental characteristics of a galaxy .
accurate oxygen abundances are mandatory in investigations of different aspects of the formation and evolution of galaxies .
accurate oxygen abundances in hii regions can be derived via the classic t@xmath4 method . in the first step ,
the electron temperature @xmath5 within the [ oiii ] zone and the electron temperature @xmath6 within the [ oii ] zone are determined .
then the abundance is derived using the equations linking the ionic abundances to the measured line intensities and electron temperature .
( the usual notation ( o / h)@xmath7 for the oxygen abundance derived with the t@xmath4 method will be used throughout the paper . )
the ratio of nebular to auroral oxygen line intensities q@xmath8 = [ oiii]@xmath9/[oiii]@xmath10 is used for the t@xmath3 determination .
the ratio of nebular to auroral oxygen line intensities q@xmath11 = [ oii]@xmath123727/[oii]@xmath127320+@xmath127330 or the ratio of nebular to auroral nitrogen line intensities q@xmath13 = [ nii]@xmath14/[nii]@xmath15 are used for the t@xmath0 determination .
unfortunately , the auroral lines are faint and drop below detectability in the spectra of high - metallicity hii regions .
this prevents the application of the t@xmath4 method to high - metallicity hii regions .
is there another way to estimate electron temperature in hii region where the faint auroral lines are not detected ?
the law of energy conservation for free electrons ( e.g. * ? ? ?
* ) shows that the electron temperature in a nebula is mainly defined by the hardness of the ionizing radiation and by the emission in the forbidden lines .
since the excitation parameter p is an indicator of the hardness of the ionizing radiation , one can expect that the excitation parameter p coupled with the measured intensities of the nebular oxygen lines can be used to estimate the electron temperature in hii regions .
we have suggested that the parametric calibration links the oxygen abundance to the measured nebular oxygen line intensities and the excitation parameter p ( p calibration or p method ) @xcite .
( the oxygen abundance derived with the p calibration will be referred to as ( o / h)@xmath16 . ) comparisons between ( o / h)@xmath7 and ( o / h)@xmath16 abundances in hii regions with direct measurement of the electron temperature as well as the comparison between radial distributions of ( o / h)@xmath7 and ( o / h)@xmath16 abundances in the disks of a few well studied spiral galaxies show that ( o / h)@xmath16 abundances are in agreement with ( o / h)@xmath7 abundances @xcite .
this indirectly confirms that the physical conditions in the hii region can be estimated via the nebular oxygen line intensities and the excitation parameter p. ( or one
may also say that the physical conditions in the hii region can be estimated via the nebular oxygen line intensities only , since the excitation parameter p is expressed in terms of these lines . )
if this is the case , then one can expect that the diagnostic line ratios can be expressed in terms of the intensities of the oxygen nebular lines .
indeed , it has been found empirically that there is a relationship ( the ff relation ) between auroral [ oiii]@xmath10 and nebular oxygen line intensities in spectra of hii regions @xcite . in other words , the diagnostic line ratio q@xmath8 , used for the determination of the electron temperature within the [ oiii ] zone ,
can be expressed in terms of the nebular oxygen line intensities .
it has also been suggested that the electron temperature @xmath6 within the [ oii ] zone is related to the electron temperature @xmath5 within the [ oiii ] zone @xcite .
then one can expect that the diagnostic line ratio used for the determination of the electron temperature within the [ oii ] zone , can also be expressed in terms of the nebular oxygen line intensities .
the goal of this paper is to establish this relation .
the paper is organized as follows . a relation between the ratio of nebular to auroral nitrogen line intensities and oxygen nebular line intensities , q@xmath13 = f(r@xmath0,p ) ,
is derived in section 2 . a relation between electron temperatures t@xmath0 and
t@xmath3 is discussed in section 3 .
we summarize our conclusions in section 4 .
we will be using the following notations throughout the paper : r@xmath0 = i@xmath17 \lambda 3727 + \lambda 3729}}$]/i@xmath18 , r@xmath3 = i@xmath19 \lambda 4959 + \lambda 5007}}$]/i@xmath18 , r = i@xmath19 \lambda 4363}}$]/i@xmath18 . with these definitions ,
the excitation parameter p can be expressed as : p = r@xmath3/(r@xmath3+r@xmath0 ) .
the electron temperature @xmath6 within the o@xmath20 zone in hii regions can be derived from the diagnostic line ratio q@xmath11 or from the diagnostic line ratio q@xmath13 .
it is believed that the measured q@xmath13 results in more reliable t@xmath0 value .
starting from our basic idea that the diagnostic line ratio can be expressed in terms of the strong oxygen nebular line intensities , we will search for a relationship of the type q@xmath13 = f(r@xmath0,p ) here .
it should be noted that the ratio of nebular to auroral nitrogen line intensities q@xmath13 = [ nii]@xmath14/[nii]@xmath15 is an indicator of the electron temperature t@xmath0 and therefore this ratio depends on the electron temperature but not on the nitrogen abundance .
recent measurements of the oxygen and nitrogen line intensities in high - metallicity ( 12+logo / h @xmath21 8.25 ) hii regions were taken from @xcite and @xcite .
these spectroscopic data ( 40 data points ) form the basis of the present study .
the value of q@xmath13 is shown as a function of the intensity of the nebular oxygen line r@xmath22 in fig .
[ figure : ff ] ( the points used in the determination of the final relation are shown ) .
the filled circles are hii regions with 0 @xmath23 p @xmath23 0.3 .
the plus signs are those with 0.3 @xmath23 p @xmath23 0.6 .
the open circles are those with p @xmath21 0.6 .
inspection of fig .
[ figure : ff ] shows that the value q@xmath13 is linked to the intensity of the nebular oxygen line r@xmath22 through a relation of the type @xmath24 ^ 2 \nonumber \\ & + & b_1 \ , \log ( 1-{\rm p } ) + b_2 \ , [ \log ( 1-{\rm p})]^2 .
\label{equation : ffa}\end{aligned}\ ] ] using a sample of hii regions with measurements of oxygen and nitrogen line intensities , the values of the coefficients in eq.([equation : ffa ] ) can be derived .
the values of the coefficients in eq.([equation : ffa ] ) are derived by using an iteration procedure . in the first step ,
the relation is determined from all data using the least - square method .
then , the point with the largest deviation is rejected , and a new relation is derived .
the iteration procedure is pursued until two successive relations have all their coefficients differing by less 0.001 and the absolute value of the largest deviation is less than 0.1 dex .
the following relation was obtained @xmath25 ^ 2 \nonumber \\ & + & 1.085 \ , \log ( 1-{\rm p } ) + 0.382 \ , [ \log ( 1-{\rm p})]^2 .
\label{equation : ff}\end{aligned}\ ] ] the final relation was obtained using 31 data points out of an original set of 40 points .
the relations corresponding to eq.([equation : ff ] ) for different values of the excitation parameter are shown in fig .
[ figure : ff ] by the solid ( p = 0.0 ) , long - dashed ( p = 0.3 ) , short - dashed ( p = 0.6 ) , and dotted ( p = 0.9 ) lines .
the differences @xmath26(logq@xmath13 ) between measured values of logq@xmath27 and those logq@xmath28 computed with obtained relation are shown as a function of oxygen line intensity in fig.[figure : f - df ] ( top panel ) and as a function of excitation parameter p ( bottom panel ) .
the points used in the determination of the final relation are shown by filled circles .
the line is the linear best fit to those data derived through the least squares method . for comparison
, we also show by open circles the objects that have been rejected by the iteration procedure . fig .
[ figure : f - df ] shows that the deviations do not show a correlation either with oxygen line intensity or with excitation parameter .
examination of fig.[figure : ff ] and fig.[figure : f - df ] shows that eq.([equation : ff ] ) gives a satisfactory fit to the observational data .
it should be noted however that the particular form of the analytical expression adopted here may be questioned . we have chosen a simple form , eq.([equation : ffa ] ) , but perhaps a more complex expression may give a better fit to the data .
furthermore , the error in measurements of the line [ nii]@xmath15 is in excess of 20% ( as indicated in original papers ) in a number of calibrating hii regions .
perhaps more accurate measurements may result in a more precise relation .
clearly , high - precision measurements of oxygen and nitrogen lines in spectra of hii regions are needed to check the derived relation5755 is [email protected] on a scale where i@xmath30=1 , and eq.([equation : ff ] ) results in 0.00600 .
the measured intensity of [ oiii]4363 is [email protected] , and eq.([equation : ffo ] ) results in 0.00267 .
thus , the differences between measured line intensities and those predicted by our relation is not in excess of the reported uncertainty of the measurement both for [ nii]5755 and [ oiii]4363 lines . ] . thus , eq.([equation : ff ] ) confirms our basic idea that the diagnostic line ratio , which is an indicator of the electron temperature t@xmath0 , can be expressed in terms of the oxygen nebular line intensities .
since the oxygen nebular lines are strong and , consequently , are easily observable , eq.([equation : ff ] ) solves the problem of the electron temperature determination in o@xmath20 zone of hii region where the faint auroral nitrogen line is not detected .
it should be emphasized that the obtained relation q@xmath13 = f(r@xmath0,p ) is a purely empirical relation in the sense that this is the relation between directly measured values and , consequently , this relation is not based on any assumption .
here we will consider the t@xmath0
t@xmath3 relation for our sample of hii regions .
the measured q@xmath13 values are used to determine the electron temperatures t@xmath0 . to establish an expression that links the electron temperature t@xmath0 to the value of the q@xmath13
, we have performed a five - level - atom calculation using recent atomic data .
the einstein coefficients for spontaneous transitions and the energy levels for five low - lying levels were taken from @xcite .
the effective cross sections or effective collision strengths for electron impact were taken from @xcite .
the effective cross sections are continuous functions of temperatures and are tabulated by @xcite at fixed temperatures .
the actual effective cross sections for a given electron temperature used here were derived from two - order polynomial fits of the data from @xcite as a function of temperature .
the expression for the h@xmath31 emissivity was taken from @xcite .
the five - level - atom solution for ion n@xmath20 results in the following simple expression for the determination of t@xmath0 @xmath32 the majority of extragalactic hii regions are in the low - density regime @xcite .
therefore the electron density n@xmath4 = 100 @xmath33 was adopted for the five - level - atom calculation , and for this reason there is no n@xmath4-term in eq.([equation : t2n ] ) .
the electron temperature t@xmath3 is derived from the q@xmath8 = r@xmath34/r ratio .
the relation between t@xmath3 and q@xmath8 is based on the five - level - atom solution .
the einstein coefficients for spontaneous transitions and the energy levels were taken from @xcite , while the effective cross sections for electron impact were taken from @xcite .
again , the actual effective cross sections for a given electron temperature were derived from two - order polynomial fits of the data from @xcite as a function of temperature .
the five - level - atom solution for ion o@xmath2 results in following simple expression for the determination of t@xmath3 @xmath35 it should be noted that the electron temperatures t@xmath3 derived from eq.([equation : t3o ] ) are close to those derived from analogous expressions given in @xcite and in @xcite . fig . [
figure : tt - singl ] shows the t@xmath0 versus t@xmath3 diagram for our sample of hii regions . since the auroral line
[ oiii]@xmath10 is not detected in hii regions from our sample , the ff relation @xmath36 derived in @xcite was used to estimate the value of r. the filled circles are individual hii regions .
the solid line is the linear best fit to those data @xmath37 derived through the least squares method . the mean value of residuals to this t@xmath0
t@xmath3 relation is 0.046 .
recently we have derived a new @xmath6
@xmath5 relation based on the idea that the equation of the t@xmath4 method for o@xmath2/h@xmath20 applied to the o@xmath2 zone and the equation for o@xmath1/h@xmath20 applied to the o@xmath1 zone must result in the same value of the oxygen abundance @xcite .
the following @xmath6
@xmath5 relation has been derived : @xmath38 this relation is shown in fig . [ figure : tt - singl ] by the dashed line .
inspection of fig .
[ figure : tt - singl ] as well the comparison between eq.([equation : ttsingl ] ) and eq.([equation : tt06 ] ) shows that the two @xmath6 @xmath5 relations derived in different ways are close to each other
. generally speaking , one might expect that there is no one - to - one correspondance between t@xmath0 and t@xmath3 temperatures , and that instead the t@xmath0 t@xmath3 relation is a function of the excitation parameter p. we now examine this possibility . fig .
[ figure : t2-t3 ] shows the t@xmath0 versus t@xmath3 diagram for our sample of hii regions .
the filled squares are hii regions with 0 @xmath23 p @xmath23 0.3 .
the open circles are those with 0.3 @xmath23 p @xmath23 0.6 .
the filled circles are those with 0.6 @xmath23 p @xmath23 0.9 .
close examination of fig .
[ figure : t2-t3 ] suggests that the t@xmath0
t@xmath3 relationship is not unique but instead is dependent on and additional parameter
the excitation parameter p.
a fit to the data ( 31 data points ) results in @xmath39 the relations corresponding to eq.([equation : ttnit ] ) for various values of the excitation parameter are shown in fig .
[ figure : t2-t3 ] by the solid ( p = 0.1 ) , long - dashed ( p = 0.3 ) , short - dashed ( p = 0.6 ) , and dotted ( p = 0.9 ) lines .
the mean value of residuals of this t@xmath0
t@xmath3 relation is 0.030 , which is lower by a factor of @xmath40 1.5 than the analogous value associated with the one - dimensional t@xmath0 - t@xmath3 relation .
thus , we have found evidence of that there is no one - to - one correspondence between t@xmath0 and t@xmath3 temperatures ; instead the t@xmath0
t@xmath3 relation is dependent on the excitation parameter .
the validity of the two - dimensional t@xmath0 t@xmath3 relation can be confirmed ( or questioned ) by the consideration of an additional sample of hii regions .
@xcite have extracted from the data release 3 of the sloan digital sky survey ( sdss ) around 4500 spectra of hii regions with an [ oiii ] @xmath10 emission line detected at a level better than 1@xmath41 , and have carefully measured the line intensities in each spectrum .
yuri izotov and natalia guseva have kindly provided me with the total list of their measurements , as only part of these have been published @xcite .
it constitutes one of the largest and most homogeneous data sets now available , being obtained and reduced in the same way .
the ff relation provides a way to select out the hii regions with high quality measurements @xcite ; the discrepancy index d@xmath42 allows us to eliminate low - quality measurements with large d@xmath42 , while retaining high - quality ones with small d@xmath42 .
we have extracted a subsample of high - quality ( with absolute value of d@xmath42 less than 0.01 ) measurements of high - metallicity ( 12+log(o / h ) @xmath21 8.25 ) hii regions .
the subsample of 80 hii regions meets that ( very hard ) precision criterion .
the electron temperature t@xmath34 in those hii regions is derived from the measured q@xmath8 ratio , the electron temperature t@xmath22 is derived from the q@xmath13 ratio given by eq.([equation : ff ] ) .
those data coupled with the data considered above form the total ( extended ) sample of 111 data points . fig .
[ figure : tt - slon ] shows the t@xmath0
t@xmath3 diagram for the total sample of hii regions .
the filled squares are hii regions with 0 @xmath23 p @xmath23 0.3 .
the open circles are those with 0.3 @xmath23 p @xmath23 0.6 .
the filled circles are those with 0.6 @xmath23 p @xmath23 0.9 .
inspection of fig .
[ figure : tt - slon ] confirms that the t@xmath0
t@xmath3 relationship is dependent on the excitation parameter p. a fit to those data ( 111 data points ) gives @xmath43 the relations corresponding to eq.([equation : tt ] ) for different values of the excitation parameter are shown in fig .
[ figure : tt - slon ] by the solid ( p = 0.1 ) , long - dashed ( p = 0.3 ) , short - dashed ( p = 0.6 ) , and dotted ( p = 0.9 ) lines .
examination of eq.([equation : ttnit ] ) and eq.([equation : tt ] ) shows that the t@xmath0 t@xmath3 relations derived from the two different samples of hii regions are very similar and agree within the formal uncertainties .
thus the derived t@xmath0
t@xmath3 relation is rather robust . in the following
, we will adopt as the t@xmath0
t@xmath3 relation @xmath44 thus , the consideration of an additional sample of hii regions strengthens the conclusion that the t@xmath0
t@xmath3 relation is two - dimensional or parametric .
several versions of the @xmath6
@xmath5 relation have been proposed .
a widely used relation is the one by @xcite ( see also @xcite ) based on the hii region models of @xcite .
@xcite has found that the @xmath6
@xmath5 relationship can be parametrized as @xmath45 another relation has been proposed by @xcite , also based on hii region model calculations by @xcite , is : @xmath46 @xcite , who also fitted the hii region models of @xcite , have proposed the following expression : @xmath47 based on hii region model calculations by @xcite ) , @xcite have suggested the following relation : @xmath48 @xcite have found that the campbell et al .
relation is reasonable for @xmath49 @xmath21 1.0 .
however at lower temperatures , the models are more consistent with an isothermal nebula .
they consequently adopted the formulation : @xmath50 we now compare the t@xmath0
t@xmath3 relation obtained here with those obtained by other authors . since our t@xmath0
t@xmath3 relation is derived for cool high - metallicity hii regions , the high - temperature low - metallicity part of relation is not considered here .
the lines in fig .
[ figure : compar ] are t@xmath0 t@xmath3 relations from @xcite .
our one - dimensional t@xmath0
t@xmath3 relation is shown by the plus signs .
our two - dimensional t@xmath0 t@xmath3 relations for different values of the excitation parameter are shown by the triangles ( p = 0.9 ) , the open circles ( p = 0.5 ) , and the open squares ( p = 0.1 ) . examination of fig .
[ figure : compar ] shows that the t@xmath0 t@xmath3 relations of @xcite and @xcite are relatively close to our two - dimensional relation for high excitation hii regions , while at low temperatures , the t@xmath0 t@xmath3 relations of @xcite and @xcite are relatively close to our two - dimensional relation for moderate excitation hii regions .
a relationship between the ratio of nebular to auroral nitrogen line intensities , which is an indicator of the electron temperature t@xmath0 in the o@xmath1 zone of hii regions , and nebular oxygen line intensities in spectra of high - metallicity hii regions was derived . since the oxygen nebular lines are strong and , consequently , are easily observable , the derived relation coupled , with the ff relation derived in our previous studies @xcite , solve the problem of the determination of the electron temperatures t@xmath0 and t@xmath3 in high - metallicity hii regions where faint auroral lines are not detected .
it should be emphasized that the relation obtained here and the ff relation are purely empirical relations in the sence that they are relations between directly measured values .
consequently , there is no assumptions at the base of those relations .
the derived relation and the ff relation confirm our idea that the diagnostic line ratios , which are indicators of the electron temperatures , can be expressed in terms of the strong oxygen nebular lines .
this , in turn , confirms the basic assumption of the `` empirical '' method , proposed by @xcite a quarter of a century ago , that the oxygen abundance in hii regions can be estimated from strong oxygen line measurements only . the relation between electron temperatures in the o@xmath2 and o@xmath1 zones in high - metallicity hii regions was investigated .
it was found that there is no one - to - one correspondance between t@xmath0 and t@xmath3 temperatures . instead the t@xmath0
t@xmath3 relation is dependent on the excitation parameter .
i thank yuri izotov and natalia guseva for providing me with their measurements of the line intensities in spectra of hii regions extracted from the data release 3 of the sloan digital sky survey ( sdss ) , the majority of which are unpublished .
i am grateful to r.b.c .
henry for constructive comments as well as improving the english text .
i thank g.stasiska for usefull discussion .
i am gratefull to an anonymous referee for useful comments that improved the presentation of the paper .
the author acknowledges the work of the sloan digital sky survey ( sdss ) team .
funding for the sdss has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , the korean scientist group , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington .
aller l.h . , 1984 , physics of thermal gaseous nebulae , d. reidel publishing company , dordrecht .
aggarwal k.m . ,
keenan f.p . , 1999 ,
apjs , 123 , 311 bresolin f. , 2006 , apj , in press , ( astro - ph/0610690 ) bresolin f. , garnett d.r . , kennicutt r.c . , 2004 , apj , 615 , 228 bresolin f. , schaerer d. , gonzlez delgado r.m . , stasiska g. , 2005 , a&a , 441 , 981 campbell a. , terlevich r. , melnick j. , 1986 , mnras , 223 , 811 castellanos m. , daz a.i .
, terlevich e. , 2002 , mnras , 329 , 315 deharveng l. , pea m. , caplan j. , costero r. , 2000 , mnras , 311 , 329 galavs m.e . ,
mendoza c. , zeippen c.j . , 1997 , a&as , 123 , 159 garnett d.r . , 1992 , aj , 103 , 1330 hudson c.e . , bell k.l . , 2005 ,
a&a , 430 , 725 izotov y.i .
, stasiska g. , guseva n.g .
, thuan t.x . , 2004 ,
a&a , 415 , 87 izotov y.i . , stasiska g. , meynet g. , guseva n.g .
, thuan t.x . , 2006 ,
a&a , 448 , 955 izotov y.i . ,
thuan t.x .
, lipovetsky v.a . , 1997 , apjs , 108 , 1 kennicutt r.c . , bresolin f. , garnett d.r . , 2003 , apj , 591 , 801 luridiana v. , esteban c. , peimbert m. , peimbert a. , 2002 , rev .
, 38 , 97 oey m.s . , shields j.c . ,
2000 , apj , 539 , 687 pagel b.e.j . , edmunds m.g . ,
blackwell , d.e . ,
, smith g. , 1979 , mnras , 189 , 95 pagel b.e.j . ,
simonson e.a .
, terlevich r.j . , edmunds m.g . , 1992 , mnras , 255 , 325 peimbert a. , 2003 , apj , 584 , 735 pilyugin l.s
a&a , 362 , 325 pilyugin l.s . , 2001a , a&a , 369 , 594 pilyugin l.s . ,
2001b , a&a , 373 , 56 pilyugin l.s .
, 2003 , a&a , 399 , 1003 pilyugin l.s .
, 2005 , a&a , 436 , 1l pilyugin l.s . , ferrini f. , shkvarun r.v .
, 2003 , a&a , 401 , 557 pilyugin l.s . ,
thuan t.x . , 2005 ,
apj , 631 , 231 pilyugin l.s .
, thuan t.x .
, vlchez j.m . , 2006a , mnras , 367 , 1139 pilyugin l.s . , vlchez j.m . , contini t. , 2004 , a&a , 425 , 849 pilyugin l.s . ,
vlchez j.m . ,
thuan t.x . , 2006b , mnras , 370 , 1928 rubin r.h .
, 1986 , apj , 309 , 334 sobolev v.v . , 1967 ,
course of theoretical astrophysics , moscow , nauka stasiska g. , 1982 , a&as , 48 , 299 stasiska g. , 1990 , a&as , 83 , 501 stasiska g. , schaerer d. , 1997 , a&a , 322 , 615 tsamis y.g . ,
barlow m.j . , liu x .- w . ,
danziger i.j . ,
storey p.j . , 2003 , mnras , 338 , 687 zaritsky d. , kennicutt r.c .
, huchra j.p . , 1994 , apj , 420 , 87 | the problem of determination of the electron temperature t@xmath0 in the o@xmath1 zone of high - metallicity hii region was examined .
it was shown that the ratio of nebular to auroral nitrogen line intensities , which is an indicator of the electron temperature t@xmath0 , can be expressed in terms of the nebular line intensities of oxygen .
this solves the problem of the determination of the electron temperature t@xmath0 , since the oxygen nebular lines are strong and , consequently , are readily observable .
a relation between electron temperatures in the o@xmath1 and o@xmath2 zones in high - metallicity hii regions was studied .
it was found that there is no one - to - one correspondance between t@xmath0 and t@xmath3 temperatures . instead the t@xmath0
t@xmath3 relation is dependent on excitation parameter .
galaxies : abundances ism : abundances hii regions |
w49n is a well - known and extensively - studied massive star - forming complex in the milky way .
the oh and h@xmath3o masers in this region represent some of the most luminous such sources in our galaxy .
the w49n complex , in the galactic direction @xmath4 , is located on the far - side of the solar circle .
its large distance ( @xmath011.4 kpc ; gwinn et al . 1992 ) and low galactic latitude together make this object attractive for studying various propagation effects due to the intervening medium .
the interstellar scattering in this case is comparable to that in the vela direction ( gwinn et al .
1993 ) , but the vela pulsar is at a much smaller distance ( @xmath51 kpc ) . in our efforts to study intrinsic short timescale variability in w3oh ( ramachandran et al .
2006 ; also laskar et al .
2012 ) we were able to estimate and remove possible variability due to interstellar scintillation using the fact that the decorrelation bandwidth is larger than the measured line widths .
such velocity - resolved analysis of w3oh data has suggested intrinsic variability on 15 - 20 minute time scale ( ramachandran et al .
2006 ) , whereas w49n data show variations on timescales of 1 hour or longer ( goss et al . 2007 : talk at iaus 242 , alice springs ) .
the timescales of apparent intrinsic variability , combined with light - travel - time - argument , suggest spatial scales for the maser columns to be a few au for the oh sources in w3oh complex ( ramachandran et al .
the reliability of similar estimation in the case of w49n sources depends , however , on how well the contamination due to interstellar scintillation can be separated from the observed variability , particularly when the scintillation decorrelation bandwidth may not be much wider than the line widths .
in fact , for the purpose of this variability study , w49n was included in our vlba observations as a _ comparison _ source , where scattering effects are expected to dominate in the apparent relative variability of different spectral features due to the interstellar scintillations ( with decorrelation bandwidth comparable or smaller than the line widths ) . the subsequent analysis of the data on this source has turned out to be far more fruitful than our initial expectations , revealing several interesting aspects that we describe and discuss in this paper .
measurements of galactic magnetic fields and of their distribution over a range of scales continue to be great interest , given their important role in the physical processes in the interstellar medium .
magnetic fields present in molecular clouds become directly measurable through the zeeman splitting of the narrow spectral lines associated with several species ( e.g. , sarma et al .
2000 , and references therein ) , and those threading across the intervening ionized medium manifest themselves through faraday rotation and anisotropic scattering .
while both the zeeman splitting of nonmaser lines and the faraday rotation depend on the line - of - sight component of the magnetic field , the anisotropy in scattering relates to the field component in the plane of the sky .
however , in cases where the zeeman splitting exceeds the individual line widths ( as is often the case for narrow maser lines such as oh ) , the line separation is then dependent on the _ total _ field strength ( heiles et al . 1993 ) and not just the component along the line of sight . in an earlier study of oh maser sources in w49n , desai , gwinn & diamond ( 1994 ;
hereafter dgd94 ) have found significant anisotropic scattering , attributable to electron - density irregularities that are preferentially elongated in the galactic plane , due to a magnetic field parallel to the plane .
our vlba observations , and the resultant images of several dozen maser spots ( including a few dozen zeeman pairs ) , provide significantly improved statistics to investigate in greater detail the anisotropic scattering , as well as to probe the distribution of magnetic fields within the w49n complex .
some of our findings are in good qualitative agreement with those of dgd94 ; they however differ quantitatively from their estimates of scatter - broadened source sizes as well as the mean source position angle ( hereafter spa ) characterizing the apparent shapes of the maser spots .
these observations provide a rare opportunity to look for another imprint of the magnetic fields in the scattering medium , namely , possible differential diffraction experienced by opposite circular polarization components of propagating electromagnetic waves .
this expectation is based on the same refractive index differences for the two circulars that result in faraday rotation of the position angle of the linear component of incident polarization . our data on 32 zeeman pairs
, providing estimates of magnetic field within the maser column , enables us to study the distribution and variation of the magnetic field in this star - forming complex . in the following section ,
we describe briefly our observations and data which form the basis for the analysis and results discussed in the paper .
section 3 presents our general findings on the anisotropic scattering observed in the direction of the oh sources in w49n .
the data suggesting significant differential anisotropic scattering , probed using zeeman pairs , are described in section 4 .
the structure functions of the magnetic field local to the source , as well as the scattering parameters estimated for the ensemble , are presented in section 5 .
the implications of our results are discussed and summarized in the final section .
our visibility data from 12 + hour synthesis observations with vlba ( br107 ; 2005 july ) were calibrated & processed in the standard way using aips .
the data were self - calibrated ; absolute coordinates are not available due to an absence of phase reference source observations .
the spectral - line image cubes , obtained separately at 1612 , 1665 & 1667 mhz , provided high angular - resolution images of the set of sources for each of the transitions , for each of the circular polarizations . the total imaged angular extent of about 8 arc - seconds corresponds to a transverse span of @xmath0 0.5 pc at the distance of the source , although a majority of the sources are within the central 2 arc - second region . as a result of exclusion of the outer 2 antennas of the vlba ( st .
croix & mauna kea ) from self - calibration , due to lack of visibility of the sources on those baselines in this imaging , the synthesized beam size across ( ra , dec ) is @xmath6 mas .
( the corresponding spatial resolution is about @xmath0175 au at the distance of w49n ) .
the lack of visibility on the longest baselines is consistent with , and is in fact due to , the apparent angular sizes of the maser sources in the range 20 - 40 mas .
each data set consisting of cleaned and restored images across 240 spectral channels spanning a 22 km s@xmath2 velocity range , providing velocity resolution of @xmath00.1 km s@xmath2 , was used to identify discrete maser sources ( avoiding cases with significant velocity gradients ) . for each of
clearly identifiable discrete ( or isolated ) maser sources , best - fit estimates of the _ deconvolved _ source shape and size were obtained , in addition to the estimates of the mean location and velocity , along with estimates of the respective uncertainties . for these estimations , we have used a procedure in aips ( jmfit ) , which facilitates fitting of a 2-d gaussian to a source image and returns the best - fit parameters , including errors . the shape and size together
are parametrized in terms of major & minor axes and position angle associated with a model ellipse describing the cross - section of a 2-d gaussian sampled at half - power points .
although the absolute position information is uncertain by an unknown amount , the relative positions of the sources at both 1665 & 1667 mhz are reliable within the estimation uncertainties , as same self - calibration applies to the data at these two frequencies .
the resultant data on a total of 205 sources ( 194 of these at 1665 & 1667 mhz ) were examined for positional proximity of left and right circular polarization ( lcp and rcp , respectively ) source pairs ( @xmath5 10 mas ) , and 32 zeeman pairs were thus identified .
figure 1 shows the distribution of all discrete sources in the ra - dec plane ; the size of the symbol is proportional to the mean velocity associated with the observed oh line .
the circles and stars denote the sources with rcp and lcp , respectively .
most of the sources are within an area extending from ra - dec offset ( in mas ) of ( -2000,-2000 ) to ( + 4000,+4000 ) .
the velocities ( indicated crudely by the symbol size ) on the average differ systematically between the two halves about the center of this area ( roughly at ra - dec offset of 1000,1000 mas ) .
an overall correlation between the locations of the sources on the sky plane and their line - of - sight velocities , such as that evident for the distribution in the figure 1 , is to be expected in system morphologies dictated by rotation and/or bipolar outflows that are commonly encountered in star - forming regions ( see , for example , shepherd and churchwell 1996 ) .
scatter - broadened shapes and orientations of the maser spots provide valuable information on the nature of the scattering medium .
figure 2 displays the deconvolved sizes of the sources , giving the dimensions along the major and the minor axes of the respective best - fit ellipses .
the elongations or the axial ratios ( ratio of major to minor axis of the best - fit ellipse ) for a majority of the sources are in the range 1.5 to 3 , indicating significant anisotropic scattering , consistent with the finding of dgd94 .
however , we find the overall apparent sizes of the sources to be significantly smaller ( by a factor of @xmath7 2 ) than those reported by dgd94 .
our size estimates are nonetheless consistent with the suitably scaled values of the scatter - broadening of h@xmath3o masers reported by gwinn ( 1994 ) , i.e. about 40 mas ( 200 @xmath8as times [ 18/1.3]@xmath9 ) .
interesting , dgd94 themselves mention about 45 mas as the expected amount of scatter broadening at 18 cm wavelength for the w49n sources .
possible reasons for what appears to be an overestimation of the sizes in dgd94 are entirely unclear at present .
however , we note that the larger apparent angular sizes for the sources reported by dgd94 would necessarily imply significant loss of visibility on a larger fraction of the vlba baselines , an effect which should be readily apparent .
one of the most important aspects discussed by dgd94 ( also see references therein ) is that the apparent image sizes / shapes are expected to be elongated orthogonal to the scattering irregularities due to anisotropic diffraction .
the presence of a magnetic field would induce such anisotropy in the electron - density irregularities ; then the implied field direction would be perpendicular to the resultant image position angle ( i.e. spa ) . based on their limited sample , of the apparent image shapes for 27 spectral components from 6 of the oh maser sources in w49n
, dgd94 suggested that the apparent anisotropy is induced by a magnetic field in and parallel to the galactic plane .
in figure 3 , we present a distribution of our estimates of the spas for our significantly ( @xmath0 30 times ) larger sample . some random spread in spa
is apparent in both our and dgd94 s reported spa values ; such a spread ( @xmath05 degrees r.m.s . ) is not unexpected , given the estimation uncertainties as well as possible differences in the scattering sampled by different lines of sight .
however , we find that our spa distribution has a significant offset from the spa of @xmath0117 degrees , the expected spa if the density irregularities are elongated parallel to the galactic plane .
our sample gives a mean spa of 107@xmath103 degrees , implying a significant mean deviation , of @xmath010 degrees , in the orientation of the density irregularities from the galactic plane .
the dgd94 sample may have been too limited to make such an offset detectable .
we discuss below possible reasons for this systematic offset .
if the large - scale magnetic fields in the intervening region has a 10 degree orientation with respect to the galactic plane , or if the galactic warp ( e.g. , burton 1988 ) is responsible for this offset , then the observed orientations for the scatter - broadened images would be consistent with a correspondingly revised expectation .
however , w49n is located almost on the farther edge of the inner ( solar ) circle of our galaxy , at a galactic longitude of about 43 degrees , and the effect of the warps is observed to be more pronounced in the outer regions of our galaxy . in any case , a look at the magnetic field structure along the w49n direction would be instructive .
interestingly , the sight - line to w49n is through the well - known north polar spur ( nps ) feature ( distance to the nps is about 250 pc ) . from the study , by wolleben ( 2007 ) and others of the nps
, it is evident that significantly different magnetic field structure as well as enhanced scattering would be expected for the medium within 0.5 kpc of the sun .
we note that the field orientation in this region , corresponding to the nps , might be at a large inclination , if not almost orthogonal , to the galactic plane ( see wolleben 2007 for details ) . also , given that the scattering medium closer to the observer is expected to make relatively higher contribution to the angular broadening ( e.g. , gwinn et al .
1993 , deshpande and ramachandran 1998 ) , the above mentioned spa deviation can be caused by the more local scattering in the nps region , with density anisotropy at a possibly large angle with respect to the galactic plane .
the scattering in the w49n direction is unlikely to be dominated by any single thin screen , but is contributed by the entire magneto - ionic medium distributed between the sun and the source .
although a detailed modelling of the magneto - ionic medium incorporating the relative strengths and sense of anisotropic scattering , associated magnetic field strengths and orientations for the different regions along the sight - line is needed to assess the net effect , we have carried out a simple simulation of propagation through multiple screens in order to examine the basic situation .
for each screen , a random column - density distribution of free electrons , following a power - law spatial spectrum ( with kolmogorov index @xmath11 ) was used to simulate a 2-d scattering screen with a mild ( 1.6 ) anisotropy across a transverse extent of a few times the fresnel scale ( @xmath12 , where @xmath13 is the observing wavelength and @xmath14 is the distance to the source ) .
the orientation of the mean anisotropy in the screen nearest the observer was chosen to be orthogonal to those in the rest of the screens , consistent with the possible elongation of density irregularities in nps region with respect to that parallel to the galactic plane .
a model with incoherent combination of the scatter broadening from such screens can not produce a resultant image with an in - between orientation or spa , for example , with 10 deg offset as mentioned above , but would instead dilute the aspect ratio .
hence , the spatial scales of relevant density irregularity in the phase - screens are likely to be very much larger than the sizes estimated based on the observed angular broadening , assuming a single thin screen
. the relevant spatial scale of density irregularity should be adequately large , so that emerging wavefront ( or field ) retains memory of and is the result of the _ coherent _ combination of phase structure contributed by all sub - screens along the sight - line .
the relevant phase structure should necessarily contain higher order variations and/or a finite random spread in orientations of phase curvatures .
thirty two zeeman pairs can be identified from the large sample of sources using a positional - proximity criterion ( @xmath5 10 mas ) . based on these zeeman pairs
we can probe a previously unobserved aspect of scattering caused by the magneto - ionic component of the interstellar medium . a magneto - ionic medium would , in principle , have differing refractive indices for signals corresponding to the two hands of circular polarization .
diffractive scintillation and scatter - broadened image shapes should therefore differ between the lcp and rcp due to any line - of - sight component of the magnetic field in the intervening medium . hence scattering - dominated images of even a randomly polarized source might show circular polarization in unmatched parts of the images , if faraday rotation is significant .
macquart & melrose ( 2000 ) consider this possibility , but estimate the effect to be too small ( circular fraction @xmath15 ) to be observable .
contrary to that expectation , the scatter - broadened images of some of the w49n oh maser sources seem to significantly differ in lcp and rcp , i.e. within a given zeeman pair . for the remainder of the zeeman pairs ,
the spa differences are either small or within the respective uncertainties .
figure 4 shows the observed differences in the image shape parameters for a subset of our sample of zeeman pairs .
the subset includes only those cases for which a significant difference ( @xmath7 6@xmath1 ) in image spa is observed .
although there are only a few such zeeman pairs , the spa differences range between 6 to 30 degrees .
difference in the line velocities within most of these pairs is too small to account for the apparent differential scattering .
position differences are also within a few mas ( i.e. much smaller compared to the scatter broadening ) , and are unlikely to contribute to the observed spa differences .
one such example of a significant spa difference is shown in figure 5 . to explore the issue further
, we have made preliminary attempts to simulate signal propagation through a magneto - ionic medium with a mild anisotropy . here a single , thin scattering screen was simulated , following the procedure outlined in an earlier section . for simplicity ,
a uniform magnetic field is assumed , so that the phase screens for the two circular polarizations are merely scaled versions of each other . the net circular polarization ,
if viewed with coarse resolution would be negligible , consistent with the conclusion of macquart & melrose ( 2000 ) .
however , when observed with adequately high angular resolution ( e.g. 10 milli - arc - second ) , the simulations do reveal noticeable differential diffractive effects , hence differing shapes and sizes of scatter - broadened images for the two circular polarizations .
the differential phase contribution due to the faraday rotation should be significant , even if relatively small in comparison with the phase variation common to both polarizations . the data on rotation measures to several pulsars ( manchester et al .
2005 ; atnf pulsar catalog ) close to the direction of w49n suggest that the expected faraday rotation at 18 cm would correspond to about 30 radians of differential phase between the two circular polarization signals . in comparison ,
the phase structure function ( armstrong , rickett & spangler 1995 ) value at 1 au scale ( close to the fresnel scale , given the distance to w49n and the wavelength of 18 cm ) would suggest an rms phase fluctuation of the order of 300 radians , common to both polarizations .
hence , the differential phase contribution here can lead to noticeable differential scattering .
of course , more detailed and realistic simulations , i.e. with thick screens , are required to assess these issues in detail .
however , the faraday rotation contribution leading to differential anisotropic scattering for the two circular polarization signals remains a more natural explanation for the detected differences in source position angles .
if true , the above would be the first such example of this effect in an astronomical context , as far as we are aware .
the sight - lines to the two hundred or so sources spread across the w49n region potentially sample variations on a range of transverse scales within the region , as well as the intervening scattering medium .
the estimated parameters for this set of scatter - broadened maser sources thus provide an excellent opportunity for probing the spatial scales associated with the variation in these parameters .
however , since this sampling is rather non - uniform , we prefer to compute spatial structure functions using these data instead of trying to estimate the associated spatial power spectra .
although we treat each of the maser sources as a separate source , irrespective of their zeeman pairing , we compute the structure functions separately for the pairs of sources with matching and opposite signs of circular polarization , along with the structure function for the entire source sample . given the limited signal - to - noise ratio in estimation of most of the parameters , the estimated structure functions will have corresponding noise bias . to assess the latter , we also computed the structure function contribution expected from uncertainties in the estimations of the relevant parameters .
figures 6 , 7 and 8 show a few examples of the structure functions estimated in this manner , where the noise bias is relatively small .
specifically , in figure 6 we present the structure function of the ( line - of - sight ) velocity associated with the sources . the clear linear trend in the log - log plot
is consistent with the underlying velocity gradient across the region , manifesting the possible bipolar outflow morphology .
figure 7 shows the structure functions for the scatter - broadened size of the sources , in this case the minor axis associated with the elliptical source shape .
significant variations in the scattering property is apparent on an angular scale of about 200 - 300 mas . for an average distance to the scatterer of about half - way to the source ( 5.7 kpc )
, this angular scale would correspond to a spatial scale of 1000 - 1500 au over which scatter broadening appears to change significantly .
the spatial structure function computed for the magnetic field using estimates available for the set of zeeman pairs , as presented in figure 8 , shows no significant monotonic trend .
this poor sampling of the structure function is largely due to the uneven and sparse sampling of the spatial scales that the available zeeman pairs in this region offer .
however , the scale on which the magnetic field appears to be decorrelated is about 150 mas ( or 1500 au spatial scale at the source distance ) . for 10 out of the 32 pairs ,
the zeeman splitting is smaller than the line widths .
hence in these cases , the estimated magnetic field would correspond to only the line of sight component .
the oh maser sources in w49n show anisotropic scattering ; the apparent scatter broadening is much less than reported earlier .
the implied scintillation decorrelation bandwidth , corresponding to the observed sizes of the w49n sources ( @xmath720 mas ) , would be about 100 hz ( typical line widths are 2.5 khz ) .
this would correspond to a velocity width of 20 m / s , less than our velocity resolution ( 100 m / s ) .
the sizes reported by dgd94 would have implied even narrower decorrelation bandwidths .
hence , we expect interstellar scintillations to be significantly quenched , reducing potentially contaminating contribution to any apparent intensity variability .
the position angles of the source images deviate significantly from the value expected if scattering density irregularities were to be stretched " due to magnetic field strictly aligned parallel to the galactic plane , indicating significant scatter - broadening contribution from differently aligned density irregularities .
this contribution may be associated with the north polar spur , within a distance of 0.5 kpc .
some of our data also reveal differential scattering during propagation of the two circular polarizations , likely caused by faraday rotation . through various aspects discussed above ,
the attractive , but yet unexplored , potential of the high - resolution maser observations for probing the intervening magneto - ionic medium is certainly evident .
the structure function analysis of the various parameters suggests spatial scales of a few thousand astronomical units over which the scattering properties in the intervening medium as well as the magnetic field in the maser region appear to be changing .
it is a pleasure to acknowledge the contributions from r. ramachandran and sarah streb at different stages of the reported work .
we thank don melrose and the anonymous referee for their valuable comments .
aad gratefully acknowledges fruitful discussions with rajaram nityananda , simon ellingsen and john dickey .
the national radio astronomy observatory is a facility of the national science foundation operated under a cooperative agreement by associated universities , inc .
+ _ facilities : _ | our analysis of a vlba 12-hour synthesis observation of the oh masers in a well - known star - forming region w49n has yielded valuable data that enables us to probe distributions of magnetic fields in both the maser columns and the intervening interstellar medium ( ism ) .
the data consisting of detailed high angular - resolution images ( with beam - width @xmath020 milli - arc - seconds ) of several dozen oh maser sources or _
spots _ , at 1612 , 1665 and 1667 mhz , reveal anisotropic scatter broadening , with typical sizes of a few tens of milli - arc - seconds and axial ratios between 1.5 to 3 .
such anisotropies have been reported earlier by desai , gwinn & diamond ( 1994 ) and interpreted as induced by the local magnetic field parallel to the galactic plane .
however , we find a ) the apparent angular sizes on the average a factor of about 2.5 less than those reported by desai et al .
( 1994 ) , indicating significantly less scattering than inferred earlier , and b ) a significant deviation in the average orientation of the scatter - broadened images ( by @xmath010 degrees ) from that implied by the magnetic field in the galactic plane .
more intriguingly , for a few zeeman pairs in our set , significant differences ( up to 6@xmath1 ) are apparent in the scatter broadened images for the two hands of circular polarization , even when apparent velocity separation is less than 0.1 km s@xmath2 .
this may possibly be the first example of a faraday rotation contribution to the diffractive effects in the ism . using the zeeman pairs
, we also study the distribution of magnetic field in the w49n complex , finding no significant trend in the spatial structure function . in this paper , we present the details of our observations and analysis leading to these findings , discuss implications of our results for the intervening anisotropic magneto - ionic medium , and suggest the possible implications for the structure of magnetic fields within this star - forming region . |
fesb@xmath2 is a narrow - gap semiconductor which attracted a lot of attention because of its unusual magnetic,@xcite thermoelectric@xcite and transport properties.@xcite the magnetic susceptibility of fesb@xmath2 is nearly constant at low temperatures with paramagnetic to diamagnetic crossover at around 100 k for a field applied along the @xmath8-axis , similar to fesi.@xcite the electrical resistivity along the @xmath9- and @xmath10 - axes shows semiconducting behavior with rapid increase for t @xmath11 100 k. along the @xmath8 -axis resistivity exhibits a metal to semiconductor transition at around 40 k.@xcite based on the measurements of the electrical resistivity , magnetic susceptibility , thermal expansion , heat capacity and optical conductivity the fesb@xmath2 has been characterized as a strongly correlated semiconductor.@xcite it was also shown that fesb@xmath2 has colossal seebeck coefficient @xmath12 at 10 k and the largest power factor @xmath13 ever reported.@xcite thermal conductivity @xmath14 of fesb@xmath2 is relatively high and is dominated by phonons around 10 k with phonon mean free path @xmath15 m several orders of magnitude larger than electronic mean free path.@xcite in the recent room temperature study we have observed , for the first time , all six raman active modes of fesb@xmath2 predicted by theory.@xcite racu _
et al_.@xcite measured polarized raman scattering spectra of fesb@xmath16 single crystals below room temperature and found only anharmonicity of a@xmath7 and b@xmath6 symmetry modes with no additional electron - phonon coupling . in this work
we have measured at different temperatures polarized raman scattering spectra of pure fesb@xmath2 single crystals and fesb@xmath2 crystals alloyed with co and cr . the b@xmath6 mode asymmetry and
broadening is analyzed using breit - wigner - fano profile model .
the coupling between single phonon and the electronic background is drastically reduced for temperatures bellow 40 k , fully in agreement with transport properties measurements.@xcite alloying of fesb@xmath2 with co and cr also reduces the coupling , i.e. leads to the b@xmath6 mode narrowing .
we have also observed strong a@xmath7 symmetry mode mixing .
single crystals of fesb@xmath2 , fe@xmath0co@xmath1sb@xmath2 and fe@xmath0cr@xmath1sb@xmath2 ( [email protected] ) were grown using the high - temperature flux method , which is described in details in refs.@xcite sample structure and composition were determined by analyzing the powder x - ray diffraction data of fe(co , cr)sb@xmath2 single crystals collected using a rigaku miniflex diffractometer with cu k@xmath18 radiation.@xcite the samples stoichiometry was determined by an energy dispersive jeol jsm-6500 sem microprobe .
analysis of several nominal x=0.25 samples showed that the uncertainty in co and cr concentrations among samples grown from different batches was @xmath19x=0.04 .
the raman scattering measurements were performed using jobin yvon t64000 raman system in micro - raman configuration .
the 514.5 nm line of an ar@xmath20/kr@xmath20 mixed gas laser was used as an excitation source .
focusing of the laser beam was realized with a long distance microscope objective ( magnification @xmath21 ) .
we have found that laser power level of 0.02 mw on the sample is sufficient to obtain raman signal and , except signal to noise ratio , no changes of the spectra were observed as a consequence of laser heating by further lowering laser power . the corresponding excitation power density was less then 0.1 kw/@xmath22 .
all raman scattering measurements presented in this work were performed using the ( 10@xmath23 ) plane of fesb@xmath2 orthorhombic crystal structure .
low temperature measurements were performed between 15 k and 300 k using konti cryovac continuous helium flow cryostat with 0.5 mm thick window .
fesb@xmath2 crystallizes in the orthorhombic marcasite - type structure of the centrosymetric pnnm ( d@xmath24 ) space group , with two formula units ( z=2 ) per unit cell.@xcite basic structural unit is built up of fe ion surrounded by deformed sb octahedra .
these structural units form edge sharing chains along the @xmath8- axis . according to the factor - group analysis
there are 6 raman active modes ( 2a@xmath7 + 2b@xmath6+b@xmath25+b@xmath26 ) , which were observed and assigned in our previous work.@xcite the a@xmath7 and b@xmath6 symmetry modes are bond stretching vibrations , whereas the b@xmath25 and b@xmath26 symmetry modes represent librational ones.@xcite single crystals in the ( x@xmath27y ) configuration ( b@xmath6 symmetry modes ) measured at different temperatures.,scaledwidth=30.0% ] fig .
[ fig2 ] shows raman scattering spectra of fesb@xmath2 single crystals in the ( x@xmath28y ) configuration ( @xmath29 $ ] , @xmath30$])@xcite measured at different temperatures in the spectral range of the highest energy b@xmath6 symmetry mode . for this configuration b@xmath6 and b@xmath26 modes
are raman active , see ref .
one can notice an asymmetry of the b@xmath6 mode towards lower wave numbers .
this broad , asymmetric structure is analyzed using a breit - wigner - fano ( bwf ) interference model.@xcite the resonance usually involves an interference between raman scattering from continuum excitations and that from a discrete phonon , provided two raman - active excitations are coupled .
the bwf model line shape is given by : @xmath31 where @xmath32 and @xmath33 is the degree of coupling which describes the departure of the lineshape from a symmetric lorentzian function .
the @xmath34 is the intensity and @xmath35 and @xmath36 are the real and imaginary part of phonon self energy , respectively .
the spectra calculated using eq .
( @xmath37 ) are shown as solid lines in fig .
[ fig2 ] .
.parameters obtained by fitting of the b@xmath6 symmetry mode spectra of pure fesb@xmath5 with the bwf line shape model . [ cols="^,^,^,^",options="header " , ] from the bwf analysis of the experimental data for fe@xmath0(co , cr)@xmath38sb@xmath5 ( solid lines at the right panel of fig [ 4 ] ) , one can see a large increase of the @xmath39 with an increase of @xmath40 , indicating a decrease of electron - phonon interaction by co and cr alloying .
temperature dependance of energy and linewidth for fe@xmath0(co , cr)@xmath38sb@xmath5 alloys are shown in the left panel of fig .
experimental data are represented by symbols .
calculated spectra , obtained using eqs .
( [ 10 ] ) and ( [ 4 ] ) , are represented by dashed and solid lines .
best fit parameters are presented in table [ tab.2 ] .
@xmath41 decreases significantly in fe@xmath0(co , cr)@xmath38sb@xmath5 alloys compared to the pure fesb@xmath5 ( table [ tab.2 ] ) , although crystal disorder increases with increasing co and cr concentration ( for @xmath42 ) .
this is a consequence of drastic decrease of electron - phonon interaction contribution with increasing @xmath40 .
one can also notice that values of anharmonic constants decrease with increasing co and cr concentrations , what can be a consequence of change of electronic structure of material by alloying .
( left panel ) and fe@xmath43cr@xmath44sb@xmath5 ( right panel ) single crystals in the ( x@xmath27x@xmath27 ) configuration ( a@xmath7 symmetry modes ) measured at different temperatures.,scaledwidth=40.0% ] the polarized raman scattering spectra for pure fesb@xmath2 and fe@xmath43cr@xmath45sb@xmath2 single crystals in the ( x@xmath28x@xmath28 ) configuration ( a@xmath46 symmetry modes ) measured at different temperatures , are presented in fig . [ fig5 ] .
we have observed structure at about 155 @xmath47 which shows asymmetry towards higher wavenumbers .
however , this asymmetry can not be ascribed to the electron - phonon interaction , but is a consequence of the existence of two a@xmath46 symmetry modes , as we have already reported in our previously published paper.@xcite low temperature measurement confirmed our previous assignation .
the lorentzian lineshape profile has been used for extraction of mode energy and linewidth .
these modes have nearly the same energies what imposes the existence of the mode mixing , manifested by energy and intensity exchange .
the mixing is specially pronounced when the intensities of the modes are nearly the same.@xcite for fesb@xmath2 the mixing is strongest in the temperature range between 200 k and 250 k and for fe@xmath43cr@xmath44sb@xmath2 between 120 k and 180 k , see fig .
the temperature study of polarized raman scattering spectra of the fe@xmath0m@xmath38 sb@xmath5 ( m = cr , co ) single crystals has been performed . the linewidths and energies of the raman modes
were analyzed as a function of @xmath40 and temperature .
strong electron - phonon interaction , observed for the b@xmath6 symmetry mode of pure fesb@xmath5 , produces significant mode asymmetry .
the coupling constant reaches highest value at about 40 k and after that remains temperature independent .
additional broadening comes from the temperature induced anharmonicity . with increasing concentration of co and cr in fe@xmath0(co , cr)@xmath38sb@xmath5 alloys
the electron - phonon interaction is drastically reduced .
we have also observed mixing of the @xmath48 symmetry phonon modes in pure and cr doped sample .
we have pleasure to thank dr zorani dohevi - mitrovi for helpful discussion .
this work was supported by the serbian ministry of science and technological development under project no .
part of this work was carried out at the brookhaven national laboratory which is operated for the office of basic energy sciences , u.s .
department of energy by brookhaven science associates ( de - ac02 - 98ch10886 ) .
e. kroumova , m. i. aroyo , j. m. perez mato , a. kirov , c. capillas , s. ivantchev and h. wondratschek .
`` bilbao crystallographic server : useful databases and tools for phase transitions studies '' .
phase transitions * 76 * , nos . 1 - 2 , 155 - 170 ( 2003 ) . | we have measured polarized raman scattering spectra of the fe@xmath0co@xmath1sb@xmath2 and fe@xmath0cr@xmath1sb@xmath2 ( [email protected] ) single crystals in the temperature range between 15 k and 300 k. the highest energy @xmath4 symmetry mode shows significant line asymmetry due to phonon mode coupling width electronic background .
the coupling constant achieves the highest value at about 40 k and after that it remains temperature independent .
origin of additional mode broadening is pure anharmonic .
below 40 k the coupling is drastically reduced , in agreement with transport properties measurements .
alloying of fesb@xmath5 with co and cr produces the b@xmath6 mode narrowing , i.e. weakening of the electron - phonon interaction . in the case of a@xmath7 symmetry modes
we have found a significant mode mixing . |
exoplanet surveys have recently measured the frequency of exoplanets as a function of a host of parameters like stellar mass , metallicity , orbital separation , and planetary mass @xcite .
these parameters help to inform our understanding of how and where exoplanets form .
the observed frequency of gas giants at small ( @xmath112 au ) separations rises from @xmath23% for dm stars to @xmath214% for solar metallicity a - type stars ; this dearth of massive planets around dm stars is consistent with theoretical predictions @xcite of the core accretion model @xcite . at larger orbital separations ( 10s-100s of au ) , the frequency of gas giants around dm stars is @xmath116% @xcite .
recent results from kepler have shown both that the frequency of small mass planets at short orbital periods increases around low mass stars @xcite and there is a lack of planets larger than 2.5 r@xmath12 surrounding dm stars at short orbital periods .
high contrast imaging investigations have similarly begun to discover gas giant exoplanets located at large orbital separations from their stars ( fomalhaut b , @xcite , @xmath13 pictoris b , @xcite ; hr 8799 bcde , @xcite ; @xmath14 and .
b , @xcite ; 51 eri b , @xcite ; hd 100546 bc , @xcite , @xcite ) .
while most of these directly imaged gas giants surround early - type stars , detections have been reported around solar - analogs ( gj 504b , @xcite ) and dm stars ( e.g. roxs42b , @xcite ; gu psc , @xcite ) . yet
the formation mechanisms responsible for these systems is still under debate . a growing number of objects with wide orbits and modest mass ratios ( eg . hd 106906b , @xcite ; roxs42bb , @xcite ; 1rxsj1609 , @xcite ; 2 m j044144 , @xcite ) have led to suggestions that the planetary companion formed via a binary star - like process rather than the core accretion process ( @xcite , @xcite , @xcite ) .
although binary stars are common ( e.g. @xcite ) , our understanding of the frequency of exoplanets around binaries and higher order systems remains limited .
since the discovery of the first exoplanet surrounding a binary host ( kepler-16b , @xcite ) , less than a dozen similar systems have been discovered by kepler ( @xcite and references therein ) .
the analysis of publicly available kepler data led @xcite to conclude that the frequency of planets with r @xmath15 6r@xmath12 on periods of less than 300 days was similar to that of single star rates ; however , this conclusion is critically dependent on the assumed planetary inclination distribution .
while at least one bona fide planetary mass companion orbiting a binary has been imaged ( roxs 42bb * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) most dedicated direct imaging surveys for gas giant planets around binaries have not yielded any firm detections to - date @xcite .
recently , @xcite reported the detection of a planetary mass ( 11.2 @xmath16 m@xmath3 ) companion at a projected separation of 102 @xmath59 au from its host star vhs j125601.92 - 125723.9 ( hereafter vhs 1256 ) , described as a m7.5 object with an inferred mass from its bolometric luminosity of 73@xmath17 m@xmath3 , placing it near the hydrogen burning limit .
the primary was estimated to have an age of 150 - 300 myr from both kinematic membership in the local association and lithium abundance . at a distance of @xmath18 @xmath9 measured from trigonometric parallax @xcite ,
this made vhs 1256 the closest directly imaged planetary mass system to the earth .
@xcite reported a greater distance to vhs 1256 of @xmath19 @xmath9 based on spectrophotometry of the system . from the standpoint of substellar atmospheres and atmospheric evolution ,
vhs 1256 b is a particularly unique object .
its near - infrared properties resemble those of the hr 8799 planets and a select few other young ( t @xmath20 30 myr ) and very low mass ( m @xmath20 15 @xmath21 ) substellar objects , occupying roughly the same near - infrared color - magnitude space @xcite : a continuation of the l dwarf sequence to fainter magnitudes and cooler temperatures . indeed , as shown by atmosphere modeling , the near - infrared properties of objects like hr 8799 bcde and 2 m 1207 b reveal evidence for thicker clouds than field brown dwarfs of the same effective temperatures @xcite .
vhs 1256 b then offers a probe of clouds at ages intermediate between these benchmark objects and gyr - old field objects and thus some insights into the atmospheric evolution of low - mass substellar objects .
furthermore , non - equilbrium carbon chemistry can be probed by new thermal infrared photometry , in particular at @xmath1 ( e.g. * ? ? ?
much of this wavelength range is far less sensitive to carbon monoxide opacity at relevant temperatures that is a tracer of non - equilbrium carbon chemistry , while @xmath1 is far more ( uniquely ) sensitive ( e.g. see figure 7 in @xcite ) . ] .
new thermal infrared data for vhs 1256 b allows us to assess the evidence for non - equilibrium chemistry for the objects at / near the deuterium burning limit and at ages older than hr 8799 bcde . in this work ,
we present new adaptive optics imagery of vhs 1256 , providing the first detections of its wide - separation companion in major thermal ir broadband filters , @xmath0 and @xmath1 .
we use these mid - infrared photometric points and optical and near - infrared photometry from @xcite to perform the first atmospheric ( forward ) modeling of vhs 1256 b and the first assessment of how its thermal ir properties ( e.g. carbon chemistry ) compare to younger planet - mass objects with similar near - ir colors . additionally , we report our independent determination of the primary s binarity , also reported in @xcite , following our original work @xcite with additional analyses .
we will adopt the same nomenclature as stone et al .
@xcite , referring to the close partially resolved binary as vhs 1256 a and b , and the wide companion as vhs 1256 b. after discussing our observations in section [ sec : obs ] , we search for new companions around vhs 1256 and investigate the binarity of vhs 1256 in section [ sec : point ] .
next , we discuss improved photometry of vhs 1256 a , b , and b at @xmath0and @xmath1 in section [ sec : photometry ] . using the new l@xmath22 and m@xmath22 photometry
, we assess the atmospheric properties of vhs 1256 b in section [ sec : atmo ] .
finally , we discuss the implications of our study in section [ sec : discussion ] .
we observed vhs 1256 in @xmath0(@xmath23 = 3.77 @xmath24 m ) and @xmath1 ( @xmath23 = 4.68 @xmath24 m ) bands on 2015 june 6 using the infrared camera and spectrograph ( ircs ; @xcite ) on the 8.2 m subaru telescope , along with the subaru ao-188 adaptive optics ( ao ) system @xcite in laser guide star ( lgs ) mode .
all observations were made with ircs having a plate scale of 20.57 mas pix@xmath25 .
for the l band , we utilized a 0.1 second exposure time along with 100 co - adds to achieve 10 second exposures per dither frame .
we obtained a total of 13 dither sets , using a 5 point dither per set , yielding a total on - source integration time of 650 seconds . for our @xmath1-band imaging
, we utilized a 0.06 second exposure time along with 100 co - adds to achieve 6 second exposures per dither frame .
we observed vhs 1256 for a total of 3 dither sets , using a 5 point dither per set , yielding a total on - source integration time of 90 seconds .
all data were obtained in angular differential imaging ( adi ) mode ; a total field rotation of [email protected] and [email protected] was achieved in @xmath0and @xmath1 respectively .
we utilized two approaches to detect vhs 1256 b. first , we simply derotated each image to true north after sky subtraction .
we inspected the quality of each image set , and determined that the ao performance was subpar during 7 @xmath0frames ( 70 seconds of integration ) ; hence , we removed these data before median combining frames . using an identical process
, we reduced the @xmath1 data and removed 6 of the 15 frames ( 54 seconds of integration ) due to subpar ao performance .
second , we utilized advanced psf subtraction methods and a different approach with image combination to compensate for the shorter integration time and brighter sky background in @xmath1 and obtained slightly deeper limits in @xmath0 .
we use a - loci as in @xcite with a large optimization area ( 500 psf footprints ) and rotation gap ( 1.5 times the diffraction limit at l@xmath22 and m@xmath27 ) combined with a moving pixel mask to flatten the background without removing signal from vhs 1256 b. as vhs 1256 b is located about 8 from the primary , even a small [email protected] parallactic motion is sufficient to apply psf subtraction methods like loci @xcite ] .
we then applied a 4 @xmath28/d spatial filter and combined the derotated data to obtain a mean image using a 3-@xmath29 outlier rejection .
this allowed us to obtain a higher signal - to - noise detection of vhs 1256 b , which was barely detectable in @xmath1 at the 3-@xmath29 level using the first image processing method , obtaining about a 40% deeper background limit at @xmath0 .
we utilized observations of the @xmath30 standard fs 138 @xcite obtained on 2015 june 7 using ngs and observations of the @xmath31 standard hr 5384 @xcite obtained on 2015 june 8 without ao to transform our photometry to a standard system .
we did not obtain successful images of a dense stellar cluster that would be needed to construct a robust distortion correction for our @xmath0and @xmath1 imagery ; hence , we have included no such correction to our data .
our lack of a distortion correction introduces some uncertainty into vhs 1256 b s separation , but does not affect our other results .
our fully reduced @xmath0- and @xmath1-band imagery ( figure [ fig : l_m ] ) clearly reveals the presence of both the composite source of vhs 1256 a and b ( a+b ) and vhs 1256 b reported by @xcite .
we detect the companion at snr @xmath2 130 ( 100 ) in the @xmath0 data and at 4.5 ( 3 ) in the @xmath1 data using the a - loci ( classical ) reduction , where the latter detection is roughly comparable to the detection significance of hr 8799 bcd in @xcite .
we do not identify any additional point sources .
we determined the centroid positions of both objects , and found that vhs 1256 b is located 8@xmath413 @xmath5 0@xmath404 from vhs 1256 a+b at a position angle of @xmath32 in @xmath0and 8@xmath417 @xmath5 0@xmath404 at a position angle of @xmath32 in @xmath1 ( table [ tbl : properties ] ) .
this separation is consistent with the angular separation observed by @xcite of 8@xmath406 @xmath5 0@xmath403 at a position angle of @xmath33 .
note that we did not utilize a distortion correction . to compute the 5-@xmath29 point source detection limit in @xmath0
, we followed standard methods used in high - contrast imaging data ( e.g. * ? ? ? * ) .
we convolved the image with a gaussian profile having a fwhm set by the observed fwhm of vhs 1256 b and determination the robust standard deviation of convolved pixels at each angular separation .
the limiting 5-@xmath29 point source detection limits of our @xmath0 imagery based on this method is roughly 16.4 mag exterior to 25 ( r@xmath34 @xmath2 32 au for @xmath35 = 12.7 @xmath9 ) , degrading to vhs 1256 b s brightness at 05 . due to the thermal background , limited integration time , and the partial fov coverage
, the @xmath1 imagery do not provide stringent constraints on the presence of additional point sources compared to the @xmath0imagery .
it is apparent from visual inspection of figure [ fig : l_m ] that the vhs 1256 a+b is much more elongated in both @xmath0and @xmath1 -bands than vhs 1256 b. in fact , simple gaussian fits to the data reveal that the fwhm of the central star is significantly broader than the companion in both band - passes .
zooming in on the region around vhs 1256 a+b , figure [ fig : contour ] demonstrates that the central star is comprised of two marginally resolved sources . using a modified elliptical least squares minimization fitting routine written by nicky van foreest
, we compute the average ellipticity ( e = ( ( a@xmath36 - b@xmath36)/a@xmath36)@xmath37 ) ) of our sources .
the average ellipticity of the central star , @xmath38 and @xmath39 in the @xmath0and @xmath1 bands respectively , is significantly greater than the average ellipticity observed for the companion ( @xmath40 in @xmath0 ) .
these results all indicate that the central star is clearly comprised of the superposition of two separate sources . to help ascertain the probability that the observed multiplicity of vhs 1256 a and b arises from chance alignment of vhs 1256 with a background source , we utilize archival wise w1-band imagery , which has a similar band - pass as our @xmath0-band imagery .
the local density of w1 sources between 10.2 and 10.8 mag is @xmath2 34 @xmath41 hence , the chance alignment of two sources , within 0@xmath4106 is @xmath42 .
aperture photometry was performed on all components of the vhs 1256 system .
both the l@xmath27 photometry of vhs 1256 a+b ( 9.757 @xmath5 0.04 magnitude ) and vhs 1256 b ( 12.93 @xmath5 0.02 magnitude ) and the m@xmath27 photometry of vhs 1256 a+b ( 9.65 @xmath50.04 magnitude ) and vhs 1256 b ( 12.66 @xmath50.26 magnitude ) are broadly consistent with the literature @xcite wise w1 and w2 photometry ( table [ tbl : photometry ] ) .
we utilized a 30 pixels radius for aperture photometry .
in addition to the standard uncertainty terms ( poisson noise and standard deviation of background ) , we included read noise , shot noise , and the errors from the zero point terms .
we do not include error terms due to variations in the atmospheric transmission during our observations , or due to variations that arise between the use of ngs versus lgs observing modes .
note that our @xmath0and @xmath1 vhs 1256 b photometry improves on the respective wise w1 and w2 errors previously presented . the relative astrometry and flux of vhs 1256 a and b were constrained by fitting a linear combination of two psfs to our @xmath43 images
; we adopt vhs 1256 b as the psf .
we first align the dithered images of the central binary by maximizing their cross - correlation and then combine them by averaging . averaging the frames
is optimal in the case of gaussian errors ; in our case , it reduces the noise by nearly a factor of 2 compared with a median combination .
we use bicubic spline interpolation to translate the psf of vhs 1256 b. we then simultaneously adjust the positions and normalizations of two psfs to minimize @xmath44 where @xmath45 is the variance from photon noise and @xmath46 is the variance from read noise and the thermal background .
we estimate @xmath47 from the instrument gain and measure @xmath46 from empty regions of the image .
the variance is the sum of variance in the image and scaled variance from the noisy companion psf . in practice ,
most of the noise arises from scaling the companion psf , as the noise increases linearly with the scaling factors .
a good fit should have @xmath48 , where @xmath49 is the number of pixels in the region being fit . with a simple estimate of the background and photon noise
, we find that our best - fit @xmath50 is about four times the number of pixels ; we therefore increase our estimated uncertainties to achieve a reduced @xmath50 of unity .
after rescaling our noise to achieve a minimum @xmath51 , we compute our confidence regions by integrating the likelihood @xmath52 \label{eq : like}\ ] ] and computing the regions containing 68% of the likelihood ( for our 1@xmath29 confidence intervals ) .
figure [ fig : vhs ] shows the results of this fit : the observed @xmath43 intensity of vhs 1256 a and b ( left panel ) , the best - fit translated and scaled linear combination of companion psfs ( middle panel ) , and the residuals ( right panel ) , expressed in units of the peak intensity of the original data . the fit is good to a few percent near the psf cores , comparable to the expected uncertainties from the noise and from interpolations .
spatial correlations in the noise are visible in the residual image even far from the psf core ; it is these correlations that force us to scale our errors .
a full treatment of the problem would modify @xmath50 ( equation [ eq : chi2 ] ) to account for the data s non - diagonal covariance matrix .
table [ tbl : properties ] lists our fitted parameters : a separation of @xmath53 ( @xmath54 pixels ) , a position angle of @xmath55 , and a flux ratio of @xmath56 .
we thus obtain 1@xmath57 measurements of the relative photometry and separation .
the discovery work by @xcite suggested that vhs 1256 b , based on its spectral shape and color - magnitude diagram position , is a red l / t transition object similar to hr 8799 bcde ( e.g. * ? ? ?
* ; * ? ? ?
* ) . with sed information spanning the red optical to mid - infrared , we can more thoroughly compare vhs 1256 b s properties to those of other substellar objects and fit atmosphere models with a range of assumptions about clouds to reproduce vhs1256 b s spectrum .
many free floating l / t dwarfs with red optical to near - infrared photometry and spectra have been studied ( e.g. * ? ? ?
however , even considering young brown dwarfs as well as the field , vhs 1256 b occupies an extremely sparsely populated near - infrared color - magnitude diagram position shared by planet mass objects like hr 8799 bcde and 2 m 1207 b @xcite .
furthermore , these objects are young , typically less than @xmath2 30 @xmath58 old @xcite , while vhs 1256 is at least older than 200 @xmath58 ( * ? ? ?
* sect 6.2 of this work ) . while the similarity of vhs 1256 b s near - infrared colors to those of hr 8799 bcde suggest that their atmospheric properties likewise share some similarity ( e.g. perhaps thick clouds ) , the addition of thermal infrared data in this work allows us to further quantify this feature and assess whether or not vhs 1256 b shows clear evidence for non - equilibrium carbon chemistry .
compared to hr 8799 bcde , vhs 1256 b s available suite of photometry extends far bluer , into the optical .
thus , the companion provides a new test of atmospheric models used to reproduce at least some bona fide directly - imaged planets .
furthermore , as shown in @xcite , young l dwarfs tend to follow a reddened version of the field sequence . comparing vhs 1256 a+b s colors to those of the field sequence and young associations like ab dor
can then provide a very coarse assessment of the system s youth .
figure [ fig : cmd ] compares vhs 1256 a , b , and b s near - infrared and mid - infrared color magnitude diagrams ( cmds ) to the field sequence , planetary companions like hr 8799 bcde , and substellar objects in ab dor ( @xmath59 @xmath2 125 @xmath58 ) .
the primary components to vhs 1256 appear indistinguishable from the field sequence and bluer than ab dor members .
even if vhs 1256 b is at 17.1 @xmath9 and thus intrinsically brighter , it still appears somewhat under luminous " , along an extension of the l dwarf sequence to fainter magnitudes and presumably lower effective temperatures .
depending on the distance , vhs 1256 b s @xmath0/h-@xmath0 position is either consistent with hr 8799 b s or appears more comparable to the inner three planets .
in contrast , vhs 1256 b is significantly brighter at @xmath1 , relative to hr 8799 bcde .
vhs 1256 b s h-@xmath1 color is 0.61.1 magnitudes redder that hr 8799 bcd , suggesting that at @xmath1 it is roughly 1.72.7 times as bright . the @xcite and @xcite compilations of m
/ l / t objects ( figure [ fig : cmd ] ) are more sparsely populated in @xmath1 for late l and t dwarfs making comparisons with the field sequence difficult . as the faint @xmath1 brightnesses for l / t transition objects and the hr 8799 planets signaled evidence for non - equilibrium carbon chemistry @xcite , vhs 1256 b s brighter @xmath1 flux density may indicate that non - equilibrium carbon chemistry is not significant for every l / t transition object .
we perform atmospheric forward modeling to estimate vhs 1256 b s temperature , surface gravity , and radius , and to see if its photometry can be reproduced by atmospheric models in chemical equilibrium .
we follow the studies of @xcite and @xcite for hr 8799 s planets and @xcite for field brown dwarfs , comparing the data with atmosphere models from a. burrows covering a range of parameterized cloud prescriptions but all assuming chemical equilibrium .
we consider the e60 " models , which , have a modal particle size of 60 @xmath7 and clouds sharply truncated at depth , well below the planet photosphere .
we then consider a60 " and ae60 " cloud models , which have an identical modal particle size but simulate very thick `` and ' ' thick " cloud model presciptions ( see @xcite and @xcite for details ) .
from the model fitting , we identify those consistent with the data to within 3-@xmath29 using a simple @xmath60 threshold , setting the minimum photometry uncertainty to be 10% as in @xcite .
this method then yields a plausible range of temperature , surface gravity , radius , luminosity and mass from the ensemble of acceptably - fitting models .
we perform two sets of model fits , one where we allow the planet radius to freely vary and another when we pin it to values implied in the @xcite evolutionary models .
we assess how the system s uncertain distance affects our best - fit derived by performing fits , for the fixed - radius case , assuming either distances ( 12.7 and 17.1 @xmath9 ) .
figure [ fig : sedmod ] displays best - fit models using each cloud model assumption and tables [ tbl : atmosfit ] and [ tbl : atmosfitfixed ] summarize our model fitting results .
vhs 1256 b s sed can not be reproduced by the thin - cloud , e60 models : like hr 8799 s planets , such models badly under - predict the companion s brightness at the shortest wavelengths and over - predict it in the thermal infrared ( see * ? ? ?
the a60 , very - thick cloud models yield a drastically improved fit at temperatures between 900 @xmath61 and 1200 @xmath61 , where a model with @xmath62 = 1000 k , log(g ) = 4.5 is marginally consistent with the optical to mid - infrared sed at the 3-@xmath29 confidence limit
. models with slightly thinner clouds , ae60 , fare better , yielding a wider range of temperature and surface gravity phase space able to match the data .
the best - fit values systematically skew towards lower temperatures and surface gravities , where the best fit model has @xmath62 = 800 @xmath61 and log(g ) = 3.8 when the radius can freely vary and has a slightly higher gravity ( log(g ) = 4.1 ) when the radius is fixed . assuming a distance of 17.1 @xmath9 , the implied luminosity of vhs 1256 b ranges between log(l / l@xmath63 ) = @xmath64 to @xmath65 . adopting the 12.7 @xmath9 distance yields a luminosity consistent with that previously estimated by @xcite : log(l / l@xmath63 ) = @xmath66 to @xmath67 .
we will estimate the mass of vhs 1256 b by using these luminosities in section [ sec : masses ] . while successful at reproducing vhs 1256 b s sed , the derived model fit parameters could be revised by future modeling efforts .
for instance , the implied radii for best - fit models are generally larger than predicted for substellar object s with ages greater than 150 @xmath58 ( see * ? ? ?
* ) , and thus similarly the temperatures could be slightly larger than implied by our analysis .
subaru / ircs ao @xmath0and @xmath1 imagery has clearly revealed that the central source of the vhs 1256 system is comprised of two objects ( figure [ fig : l_m ] ) that have similar relative brightness ( @xmath0= 10.5 and 10.54 magnitude respectively ) . such binarity is observed in 22@xmath68% of very low mass stars @xcite .
@xcite assigned the central source a spectral classification of m7.5 , based on optical ( m7.0 ) and ir ( m8.0 ) spectral classifications .
we speculate that the minor differences in the optical versus ir spectral classifications derived by @xcite could be caused by minor differences in the spectral classifications of the binary components . at the observed distance to vhs 1256 ( @xmath18 @xmath9 ; @xcite ) , the @xmath69 projected separation between the binary components corresponds to a projected physical separation of @xmath21.3 au .
our results on the binarity of the central source are consistent with those independently and recently reported by @xcite .
@xcite suggested a system age of 150 - 300 myr , based on the lack of observed li in the system and kinematic age constraints from being a local association member .
however , with the discovery that the central source is a binary @xcite and our independent verification of vhs 1256 a and b in the @xmath0-band , we can reassess the age limits of the system . using the nominal distance ( 12.7 pc ) and the absolute magnitude ( m@xmath70 ; @xmath71 ) , the 300 myr upper limit age suggested by @xcite results in an inferred mass for vhs 1256 a or b of 47 m@xmath3 .
however such a mass would be too small to destroy li @xcite and produce the non - detection of this line ( figure [ fig : lithium ] ) .
rather , at this adopted distance the lower limit age of vhs 1256 must be @xmath15 400 myr to produce vhs 1256 a and b with our observed m@xmath70 and the lack of li in the system s spectra .
if one assumes the new distance of 17.1 pc proposed by @xcite and the corresponding absolute magnitude of the central components ( m@xmath70 ; @xmath72 ) , the lower age limit is @xmath15 200 myrs ( figure 5 ) .
this is broadly consistent with the lower age limit proposed by @xcite of @xmath73 .
note we used models from @xcite , while stone et al . used models from @xcite .
stone et al .
( @xcite ) suggested that vhs 1256 was consistent with being a member of the ab dor moving group , based on analysis of its uvw kinematics and a 66.85% membership probability predicted by the banyan ii software tool @xcite .
our own investigation suggests that it still has a 28% chance of being in the young field " ( age up to 1 @xmath74 ) .
additionally , vhs 1256 b is a clear outlier in uwv space ( @xmath2 8 @xmath5 1.7 km @xmath75 from the core of ab dor .
( j. gagne , pvt . comm . ) .
furthermore , membership in the 149@xmath76 myr ab dor moving group @xcite is inconsistent with the lower age limit of @xmath73 proposed by @xcite and marginally inconsistent with our lower limit of 200400 @xmath58 .
moreover , the near - to - mid infrared colors of vhs 1256 a appear indistinguishable from those in the field and potentially bluer than ab dor members ( figure [ fig : cmd ] .
thus , it is not clear that vhs 1256 is a member of the ab dor moving group , as suggested by @xcite .
as shown in figure [ fig : lithium ] , the minimum mass of each central component of vhs 1256 ( a and b ) is @xmath15 58 m@xmath3 for both of the distances discussed above .
this implies that the wide companion , vhs 1256 b , has a minimum mass ranging from 10.5 to 26.2 m@xmath3 , as shown in table [ tbl : absolute ] .
the large range is due to the uncertainty in the distance ( 12.7 or 17.1 pc ) and the range in bolometric luminosities from atmospheric fitting ( section [ sec : atmo ] ) .
though the lower estimate does dip below the deuterium burning limit , the companion is most likely in the brown dwarf regime .
we detected no other point source companions in our field of view ( fov ) , @xmath216@xmath45 x @xmath216@xmath45 in @xmath0and @xmath29@xmath43 x @xmath29@xmath43 in @xmath1 , down to our 5-@xmath29 sensitivity limits shown in figure [ fig : lims ] .
of 13.2 ( @xmath0 ; 12.5 mag at 17.1 pc ) . assuming a distance of 12.7 ( 17.1 ) @xmath9 , minimum system age of @xmath15 400 ( 200 ) myr , and no flux reversal at @xmath0 ( i.e. that more massive objects are fainter )
, we can therefore exclude the presence of additional companions more massive than vhs 1256 b beyond 6 ( 8) au . for most of the semi
major axis space we probe , comparisons with @xcite imply that companions down to 35 @xmath21 can be excluded if the system is 200400 @xmath58 old . because we have failed to identify other substellar companions orbiting the primary ,
this severely restricts the possibility that vhs 1256 b was scattered to its present orbit by dynamical interactions with another , unseen planet .
thus far , searches for close - in substellar companions to stars with imaged ( near ) planet - mass companions at 100500 au have failed to identify potential scatterers , suggesting that this class of objects formed in situ either from protostellar disk or molecular cloud fragmentation @xcite .
furthremore , the mass ratio ( q ) of vhs 1256 b ( [email protected] m@xmath3 ; median lower limit between 10.2 - 26.2 m@xmath3 ) to vhs 1256 a+b ( m@xmath77116 @xmath10 ) is @xmath20.16 .
this mass ratio is substantially larger than that observed for other imaged planetary systems such as hr 8799 ( q @xmath2 @xmath78 ; @xcite ) and roxs 42b ( q @xmath2 0.008 - 0.01 ; @xcite ) .
rather , it is more similar to that observed for low mass bds ( q @xmath2 0.01 - 0.9 ; eg .
see figure 4 , @xcite and citations there in ) .
we suggest this is indicative that the system formed via some form of fragmentation , i.e. a binary - star - like formation mechanism , rather than core accretion @xcite .
stone et al .
@xcite reached a similar conclusion of the binary - star - like formation .
although vhs 1256 b occupies a similar near - ir color - magnitude space to hr 8799 bcde @xcite , its significantly older age than the hr 8799 system enables one to probe a different time frame in planet / brown dwarf atmospheric evolution .
vhs 1256 b and hr 8799 bcd(e ? ) have different spectral energy distributions at the longest wavelengths probe ( @xmath1/4.7 @xmath7 ) . in the now - standard picture of understanding the atmospheres of the youngest and lowest - mass l / t objects , thick clouds and non - equilibrium carbon chemistry both
are due to the objects low surface gravity ( e.g. * ? ? ?
* ) . that vhs 1256 b thus far lacks evidence for non - equilibrium carbon chemistry may complicate this picture , suggesting some decoupling of gravity s two effects or that vhs 1256 b s gravity is high enough that non - equilibrium effects are less obvious than they are for , say , hr 8799 b. higher signal - to - noise detections in @xmath1 and photometry in the 34 @xmath7 range probing methane will allow us to better clarify vhs 1256 b s carbon chemistry .
multiple lines in @xmath79 band resolvable at medium resolution could better clarify the companion s surface gravity .
with other , similar objects detected at a range of ages , we can better map out the atmospheric evolution of objects of a given mass as well as the diversity of objects occupying the same reddened l / t transition region where vhs 1256 b and bona fide planets like hr 8799 bcde reside .
the vhs 1256 hierarchical triple system is poised to become an important contributor to our understanding of vlm systems .
it represents the third hierarchical triple system comprised solely of brown dwarf - mass components known @xcite .
given this projected separation ( 1.3 au ) and associated approximate orbital period ( @xmath24.7 years ) of the central binary in vhs 1256 , future ao spectroscopic monitoring of the system is poised to determine dynamical masses of all components of the triple system , which should help constrain evolutionary models ( see e.g. @xcite ) . since at least some brown dwarf binaries are believed to form via the disintegration of triple systems , and the third body in such systems are most likely also brown dwarf mass object @xcite , robustly determining the fundamental properties of the few known triple systems like vhs 1256 could help test the predictions of dynamical simulations of bd formation and evolution .
+ we thank sarah schmidt for thoughtful discussions about l / t dwarfs and multiplicity and jonathan gagne for helpful comments on vhs 1256 s possible membership in different moving groups .
we acknowledge support from nsf - ast 1009314 and nasa s origins of solar systems program under nnx13ak17 g .
this work was performed in part under contract with the jet propulsion laboratory ( jpl ) funded by nasa through the sagan fellowship program executed by the nasa exoplanet science institute .
this work is also based on data collected at subaru telescope , which is operated by the national astronomical observatory of japan .
the authors recognize and acknowledge the significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
9 allard , f. 2014 in exploring the formation and evolution of planetary systems , iau proc .
299 , 271 anthes rich , e. , wisniewski , j. p. , hashimoto , j. , et al .
2015 , aas / division for extreme solar systems abstracts , 3 , 117.10 armstrong , d.j . ,
osborn , h.p . ,
brown , d.j.a .
, faedi , f. , gomez maqueo chew , y. , martin , d.v . ,
pollacco , d. , & udry , s. 2014 , mnras , 444 , 1873 bailey , v. , meshket , t. , reiter , m. et al .
2014 , apjl , 780 , 4 baraffe , i. , chabrier , g. , barman , t. s. , et al . , 2003 , a&a , 402 , 701 bate , m. r. 2009 , , 392 , 590 bell , c.p.m . ,
mamajek , e.e . ,
& naylor , t. 2015 , mnras , 454 , 593 borucki , w.j . , koch , d.g .
, basri , g. et al .
2011 , apj , 736 , 19 bonnefoy , m. , zurlo , a. , baudino , j .- l . , et al . , 2016 ,
a&a , in press bouy , h. , martin , e.l . , brandner , w. , & bouvier , j. 2005 , aj , 129 , 511 bowler , b. p. , liu , m. c. , shkolnik , e. l. , & tamura , m. 2015 , , 216 , 7 brandt , t.d . ,
kuzuhara , m. , mcelwain , m.w .
2014 , apj , 786 , 1 bryan , m. , et al . , 2016 , , in press burrows , a. , hubbard , w. b. , lunine , j. i. , liebert , j. , 2001 , review of modern physics , 73 , 719 burrows , a. , sudarsky , d. , hubeny , i. , 2006 , , 640 , 1063 chabrier , g. , baraffe , i. , allard , f. , & hauschildt , p. 2000 , , 542 , 464 carson , j. , thalmann , c. , janson , m. , et al .
2013 , , 763 , l32 currie , t. , burrows , a. , itoh , y. , et al . , 2011 , , 729 , 128 currie , t. , debes , j. h. , rodigas , t. j. , et al . , 2012 , , 760 , l32 currie , t. , daemgen , s. , debes , j. , lafreniere , d. , itoh , y. , jayawardhana , r. , ratzka , t. , & correia , s. 2014 , apjl , 780 , 30 currie , t. , burrows , a. , girard , j. h. , et al .
2014 , , 795 , 133 currie , t. , cloutier , r. , grady , c. , et al .
, 2015 , , 814 , l27 currie , t. , lisse , c. , kuchner , m. , et al . , 2015 , , 807 , l7 duchene , g. & kraus , a. 2013 , ara&a , 51 , 269 doyle , l.r . ,
carter , j.a . ,
fabrycky , d.c .
2011 , science , 333 , 1602 dupuy , t. j. , liu , m. c. , bowler , b. p. , et al .
2010 , , 721 , 1725 dupuy , t. j. , & liu , m. c. 2012 , , 201 , 19 hayano , y. , takami , h. , guyon , o. et al .
2008 , in proc .
spie , adaptive optics systems , eds .
n. hubin , claire max , & p. wizinowich , vol 7015 , 10 hayano , y. , takami , h. , oya , s. et al .
2010 , in proc .
spie , vol .
7736 , 0 howard , a.w .
, marcy , g.w . , bryson , s.t .
2012 , apjs , 201 , 15 fabrycky , d. c. , & murray - clay , r. a. 2010 , , 710 , 1408 faherty , j. k. , riedel , a. r. , cruz , k. l. , et al .
2016 , arxiv:1605.07927 gagne , j. , lafreniere , d. , doyon , r. , malo , l. , & artigau , e. 2014 , apj , 783 , 121 gauza , b. , bejar , v.j.s .
, perez - garrido , a. et al .
2015 , apj , 804 , 96 galicher , r. , marois , c. , macintosh , b. , et al .
, 2011 , , 739 , l41 hinz , p. , rodigas , t.j . ,
kenworthy , m.a .
2010 , apj , 716 , 417 kalas , p.g . ,
graham , j.r .
, chiang , e. et al .
2008 , science , 322 , 1345 knapp , g. r. , leggett , s. k. , fan , x. , et al .
2004 , , 127 , 3553 kobayashi , h. , kawaguchi , n. , fujisawa , k. et al .
2000 , in proc .
spie , optical and ir telescope instrumentation and detectors , eds m. iye & a. f. moorwood , vol .
4008 , 1056 kuzuhara , m. , tamura , m. , kudo .
2013 , apj , 774 , 11 lafreniere , d. , marois , c. , duyon , r. , et al .
2007 , , 660 , 770 lafrenire , d. , jayawardhana , r. , & van kerkwijk , m. h. 2008 , , 689 , l153 lagrange , a .- m . , gratadour , d. , chauvin , g. , et al .
2009 , , 493 , l21 lagrange , a .- m . , bonnefoy , m. , chauvin , g. , apai , d. , ehrenreich , d. , boccaletti , a. , gratadour , d. , rouan , d. , mouillet , d. , lacour , s. , & kasper , m. 2010 , science , 329 , 57 laughlin , g. , bodenheimer , p. , & adams , f.c .
2004 , apjl , 612 , 73 leggett , s. k. , golimowski , d. a. , fan , x. , et al .
2002 , , 564 , 452 leggett , s. k. , burningham , b. , saumon , d. , et al . , 2010 , apj , 710 , 1627 liu , m.c . , magnier , e. a > , neacon , n. r. , et al . , 2013 , , 777 , l20 low ,
c. , & lynden - bell , d. 1976 , , 176 , 367 macintosh , b. , graham , j.r . , barman , t. et al .
2015 , science , 350 , 64 madhusudhan , n. , burrows , a. , currie , t. , 2011 , , 737 , 34 malo , l. , doyon , r. , lafreniere , d. , et al .
2013 , apj , 762 , 88 marley , m. , saumon , d. , cushing , m. , et al . , 2012 , , 754 , 135 marois , c. , zuckerman , b. , konopacky , q.m . ,
macintosh , b. , & barman , t. 2010 , nature , 468 , 1080 naud , m .- e . ,
artigau , e. , malo , l. et al .
2014 , apj , 787 , 5 quanz , s. , meyer , m. r. , kenworthy , m. , et al . , 2013 ,
apj , 766 , l1 pollack , j. b. , hubickyj , o. , bodenheimer , p. , et al .
1996 , icarus , 124 , 62 radigan , j. , jayawardhana , r. , lafreniere , d. et al . 2013 ,
apj , 778 , 36 raghavan , d. , mcalister , h.a . , henry , t.j .
et al . 2010 , apjs , 190 , 1 ratzka , t. , kohler , r. , & leinert , c. 2005 , a&a , 437 , 611 reipurth , b. & mikkola , s. 2015 , aj , 149 , 145 simon , m. , ghez , a.m. , leinert , c. et al .
1995 , apj , 443 , 625 skemer , a. , hinz , p.m. , esposito , s. et al .
2012 , apj , 753 , 14 spiegel , d. s. , & burrows , a. 2012 , , 745 , 174 stone , j. , et al . , 2016 , , 818 , 12 thalmann , c. , desidera , s. , bonavita , m. et al .
2014 , a&a , 572 , 91 todorov , k. , luhman , k. l. , & mcleod , k. k. 2010 , , 714 , l84 van der bliek , n.s . , manfroid , j. , & bouchet , p. 1996
, a&as , 119 , 547 winn , j.n . &
fabrycky , d.c .
2015 , araa , 53 , 409 a and b separation ( @xmath0 ) & 0@xmath41056 @xmath5 0@xmath4001 + a and b pa ( @xmath0 ) & @xmath80 + a and b est .
period ( @xmath0 ) & 4.7 yrs + a apparent magnitude ( @xmath0 ) & 10.50 @xmath50.01 mag + b apparent magnitude ( @xmath0 ) & 10.54 @xmath50.01 mag + a+b and b separation ( @xmath0 ) & 8.``13 @xmath5 0.''04 + a+b and b pa ( @xmath0 ) & 217.8 @xmath5 0.3 @xmath26 + a+b and b separation ( @xmath1 ) & 8.``17 @xmath5 0.''04 + a+b and b pa ( @xmath1 ) & 217.8 @xmath5 0.3 @xmath26 + llcccc l & 9.76 @xmath5 0.03 & 12.99 @xmath5 0.04 + m & 9.65 @xmath5 0.05 & 12.66 @xmath5 0.26 + w1 & 9.880 @xmath5 0.023 & 13.6 @xmath5 0.5 + w2 & 9.658 @xmath5 0.021 & 12.8 @xmath5 0.5 + [ tbl : photometry ] llcccc burrows / a60 & 1000 , 4.5 , 1.34 & 1000 , 4.5 , 1.34 + burrows / ae60 & 800 , 3.8 , 1.8 & 700 , 3.54 , 2.57 - 2.64 + & & 750 , 3.54.3 , 2.03 - 2.25 + & & 800 , 3.54.5 , 1.74 - 1.95 + & & 850 , 3.54.3 , 1.60 - 1.64 + & & 900 , 3.5 - 3.9 , 1.35 , 1.42 + burrows / e60 & 1000 , 4.0 , 0.98 &
+ [ tbl : atmosfit ] _ d = 17.1 @xmath9 _
+ burrows / a60 & 1000 , 4.25 , 1.45 & + burrows / ae60 & 850 , 3.8 , 1.55 & 850 , 3.53.9 , 1.481.75 + burrows / e60 & 700 , 4.0 , 1.45 & + + _ d = 12.7 @xmath9 _
+ burrows / a60 & 900 , 4.25 , 1.29 &
+ burrows / ae60 & 800 , 4.1 , 1.36 & 750 , 3.53.8 , 1.531.73 + & & 775 , 3.84.1 , 1.351.53 + & & 800 , 4.04.3 , 1.251.41 + & & 850 , 4.3 , 1.26 + burrows / e60 & 700 , 4.0 , 1.45 & + object & measurement & measurement + & at 12.7 @xmath5 1.0 pc & at 17.1 @xmath5 2.5 pc + + a absolute magnitude ( @xmath0 ) & 10.0 @xmath5 0.2 mag & 9.3 @xmath5 0.3 mag + b absolute magnitude ( @xmath0 ) & 10.0 @xmath5 0.2 mag & 9.4 @xmath5 0.3 mag + age lower limit & @xmath15 400 myrs & @xmath15 200 myrs + a and b mass bt - settll@xmath81 ( l ) & 58 m@xmath82 & 58 m@xmath82 + b absolute magnitude ( l ) & 12.5 @xmath5 0.2 mag & 11.8 @xmath5 0.3 mag + b mass bt - settl@xmath81 ( l ) & 26.2 to 74 m@xmath82 & 15.7 to 75 m@xmath82 + b bolometric luminosity log(l / l@xmath83 ) & -5.06 to -5.24 & -4.79 to -4.95 + b mass bt - settl@xmath81 bolometric & 12.6 - 20.6 m@xmath82 & 10.5 - 15.7 m@xmath82 + limiting absolute magnitude at 2@xmath48 ( l ) & 15.6 & 14.9 + limiting magnitude mass sb12@xmath84 models ( l ) & @xmath2 10 m@xmath82 & @xmath2 10 m@xmath82 + limiting magnitude mass bt - settl@xmath81 models ( l ) & 5 m@xmath82 & 7.8 m@xmath82 + | we present and analyze subaru / ircs @xmath0and @xmath1 images of the nearby m dwarf vhs j125601.92 - 125723.9 ( vhs 1256 ) , which was recently claimed to have a @xmath211 m@xmath3 companion ( vhs 1256 b ) at @xmath2102 au separation .
our adaptive optics images partially resolve the central star into a binary , whose components are nearly equal in brightness and separated by 0@xmath4106 @xmath5 0@xmath4001 .
vhs 1256 b occupies nearly the same near - infrared color - magnitude diagram position as hr 8799 bcde and has a comparable @xmath0 brightness .
however , it has a substantially redder @xmath6 - @xmath1 color , implying a relatively brighter @xmath1 flux density than for the hr 8799 planets and suggesting that non - equilibrium carbon chemistry may be less significant in vhs 1256 b. we successfully match the entire sed ( optical through thermal infrared ) for vhs 1256 b to atmospheric models assuming chemical equilibrium , models which failed to reproduce hr 8799 b at 5 @xmath7 .
our modeling favors slightly thick clouds in the companion s atmosphere , although perhaps not quite as thick as those favored recently for hr 8799 bcde .
combined with the non - detection of lithium in the primary , we estimate that the system is at least older than 200 myr and the masses of the stars comprising the central binary are at least 58 @xmath8 each .
moreover , we find some of the properties of vhs 1256 are inconsistent with the recent suggestion that it is a member of the ab dor moving group . given the possible ranges in distance ( 12.7 @xmath9 vs. 17.1 @xmath9 ) , the lower mass limit for vhs 1256 b ranges from 10.5 @xmath10 to 26.2 @xmath10 .
our detection limits rule out companions more massive than vhs 1256 b exterior to 68 au , placing significant limits on and providing some evidence against a second , more massive companion that may have scattered the wide - separation companion to its current location .
vhs 1256 is most likely a very low mass ( vlm ) hierarchical triple system , and could be the third such system in which all components reside in the brown dwarf mass regime . |
one of the well - known and fascinating mathematical problems is about the existence of integer solutions of polynomial equations of the type @xmath0 according to fermat s last theorem , which has only been proved very recently , the above equation does not have any integer solution for @xmath1 , @xmath2 , and @xmath3 . however , although it took hundreds of years for mathematicians to finally come up with such a proof , this proof may not be valid for similar equations such as @xmath4 many important mathematical problems , such as goldbach s conjecture or the distribution of zeroes of riemann zeta function , can be cast into equivalent problems of whether related polynomial equations with integer coefficients have integer solutions or not . because of this , david hilbert in his famous list of mathematical problems in 1900 included as the tenth problem @xcite the challenge : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ given any polynomial equation with any number of unknowns and with integer coefficients : to devise a universal process according to which it can be determined by a finite number of operations whether the equation has integer solutions . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ such equations are known as diophantine equations .
never was it anticipated that this tenth problem is ultimately equivalent ( as shown by davis , putnam , robinson and matiyasevich , see @xcite ) to the halting problem of turing machines of more than 30 years later . on this equivalence basis
it has been concluded that hilbert s tenth is not computable : there is no _ single _ universal process to determine the existence of integer solution or lack of it for arbitrarily given diophantine equations in as far as there is no _ single _ universal machine to determine the halting or not of arbitrarily given turing machine ( which starts with some arbitrarily given input ) .
thus , we would have to consider anew each different case of diophantine equation . in spite of this widely accepted result
, we have proposed a _ quantum algorithm _
@xcite for solving hilbert s tenth problem . in the next section we briefly summarise the algorithm .
we then next present some preliminary numerical results from the simulations of quantum processes for some very simple diophantine equations as a concrete proof of concept " .
it suffices to consider only nonnegative solutions of a diophantine equation .
let us consider the example @xmath5 with unknowns @xmath1 , @xmath2 , and @xmath3 .
starting from the observation that if we can construct the hamiltonian @xmath6 which has a spectrum bounded from below in fact , and if we can obtain the corresponding ground state ( of least energy ) then we can solve the tenth problem !
the ground state @xmath7 of the hamiltonian so constructed has the properties , for some @xmath8 , @xmath9 thus a projective measurement of the energy @xmath10 of the ground state @xmath7 will yield the answer for the decision problem : _ * _ the corresponding diophantine equation has at least one integer solution if and only if @xmath11 , and has not otherwise .
_ * _ ( if @xmath12 in our example , we know that @xmath13 from the fermat s last theorem . ) from the above observation , our ground - state oracle is thus clear : 1 . given a diophantine equation with @xmath14 unknowns @xmath1 s @xmath15 we need to simulate on some appropriate fock space the quantum hamiltonian @xmath16 2 .
measurement results of appropriate observables in the ground state will provide the answer for our decision problem .
one way , which is by no mean the only way , to obtain the ground state is guaranteed by the quantum adiabatic theorem @xcite which we will exploit in the next section . in the adiabatic approach @xcite ,
one starts with a hamiltonian @xmath17 whose ground state @xmath18 is readily achievable .
then one forms a slowly " varying hamiltonian which interpolates between @xmath17 and @xmath19 in the time interval @xmath20 $ ] @xmath21 we will adopt this approach with the proposed ( universal ) initial hamiltonian @xmath22 which admits as the ground state the coherence state @xmath23 where @xmath24 provided the conditions of the adiabatic theorem @xcite are observed , the initial ground state will evolve into our desirable ground state @xmath7 up to a phase : @xmath25 where @xmath26 is the time - ordering operator .
so our problem now is to solve the time - dependent schrdinger equation for @xmath27 $ ] @xmath28,\end{aligned}\ ] ] with the initial state being the coherence state ( [ cohstate ] ) .
we can analytically show in general the two crucial results below @xcite : * the ground state of @xmath29 is non - degenerate for @xmath30 . as the minimum energy gap between the ground state and the first excited state is non - zero
, it takes only a finite time @xmath31 for the adiabatic process , as asserted by the quantum adiabatic theorem , to generate a state which has a high probability of being the ground state of @xmath19 . * the probability of the state at time @xmath31 in some number state , @xmath32 , is greater than 1/2 _ iff _
@xmath33 is the ground state of @xmath19 .
that is , we only need to solve the equation for increasing @xmath31 until this majority condition is satisfied in order to _ identify _ the ground state .
the proofs of these important results will be available elsewhere .
* we solve the schrdinger equation numerically in some finitely truncated fock space large enough to approximate the initial coherence state to an arbitrarily given accuracy .
that is , a truncation @xmath34 is chosen such that @xmath35 has a norm less than one by some chosen @xmath36 , @xmath37 .
* at each time step @xmath38 , and up to @xmath39 , @xmath40 thus there are maximally only two creation operators , @xmath41 , and we can explore this fact to _ explore the infinite fock space _ by increasing the size of the truncated fock space by two at every time step .
* we employ the unitary solver , with conjugate gradient method , @xmath42 which approximates @xmath43 to second order in @xmath38 .
this solver , being unitary , preserves the norm of the state vector in the evolution of time even though the size of the underlying truncated fock space is allowed to increase with time if necessary . *
the time step @xmath38 at time @xmath44 is a function of time and is chosen in such a way that halving this @xmath45 only results in @xmath39 correction .
this is an equation resulted from the factoring of the number 15 into two prime factors . because of the nature of the problem we can fix the truncated size of our fock space to be @xmath47 in order to have the norm of the state vector less than unity by an amount @xmath48 at all times . in this and in all other simulations below
we choose @xmath49 . in fig .
[ fig1 ] we plot the magnitude square of maximum component ( in terms of the fock states ) of the state vector as a function of the total evolution time @xmath31 in some arbitrary unit . closed to below @xmath50 , the maximum probability components of @xmath51 are dominant by some states , denoted by ( blue ) star and ( green ) triangle symbols , none of which are the ground state .
in fact , they are , respectively , the _ first _ degenerate excited states @xmath52 and @xmath53 of @xmath19 . corresponding equation @xmath46 ; with intermediate evolution time @xmath31 . ] with the increase in @xmath31 in fig .
[ fig2 ] , one of the fock state , @xmath54 denoted by ( red ) box symbol , has probability greater than @xmath55 which is our criterion for being identified as the ground state .
we mark this regime as the quantum adiabatic regime when the ground state wins the battle for dominance . corresponding equation @xmath46 ; with @xmath31 entering the quantum adiabatic regime . ] from the ground state so identified we can infer that our diophantine equation has one solution in this domain .
we consider this extremely simple equation as an example which has no solution in the positive integers .
the simulation parameters are as in the previous , except that the truncated size of our fock space , starting with size 8 , is now allowed to vary with time in order to simulate the exploration of the whole infinite space . in fig .
[ fig3 ] we plot the probabilities of the dominant components as a function of @xmath31 .
below @xmath57 none of the two components is greater than one - half , and in fact the first excited state , @xmath58 denoted by ( blue ) triangle symbol , clearly dominates in this regime . eventually we enter the quantum adiabatic regime upon when the ( red ) box symbol rises over the one - half mark ; indeed it corresponds to the fock state @xmath59 which is the true ground state and which implies that our original diophantine equation has no integer solution at all .
corresponding equation @xmath60 . ]
we now consider a simple example which nevertheless has all the interesting ingredients typical for a general simulation of our quantum algorithm . here
we choose @xmath62 and the initial fock space has only up to @xmath63 , which does not include the true ground state of @xmath19 .
this is typical in our simulations since we in general would not be able to tell in advance whether our initial fock spaces do contain the true ground states or not .
generally , they do not .
even so , our strategy of allowing the expansion in the size of truncated fock space in time has enabled the true ground state to be found and identified . in this example , the state @xmath64 , _ which is not included in the initial truncated fock space _ ,
is eventually reached and identified as the ground state as shown as red boxes in fig .
blue triangles and green stars are corresponding to the first two excited states @xmath65 and @xmath66 , which are degenerate eigenstates of @xmath19 .
note that these competing pretenders somehow have unexpectedly probabilities greater than one - half ( around @xmath67 ) , contrary to our analytical result that only the ground state can have probability rising above one - half !
we think that this is only some artefact of finite - size time steps @xmath38 , and expect that it would go away once we employ a more sophisticated method for solving the schrdinger equation .
work is in progress to systematically extrapolate to zero - size time steps to confirm the removal of this type of finite - size effects . corresponding equation @xmath68 .
] figs .
[ fig5 ] and [ fig6 ] depict the expectation values of occupation number and of energy as functions of @xmath31 .
they are , respectively , approaching 20 and zero ( which signify the fact that our equation has an integer solution , namely 20 ) . the expectation value of occupation number , @xmath69 , as a function of @xmath31 .
] the expectation value of energy , @xmath70 , as a function of @xmath31 . ]
we here consider the issue of computability in principle , not that of computational complexity .
the simulation results support the idea of implementation and execution of our quantum algorithm with a _ physical _ process @xcite : * run the physical process ( corresponding to the diophantine equation in consideration ) for some time @xmath31 .
* repeat the process at this @xmath31 to obtain the statistics through measurements . *
if none of the measurement outcomes exhibits probability of more than 1/2 , ram up the @xmath31 and go back to the step above .
* eventually , at some sufficiently large @xmath31 , the measurement state with more - than - even probability can thus be identified as the ground state , terminating our physical implementation and execution . *
substituting the quantum numbers for the now identified ground state enables us to see if @xmath71 and thus whether the diophantine equation has a solution .
i would like to thank cristian calude , bryan dalton , peter hannaford , alan head , toby ord and andrew rawlinson for discussions and support .
i am also grateful to mathew bailes for the extensive use of swinburne supercluster facility to produce the numerical results reported herein , and to barbara mckinnon for discussions on the finer points of fortran90
. 99 y. matiyasevich , _
hilbert s tenth problem _ , mit press , 1993 .
tien d. kieu , computing the non - computable , " _ contemporary physics _ * 44 * , 51 - 71 , 2003 ; quantum algorithm for the hilbert s tenth problem , " e - print archive quant - ph/0110136v2 and references therein , 2001 ; gdel s incompleteness , chaitin s @xmath72 number and quantum physics , " e - print archive quant - ph/0111062v2 and references therein , 2001 ; a reformulation of the hilbert s tenth problem , " e - print archive quant - ph/0111063v1 and references therein , 2001 .
tien d. kieu , in preparation .
a. messiah , _ quantum mechanics _ , wiley & sons , 1958 .
e. farhi , j. goldstone , s. gutmann and m. sipser , quantum computation by adiabatic evolution , quant - ph/0001106 , 2000 . | we employ quantum mechanical principles in the computability exploration of the class of classically noncomputable hilbert s tenth problem which is equivalent to the turing halting problem in computer science .
the quantum adiabatic theorem enables us to establish a connection between the solution for this class of problems and the asymptotic behaviour of solutions of a particular type of time - dependent schrdinger equations .
we then present some preliminary numerical simulation results for the quantum adiabatic processes corresponding to various diophantine equations . |
dna and proteins are polyelectrolytes whose monomers are in a charged state that depends on the ph of the solution in which they are immersed
. the charges may fluctuate in space ( ` quenched ' ) and in time ( ` annealed ' ) . in this paper
we consider the charged polymer chain introduced in kantor and kardar @xcite .
the polymer chain is modelled by the path of a simple random walk on @xmath0 , @xmath1 .
each monomer in the polymer chain carries a _ random electric charge _ , drawn in an i.i.d . fashion from @xmath2 .
each self - intersection of the polymer chain contributes an energy that is equal to the product of the charges of the two monomers that meet ( i.e. , a negative energy when the charges have opposite sign and a positive energy when the charges have the same sign ) .
the polymer chain has a probability distribution on path space that is given by the _ gibbs measure _ associated with the energy .
our goal is to study the scaling properties of the polymer as its length tends to infinity .
very little is known mathematically about the _
quenched _ version of the model , where the charges are frozen .
the two main questions of interest are : * is the free energy self - averaging in the disorder ? * is there a phase transition from a ` collapsed phase ' to an ` extended phase ' at some critical value of the temperature ?
we expect that the answer to ( 1 ) is yes and the answer to ( 2 ) is no .
all we are able to show is the following ( see appendix [ appb ] ) : * if the average charge is non - zero , then the number of different sites visited by the polymer is proportional to its length . * in @xmath3
, if the average charge is sufficiently positive or negative and the temperature is sufficiently low , then the polymer behaves ballistically .
we expect that in any @xmath1 the scaling of the polymer is similar to that of the _ self - avoiding walk _ when the average charge is non - zero .
we further expect that the polymer is subdiffusive when the average charge is zero .
all these problems remains open . in the present paper we focus on the _ annealed _ version of the model , where the charges are averaged out
this version , which we study in @xmath3 only , is easier to deal with , yet turns out to exhibit a very rich scaling behavior .
the answer to ( 2 ) is yes for the annealed model .
we will obtain a detailed description of the phase transition curve separating a _
subballistic phase _ from a _
ballistic phase_. moreover , we show that the phase transition is first order , and show that the empirical speed and the empirical charge satisfy a law of large numbers , a central limit theorem , as well as a large deviation principle with a rate function that exhibits flat pieces .
the latter corresponds to an inhomogeneous strategy for the polymer to realise a large deviation .
we identify the scaling of the free energy in the limit of small average charge and small inverse temperature , which exhibits anomalous behaviour .
a key tool in our analysis is the ray - knight formula for the local times of the one - dimensional simple random walk .
this tool , which has been used extensively in the literature , is exploited in full throughout the paper in order to obtain the fine details of the phase diagram of the charged polymer .
the ray - knight formula is no longer available in @xmath4 . in berger , den hollander and poisat
@xcite it is shown that the phase diagram is qualitatively similar , but no detailed description of the scaling behaviour in the two phases is obtained .
the outline of the paper is as follows . in section [ ss : model ]
we define the model . in section [ ss : theoremsgeneral ]
we state six theorems with general properties and in section [ ss : theoremsasymptotics ] three theorems with asymptotic properties . in section [ ss : discussion ] we discuss these theorems .
proofs are given in sections [ freeenergy][s : asymptotics ] .
appendices [ appa][appc ] contain a few technical computations , while appendix [ appd ] states two modest results for the quenched version of the model . throughout the paper we use the notation @xmath5 and @xmath6 .
let @xmath7 be a simple random walk on @xmath0 , @xmath1 , i.e. , @xmath8 and @xmath9 , @xmath10 , with @xmath11 i.i.d .
random variables such that @xmath12 for @xmath13 with @xmath14 and zero otherwise ( @xmath15 denotes the lattice norm ) .
the path @xmath16 models the configuration of the polymer chain , i.e. , @xmath17 is the location of monomer @xmath18 .
we use the letters @xmath19 and @xmath20 for probability and expectation with respect to @xmath16 .
let @xmath21 be i.i.d .
random variables taking values in @xmath2 .
the sequence @xmath22 models the electric charges along the polymer chain , i.e. , @xmath23 is the charge of monomer @xmath18 ( see fig . [ fig - charpol ] ) .
we use the letters @xmath24 and @xmath25 for probability and expectation with respect to @xmath22 . throughout the paper
we assume that @xmath26 without loss of generality we may take ( see below ) @xmath27 to allow for biased charges , we use a tilting parameter @xmath28 and write @xmath29 for the i.i.d .
law of @xmath22 with marginal @xmath30 note that @xmath31 . in
what follows we may , without loss of generality , take @xmath32 .
* example : * the special case where the charges are @xmath33 with probability @xmath34 and @xmath35 with probability @xmath36 for some @xmath37 corresponds to @xmath38^{\otimes{\mathbb{n}}}$ ] and @xmath39 .
let @xmath40 denote the set of nearest - neighbor paths starting at @xmath41 .
given @xmath42 , we associate with each @xmath43 an energy given by the hamiltonian ( see fig .
[ fig - charpol ] ) @xmath44 let @xmath45 denote the inverse temperature . throughout the sequel the relevant space for the pair of parameters @xmath46 is the quadrant @xmath47 given @xmath48 , the _ annealed polymer measure of length _
@xmath49 is the gibbs measure @xmath50 defined as @xmath51 where @xmath52\ ] ] is the _ annealed partition function of length _ @xmath49 .
the measure @xmath50 is the joint probability distribution for the polymer chain and the charges at charge bias @xmath53 and inverse temperature @xmath45 when the polymer chain has length @xmath49 . in what follows , instead of we will work with the hamiltonian @xmath54 the sum under the square is the local time of @xmath16 at site @xmath55 weighted by the charges that are encountered in @xmath22 .
the change from to amounts to replacing @xmath45 by @xmath56 and adding a charge bias ( see section [ ss : loctimerepr ] for more details ) .
let @xmath57 be the probability matrix defined by @xmath58 which is the transition kernel of a critical galton - watson branching process with a geometric offspring distribution ( of parameter @xmath59 ) .
for @xmath48 , let @xmath60 be the function defined by
@xmath61 \quad \text { with } \quad \omega_\ell = \sum_{k=1}^\ell \omega_k , \qquad \ell \in { \mathbb{n}}_0.\ ] ] ( @xmath62 . ) for @xmath63 , define the @xmath64 matrices @xmath65 and @xmath66 by @xmath67 note that @xmath65 is symmetric while @xmath66 is not .
let @xmath68 and @xmath69 be the spectral radius of @xmath70 , respectively , @xmath66 in @xmath71 .
we will see in section [ subsec : eigenvalue ] that , for every @xmath48 , both @xmath72 and @xmath73 are continuous , strictly decreasing and log - convex on @xmath74 , tend to zero at infinity , and satisfy @xmath75 for all @xmath76 .
let @xmath77 which satisfy @xmath78 , with strict inequality as soon as @xmath79 .
we will also see that , for every @xmath48 , @xmath80 is analytic and strictly log - convex on @xmath81 , and has a finite strictly negative right - slope at @xmath41 ( see fig .
[ fig - spectral ] ) .
( 12,12)(0,-3 ) ( 0,0)(1,0)12 ( 0,-6)(0,1)14 ( 0,5)(3,0)(12,-3 ) ( 0,2)(1.5,0)(10,-3.5 ) ( -1,-.5)@xmath41 ( 12.5,-0.2)@xmath82 ( 5,.7)@xmath83 ( .6,-1.5)@xmath84 ( 13,-3)@xmath85 ( 11,-4.5)@xmath86 ( 0,5 ) ( 0,2 ) ( 5.3,0 ) ( 2.6,0 ) we begin with a _
spectral representation _ for the annealed free energy .
abbreviate @xmath87.\ ] ] [ thm : free.energy ] for all @xmath48 , the annealed free energy per monomer @xmath88 exists , takes values in @xmath89 $ ] , and satisfies the inequality @xmath90 moreover , the excess free energy @xmath91 is convex in @xmath46 and has the spectral representation @xmath92 the inequality in leads us to define two phases : @xmath93 we next show that these phases are separated by a single _ critical curve _ ( see fig .
[ fig - critcurve ] ) and that there are no further subphases .
( 12,12)(0,-2 ) ( 0,0)(12,0)12 ( 0,0)(0,8)8 ( 0,0)(5,0.5)(9,6.5 ) ( -1,-.5)@xmath41 ( 12.5,-0.2)@xmath53 ( -0.3,8.5)@xmath45 ( 8.5,2)@xmath94 ( 3,4)@xmath95 ( 8.5,7.3)@xmath96 [ thm : critcurve ] there exists a critical curve @xmath97 such that @xmath98 for every @xmath99 , @xmath96 is the unique solution of the equation @xmath100 .
moreover , @xmath97 is continuous , strictly increasing and convex on @xmath74 , analytic on @xmath81 , and satisfies @xmath101 .
in addition , @xmath102 is analytic on @xmath94 .
let @xmath103 the set @xmath104 will be referred to as the _ ballistic phase _ , the set @xmath105 as the _ subballistic phase _ ,
for reasons we explain next .
namely , we proceed by stating a _ law of large numbers _ for the empirical speed @xmath106 and the empirical charge @xmath107 , respectively , with @xmath108 in the statement below the condition @xmath109 is put in to choose a direction for the endpoint of the polymer chain .
[ thm : speed ] for every @xmath48 there exists a @xmath110 $ ] such that @xmath111 where @xmath112 for every @xmath113 , @xmath114_{\mu=\mu(\delta,\beta ) } = \left[-\frac{\partial}{\partial\mu}\lambda_{\delta,\beta}(\mu)\right]_{\mu=\mu(\delta,\beta)}.\ ] ] ( take the right - derivative when @xmath115 ; see fig .
[ fig - spectral ] . )
moreover , @xmath116 is analytic on @xmath117 .
[ thm : charge ] for every @xmath48 , there exists a @xmath118 such that @xmath119 where @xmath120 for every @xmath113 , @xmath121_{\mu=\mu(\delta,\beta ) } = \frac{\partial}{\partial\delta}\,\mu\big(\delta,\beta\big).\ ] ] moreover , @xmath122 is analytic on @xmath117 .
since @xmath123 lies in the ballistic phase @xmath104 ( recall ) , theorems [ thm : speed][thm : charge ] imply that @xmath124 and @xmath125 are _ discontinuous at criticality_. this means that the _ phase transition is first order_. see fig .
[ fig - speedcharge ] below for numerical plots of @xmath116 and @xmath126 .
in fact , _ large deviation principles _ holds for the laws of the empirical speed and the empirical charge .
let @xmath127 and note that @xmath128 .
[ thm : jointldp ] for every @xmath48 : + ( 1 ) the sequence @xmath129 conditionally on @xmath130 satisfies the large deviation principle on @xmath74 with rate function @xmath131 given by @xmath132 , \qquad \theta \in [ 0,\infty).\ ] ] ( 2 ) the sequence @xmath133 satisfies the large deviation principle on @xmath74 with rate function @xmath134 given by @xmath135 , \qquad \theta ' \in [ 0,\infty).\ ] ] ( the large deviation principle on @xmath136 is obtained from that on @xmath81 after reflection of the charge distribution . ) ( 12,8)(10,-2 ) ( 0,0)(16,0)16 ( 0,0)(0,8)8 ( 0,6)(2,3)(3,2 ) ( 3,2)(5,0)(8,0 ) ( 8,0)(11,0)(14,4 ) ( 3,2)(3,1)(3,0 ) ( -1,-.5)@xmath41 ( 16.5,-0.2)@xmath137 ( 8,0 ) ( 3,2 ) ( -1,9)@xmath138 ( 6.7,-1.3)@xmath139 ( 1.4,-1.3)@xmath140 ( 20,0)(16,0)16 ( 20,0)(0,8)8 ( 20,0)(24,2)(28,4 ) ( 28,4)(30,5)(32,7 ) ( 28,0)(28,2)(28,4 ) ( 19,-.5)@xmath41 ( 36.5,-0.2)@xmath137 ( 20,0 ) ( 28,4 ) ( 19,9)@xmath138 ( 26.4,-1.3)@xmath140 ( 12,8)(10,-2 ) ( 0,0)(16,0)16 ( 0,0)(0,8)8 ( 0,6)(2,3)(3,2 ) ( 3,2)(5,0)(8,0 ) ( 8,0)(11,0)(14,4 ) ( 3,2)(3,1)(3,0 ) ( -1,-.5)@xmath41 ( 16.5,-0.2)@xmath141 ( 8,0 ) ( 3,2 ) ( -1,9)@xmath142 ( 6.7,-1.3)@xmath143 ( 1.8,-1.3)@xmath144 ( 20,0)(16,0)16 ( 20,0)(0,8)8 ( 20,0)(24,2)(28,4 ) ( 28,4)(30,5)(32,7 ) ( 28,0)(28,2)(28,4 ) ( 19,-.5)@xmath41 ( 36.5,-0.2)@xmath141 ( 20,0 ) ( 28,4 ) ( 19,9)@xmath142 ( 26.8,-1.3)@xmath144 the two rate functions are depicted in figs .
[ fig - rfspeed][fig - rfcharge ] .
they are strictly convex , except for linear pieces on @xmath145 $ ] and @xmath146 $ ] with @xmath147_{\mu={\widetilde}\mu(\delta,\beta ) } , \qquad { \widetilde}{\rho}(\beta ) = \rho(\delta_c(\beta),\beta),\ ] ] where @xmath148 is the inverse of @xmath97 ( recall fig .
[ fig - critcurve ] ) .
note that , whereas @xmath139 in and @xmath143 in jump from a strictly positive value to zero when @xmath46 moves from @xmath104 to @xmath105 inside @xmath149 , @xmath140 and @xmath150 in are strictly positive throughout @xmath149 .
the large deviation principles in turn yield _ central limit theorems _ : [ thm : cltspeedcharge ] for every @xmath151 , @xmath152 converge in distribution to the standard normal law , with @xmath153 given by @xmath154^{-1}_{\theta = v(\delta,\beta ) } = \left[\frac{\partial^2}{\partial\gamma^2}\,\mu({\delta},{\beta},\gamma)\right]_{\gamma = 0}\\ & = v(\delta,\beta)^3 \left[\frac{\partial^2}{\partial\mu^2 } \log\lambda_{\delta,\beta}(\mu)\right]_{\mu=\mu(\delta,\beta)},\\ \sigma_\rho(\delta,\beta)^2 & = \left[\frac{\partial^2}{\partial\theta'^2}\ , i^\rho_{\delta,\beta}(\theta')\right]^{-1}_{\theta'=\rho(\delta,\beta ) } = \frac{\partial^2}{\partial\delta^2}\,\mu({\delta},{\beta } ) = \frac{\partial}{\partial\delta}\,\rho({\delta},{\beta } ) .
\end{aligned}\ ] ] the proof of theorem [ thm : cltspeedcharge ] , that we give in section [ cltproof ] , is inspired by knig @xcite .
the expression in the second line of can be written out in terms of @xmath155 , @xmath156 and second order derivatives with respect to @xmath53 and @xmath82 of @xmath85 at @xmath157 , but the resulting expression is not particularly illuminating . in theorem [ thm : asymp.cc](1 ) and theorem [ thm : fscalbeta ]
below we need to make an additional assumption on the charge distribution , namely , we require that one of the following properties holds : @xmath158 for @xmath159 and @xmath160 , let @xmath161 be the _ sturm - liouville operator _ defined by @xmath162
this is a two - parameter version of a one - parameter family of operators considered in van der hofstad and den hollander @xcite .
let @xmath163\ , { \text{\rm d}}x < \infty\right\}.\ ] ] the largest eigenvalue problem @xmath164 has a unique solution @xmath165 with the following properties : for every @xmath160 , @xmath166 ( see coddington and levinson @xcite for general background on sturm - liouville theory . )
( 8,8)(0,-3 ) ( -5,0)(10,0)10 ( 0,-5)(0,10)10 ( -4,-4)(2,-1)(4,4 ) ( 5.5,-.2)@xmath167 ( -1.3,5.6)@xmath168 ( 1.5,-1.5)@xmath169 ( 1.5,0 ) let @xmath170 denote the unique solution of the equation @xmath171 ( see fig .
[ fig - chiab ] ) .
the critical curve has the following scaling behaviour for small and for large charge bias .
[ thm : asymp.cc ] ( 1 ) as @xmath172 , @xmath173 ( 2 ) as @xmath174 , @xmath175 with @xmath176 ( with the convention @xmath177 ) .
either @xmath178 ( ` lattice case ' ) or @xmath179 ( ` non - lattice case ' ) .
if @xmath179 and @xmath180 has a bounded density ( with respect to the lebesgue measure ) , then @xmath181 the proof of , given in section [ ss : wil ] , follows van der hofstad and den hollander @xcite , but we have to address additional difficulties , due to our more complicated hamiltonian .
the scaling behaviour of the excess free energy near the critical curve shows that the phase transition is first order .
[ thm : orderphtr ] for every @xmath182 , @xmath183 , \quad \mbox { as } \beta \uparrow \beta_c(\delta),\ ] ] where @xmath184 is given by @xmath185_{\beta=\beta_c(\delta),\mu=0}.\ ] ] we close by identifying the scaling behaviour of the free energy for small charge bias and small inverse temperature .
the proof also follows der hofstad and den hollander @xcite .
[ thm : fscalbeta ] ( 1 ) for every @xmath186 , @xmath187 where @xmath188 , and @xmath189 are given by @xmath190_{a = a^*(\rho_\delta ) , \,b=\rho_\delta } , \qquad c_\delta = -\frac{{\text{\rm d}}}{{\text{\rm d}}\delta}\,a^*(\rho_\delta).\ ] ] the third statement in holds under the assumption that @xmath191 with @xmath192 defined in and @xmath193 . without this assumption
only the weaker result @xmath194 holds .
+ ( 2 ) for every @xmath195 , @xmath196 ( the notation @xmath197 means that the ratio @xmath198 stays bounded from above and below by finite and positive constants . ) we discuss the theorems stated in sections [ ss : theoremsgeneral][ss : theoremsasymptotics ] and place them in their proper context .
the quenched charged polymer model with @xmath199^{\otimes{\mathbb{n}}}$ ] interpolates between the simple random walk ( @xmath200 ) , the self - avoiding walk ( @xmath201 ) and the weakly self - avoiding walk ( @xmath202 , @xmath203 ) , for which an abundant literature is available ( see den hollander ( * ? ? ?
* chapter 2 ) for references ) .
the latter corresponds to the situation where all the charges are @xmath33 , in which case the hamiltonian in equals @xmath204 with @xmath205 the local time of @xmath16 at site @xmath55 up to time @xmath49 .
theorem [ thm : free.energy ] shows that the annealed excess free energy exists and has a spectral representation .
the latter generalizes the spectral representation derived in greven and den hollander @xcite for weakly self - avoiding walk ( see den hollander ( * ? ? ?
* chapter ix ) ) .
theorem [ thm : critcurve ] shows that there is a phase transition at a non - trivial critical curve and that there are no further subphases .
@xmath206 * 2 . *
theorems [ thm : speed][thm : charge ] and [ thm : orderphtr ] show that the annealed charged polymer exhibits a phase transition of first order .
the speed @xmath139 of the polymer chain is strictly positive in the ballistic phase and zero in the subballistic phase ( which explains the names associated with these two phases ) . in the ballistic phase
the speed is given by the spectral formula in .
the latter generalizes the spectral formula derived in greven and den hollander @xcite for the speed @xmath207 of the weakly self - avoiding walk .
the charge @xmath143 of the polymer chain is strictly positive in the ballistic phase and zero in the subballistic phase . in the ballistic phase the charge
is given by the spectral formula in .
[ fig - speedcharge ] shows a numerical plot of @xmath208 and @xmath209 when @xmath210 is standard normal .
interestingly , the speed is not monotone on @xmath211 $ ] .
this is in contrast with the monotonicity that was found ( but was not proven ) in @xcite for the weakly self - avoiding walk ( for which @xmath212 ) .
equally interesting , the charge is monotone on @xmath211 $ ] .
a rough heuristics behind the shape of @xmath139 and @xmath143 is the following . approximating the distributions of @xmath213 and @xmath214 by standard normal laws
, we get @xmath215\right)\\ & \approx \sup _ { { v \in ( 0,\infty ) } \atop { \rho \in { \mathbb{r } } } } \left[-\beta v \left(\frac{\rho}{v}\right)^2 + \delta \rho - \tfrac12(v^2+\rho^2)\right ] . \end{aligned}\ ] ] here , the supremum runs over the possible values of the empirical speed and the empirical charge , the first term arises from the hamiltonian in , the second term comes from the tilting of the charges in , together with the approximation @xmath216 , while the third term embodies the normal approximation . for fixed @xmath217 the supremum over @xmath218
is taken at @xmath219 .
substitution of this relation shows that the supremum over @xmath217 is taken at the solution of the equation @xmath220 .
hence @xmath221 these approximations are compatible with the numerical plots in fig .
[ fig - speedcharge ] .
* theorem [ thm : jointldp ] identifies the rate functions in the large deviation principles for the speed and the charge . both rate functions exhibit flat pieces in both phases , as indicated in figs .
[ fig - rfspeed][fig - rfcharge ] .
these flat pieces correspond to an inhomogeneous strategy for the polymer to realise a large deviation .
for instance , in the flat piece on the left of fig .
[ fig - rfspeed ] , if the speed is @xmath222 , then the charge makes a large deviation on a stretch of the polymer of length @xmath223 times the total length , so as to allow it to move at speed @xmath140 along that stretch at zero cost , and then makes a large deviation on the remaining stretch , so as to allow it to be subballistic along that remaining stretch at zero cost . for the weakly self - avoiding walk the presence of a flat piece in the rate function for the speed
was noted in den hollander ( * ? ? ?
* chapter 8) .
it is possible to extend theorem [ thm : jointldp ] to a joint ldp , but we refrain from doing so . * 4 . *
theorem [ thm : cltspeedcharge ] provides the central limit theorem for the speed and the charge in the interior of the ballistic regime .
the variance is the inverse of the curvature of the rate function at its unique zero , as is to be expected .
numerical plots are given in fig .
[ fig - speedchargevar ] .
it is hard to obtain accurate simulations for @xmath45 small , but the plots appear to be compatible with the assumption made in . for weakly self - avoiding walk
it was shown in van der hofstad , den hollander and knig @xcite that @xmath224 is discontinuous at @xmath225 , namely , @xmath226 .
[ fig - speedcharge ] suggests that this behaviour persists for @xmath227 .
the heuristics is that the variance of the endpoint of the polymer gets squeezed because the polymer moves ballistically .
apparently this squeezing does not vanish as the speeds tends to zero .
@xmath206 we do not deduce the central limit theorem from the large deviation principle , but rather exploit finer properties of the spectral representation for the excess free energy .
we have no result about the fluctuations at criticality .
we expect these fluctuations to be of order @xmath228 in the upward direction and of order @xmath229 in the downward direction .
* theorem [ thm : asymp.cc ] identifies the scaling behavior of the critical curve for small and for large charge bias .
part ( 1 ) shows that the scaling is anomalous for small charge bias , and implies that the critical curve is not analytic at the origin .
part ( 2 ) shows that the scaling is also delicate for large charge bias .
heuristically , it is easier to build small absolute values of @xmath230 for small values of @xmath231 when the charge distribution is non - lattice rather than lattice .
since the local times are of order one in the ballistic phase , we expect that the ballistic phase for the lattice case is contained in the ballistic phase for the non - lattice case ( because smaller values of @xmath45 are needed to compensate for the larger absolute values of @xmath232 ) .
* theorem [ thm : fscalbeta ] deals with weak interaction limits .
part ( 1 ) shows that near the horizontal axis in fig .
[ fig - critcurve ] the free energy , the speed and the charge exhibit an anomalous scaling .
this is a generalization of the scaling found in van der hofstad and den hollander @xcite for weakly self - avoiding walk .
part ( 2 ) shows that near the origin of fig .
[ fig - critcurve ] the free energy scales like the distance to the critical curve , provided the latter is approached properly .
the constants @xmath233 are expected to represent the free energy , speed and charge of a brownian version of the charged polymer with hamiltonian @xmath234 ) = \int_{\mathbb{r}}l_t^{\mathrm{\widetilde{w}}}(x)^2\,{\text{\rm d}}x , \quad l_t^{\mathrm{\widetilde{w}}}(x ) = \int_0^t { \text{\rm d}}{\mathrm{\widetilde{w}}}_s\,{\delta}(w_s - x),\ ] ] where @xmath235 $ ] is the path of the polymer , @xmath236 is the charge of the interval @xmath237 , @xmath238 $ ] is an independent brownian motion with drift @xmath239 , and the polymer measure has @xmath240 with the wiener measure as reference measure . the version without charges is known as the edwards model ( see van der hofstad , den hollander and knig @xcite ) .
the limit below is expected to represent the standard deviation in the central limit theorem as @xmath241 for the charge in the continuum model defined via .
* theorem [ thm : critcurve ] corrects a mistake in den hollander ( * ? ? ?
* chapter 8) , where it was argued that @xmath242 ( i.e. , @xmath105 covers the full quadrant , or @xmath243 ) .
the mistake can be traced back to a failure of convexity of the function @xmath244 . using the technique outlined in den hollander (
* chapter 8) , it can be shown that for every @xmath1 and every @xmath245 , @xmath246 with @xmath247 and with @xmath248 a constant that is explicitly computable .
the idea behind is the following .
for @xmath245 the empirical charge makes a large deviation under the disorder measure @xmath29 so that it becomes zero .
the price for this large deviation is @xmath249 where @xmath250 denotes the specific relative entropy of @xmath251 with respect to @xmath29 . since the latter equals @xmath252 ( recall ) , this accounts for the leading term in the free energy .
conditional on the empirical charge being zero , the attraction between charged monomers with the same sign wins from the repulsion between charged monomers with opposite sign , making the polymer chain contract to a _ subdiffusive _ scale @xmath253 .
this accounts for the correction term in the free energy .
it is shown in @xcite that under the annealed polymer measure , @xmath254 where @xmath255 denotes convergence in distribution and @xmath256 is a brownian motion on @xmath257 conditioned not to leave a ball with a certain radius and a certain randomly shifted center .
* previous results on the charged polymer model include limit theorems for the hamiltonian in .
chen @xcite proves an annealed central limit theorem and an annealed law of the iterated logarithm , and identifies the annealed moderate deviations ( see also chen and khoshnevisan @xcite ) .
asselah @xcite , @xcite derives upper and lower bounds for annealed large deviations .
hu and khoshnevisan @xcite give a law of the iterated logarithm and a strong approximation theorem : on an enlarged probability space the properly normalised hamiltonian converges almost - surely to a reparametrised brownian motion .
guillotin - plantard and dos santos @xcite prove a quenched central limit theorem in dimensions @xmath258 .
hu , khoshnevisan and wouts @xcite consider the quenched weak interaction regime ( where the hamiltonian is multiplied by @xmath259 rather than @xmath260 ) and prove a phase transition from brownian scaling to four - point localization : for small @xmath45 the polymer behaves like a simple random walk , while for large @xmath261 a large fraction of the monomers are located on four sites .
@xmath206 * 9 . *
the large deviation bounds derived by asselah @xcite , @xcite can be completed as follows . under the annealed polymer measure , the sequence @xmath262 ( recall )
satisfies the ( weak ) large deviation principle on @xmath2 with ( weak ) rate function @xmath263 given by ( see fig . [ fig - ratefunctionham ] ) @xmath264 , \qquad u \in { \mathbb{r}}.\ ] ] here we use that @xmath265 for @xmath266 and @xmath267 , to restrict the supremum to @xmath202 .
( indeed , the strategy where the charges are bounded from below by a positive constant and the walk zigzags between two consecutive sites has an entropic cost that is linear in the length of the polymer , whereas the positive energetic contribution is quadratic . ) since @xmath268 by theorem [ thm : free.energy ] , provides us with an explicit variational formula similar to .
* here are some open problems for the _ quenched _ version of the model ( see appendix [ appb ] ) : * does the quenched free energy exist for @xmath29-a.e .
@xmath22 , and is it constant ? how does it depend on @xmath53 and @xmath45 ?
trivially , it is convex in @xmath45 for all @xmath53 , but what more can be said ? * is the quenched charged polymer ballistic for all @xmath182 ?
how does the speed depend on @xmath45 and @xmath53 ?
* in the quenched model with @xmath269 , is the polymer chain subdiffusive ( like in the annealed model ; see item 3 above ) ?
the fluctuations of the charges are expected to push the polymer farther apart than in the annealed model .
is there a scaling limit for @xmath24-a.e .
@xmath22 , or does the polymer chain fluctuate so much that there is a scaling limit only along @xmath22-dependent subsequences ( `` sample dependence '' ) ? * 11 . *
still looking at a quenched model , derrida , griffiths and higgs @xcite and derrida and higgs @xcite consider the case where the steps of the random walk are drawn from @xmath270 rather than @xmath271 , which makes the model a bit more tractable , both theoretically and numerically . in @xcite
the charge disorder is _ binary _ , and numerical evidence is found for the free energy to be self - averaging and to exhibit a _ freezing transition _ at a critical threshold @xmath272 , i.e. , the quenched charged polymer is ballistic when @xmath273 and subballistic when @xmath274 . in the latter phase numerical simulation
shows that the end - to - end distance scales like @xmath275 , with @xmath276 an exponent that depends on @xmath45 . in this phase ,
long and rare stretches of the polymer that are globally neutral find it energetically favorable to collapse onto single sites .
numerical simulation indicates that @xmath277 . in @xcite
the charge disorder is _ standard normal _ and the total charge @xmath278 is _ conditioned _ to grow like @xmath279 , @xmath280 $ ] .
it is found numerically that the end - to - end distance scales like @xmath275 , with @xmath281 an exponent that depends on @xmath282 and grows roughly linearly from @xmath283 to @xmath284 , with @xmath285 .
the latter is the exponent for the quenched charged polymer when the charges are typical .
* it would be interesting to deal with charges whose interaction extends beyond the ` on - site ' interaction in ( [ eq : def.ham ] ) , like a coulomb potential ( polynomial decay ) or a yukawa potential ( exponential decay ) .
a yukawa potential arises from a coulomb potential via screening of the charges when the polymer chain is immersed in an ionic fluid . *
biskup and knig @xcite , ioffe and velenik @xcite , kosygina and mountford @xcite deal with annealed versions of various models of simple random walk in a random potential . in all these models
the interaction is either _ attractive _ or _ repulsive _ , meaning that the annealed partition function is the expectation of the exponential of a functional of the local times of simple random walk that is either _ subadditive _ or _ superadditive_. as we will see in section [ freeenergy ] , our annealed charged polymer model is neither attractive nor repulsive .
however , our spectral representation is flexible so as to include such models .
our goal in this section is to prove theorem [ thm : free.energy ] , i.e. , the existence of the annealed free energy and its characterization in terms of an eigenvalue problem . in section [ sec : markov_edge_local_times ] we show that the edge - crossing numbers of the simple random walk have a markovian structure .
in section [ ss : loctimerepr ] we rewrite the annealed partition function as the expectation of a functional of the local times of the simple random walk , which are a two - block functional of the edge - crossing numbers . in section [ ss : grcanrepr ]
we introduce the generating function of the annealed excess partition function , and show that this can be expressed in terms of the matrices defined in . the annealed excess free energy is the radius of convergence of this generating function . in section [ subsec : eigenvalue ]
we analyze the spectral radii of the matrices . in section [ ss : spec ]
we identify the annealed excess free energy in terms of these spectral radii . in section [ ss :
concl ] we put everything together to prove theorem [ thm : free.energy ] .
this section is the cornerstone of the following sections , since the representation of the partition function developed here will be used throughout the paper .
the observation that the edge - crossing numbers of the simple random walk have a markovian property goes back at least to knight @xcite .
this property can be formulated in various ways . in this section
we present a version that holds for a fixed time horizon , which is based on the well - known link between random walk excursions and rooted planar trees ( see remark [ rem : tree ] below ) .
we work conditionally on the event @xmath286 for fixed @xmath287 and @xmath288 , and w.l.o.g .
we assume that @xmath289
. then all edges are crossed the same number of times upwards and downwards , except for the edges in the stretch @xmath290 , which have one extra upward crossing .
we define the _ edge - crossing number _ @xmath291 , @xmath292 , as the number or upward crossing of the edge @xmath293 that are eventually followed by a downward crossing ( i.e. , we disregard the last upward crossing for @xmath294 ) . to keep the notation symmetric , we define @xmath295 , @xmath292 , as the number of downward crossings of the edge @xmath296 , each of which necessarily is eventually followed by an upward crossing . in formulas , @xmath297 for ease of notation
we suppress the dependence on @xmath49 .
in what follows we will work with the _ local times _ of the random walk , i.e. , the _ site visit numbers _ defined by @xmath298 these can be expressed in terms of the edge - crossing numbers as follows : @xmath299 we next define a specific _ branching process _ , which will be shown to be closely linked to the edge - crossing numbers @xmath300 , @xmath292 .
[ def : proc ] fix @xmath301 .
define a two - species _ branching process _
@xmath302 with law @xmath303 as follows : * at generation 0 there are @xmath231 individuals , which are divided by fair coin tossing into two subpopulations , labelled @xmath304 and @xmath305 . *
each subpopulation evolves independently as a _
critical galton - watson branching process with a geometric offspring distribution _ , denoted by @xmath306 and given by @xmath307 , @xmath308 .
* if @xmath309 , then there is additional immigration of a @xmath310-distributed number of individuals in the @xmath304 subpopulation , at each generation @xmath311 ( equivalently , the generations @xmath312 have an additional `` hidden '' individual , which is not counted but produces offspring ) . *
define @xmath300 as the size of the @xmath313 subpopulation in the @xmath314-th generation .
define the total population size @xmath315 and note that @xmath316 a.s . because a critical galton - watson process eventually dies out .
we can now state the main result of this section .
abbreviate @xmath317 .
[ th : bra ] fix @xmath318 such that @xmath319 , @xmath320 and @xmath321 is even .
the edge - crossing numbers @xmath322 of the simple random walk defined in conditionally on @xmath323 have the same joint distribution as the branching process with law @xmath324 defined in definition [ def : proc ] conditionally on @xmath325 . in formulas , @xmath326 for all sequences @xmath327 . taking the scaling limit of we obtain the famous ray - knight relation between brownian motion local time and squared bessel processes ( see revuz and yor @xcite ) .
we refer to tth @xcite for analogous relations involving more general processes , arising in the context of self - interacting random walks . before proving theorem [ th : bra ] , we note that the transition kernel of a critical galton - watson branching process with geometric offspring distibution is given by the matrix @xmath57 , @xmath328 , defined in .
in fact , if @xmath329 are i.i.d .
@xmath310 random variables , then @xmath330 in the presence of immigration , the transition kernel becomes @xmath331 . by definition [ def : proc ]
, @xmath322 is a markov chain on @xmath64 that is not time - homogeneous whenever @xmath332 ( due to the immigration ) .
the initial distribution of this markov chain is @xmath333 while the transition kernel factorizes , i.e. , it is the product of its marginals , because conditionally on @xmath334 the two components @xmath335 and @xmath336 evolve independently , with marginal transition kernels @xmath337 we are now ready to give the proof of theorem [ th : bra ] .
note that both sides of vanish , unless the sequences @xmath338 satisfy the conditions @xmath339 the first condition holds because @xmath340 for the random walk ( each visit to zero is preceded by a crossing of either @xmath341 or @xmath342 ) , while @xmath343 for the branching process by construction .
analogously , the second condition in holds for the branching process by the definition of @xmath344 in , while it holds for the random walk , because the total number of steps @xmath49 equals the total number of upward or downward crossings , which is given by @xmath345 ( recall that the last upward crossing of a bond in the stretch @xmath290 is not counted in @xmath291 ) .
henceforth we fix two sequences @xmath346 that satisfy .
below we will show that the number of simple random walk paths @xmath347 contributing to the event @xmath348 equals @xmath349 where @xmath350 the first product in is @xmath351 when @xmath352 , by convention .
note that @xmath338 have a finite sum by , and hence are eventually zero : @xmath353 for large enough @xmath314 .
since @xmath354 , this means that the products in are finite .
we can now prove .
the probability in the left - hand side of is obtained after dividing by @xmath355 , which is the total number of random walk paths .
recalling , and , we obtain @xmath356 which is precisely the probability in the right - hand side of , by the markov property of the process @xmath322 under the law @xmath324 ( recall - ) .
it remains to prove .
observe that @xmath357 in equals the number of ways in which @xmath358 objects can be allocated to @xmath18 boxes , i.e. , the number of sequences @xmath359 satisfying @xmath360 . as to the random walk , the crossing number @xmath361 counts the number of excursions below @xmath41 , while the crossing number @xmath362 counts the number of excursions below @xmath35 .
the key observation is that each excursion below @xmath35 is included in precisely one excursion below @xmath41 .
therefore the number of ways in which the @xmath362 excursions below @xmath35 can be `` allocated '' to the @xmath361 excursions below @xmath41 equals @xmath363 . iterating this argument
, we see that the last product in counts the number of random walk paths in the negative half - plane that are compatible with the given bond crossing - numbers @xmath364 . for the positive part of the random walk path
there is one difference .
when @xmath309 , each of the @xmath365 excursions above @xmath33 can be allocated not only to the @xmath366 excursions above @xmath41 , but also to the last incomplete excursion leading to @xmath55 .
this explains the presence of `` @xmath33 '' in the combinatorial factor @xmath367 in .
this holds until level @xmath55 , while above level @xmath55 the combinatorial factor @xmath368 applies .
the first two products in therefore count the number of random walk paths in the positive half - plane that lead to @xmath55 and are compatible with the given bond crossing - numbers @xmath369 .
it remains to combine the positive and the negative parts of the random walk path we have just built .
this can be done by alternating the @xmath370 positive excursions and the @xmath371 negative excursions in an arbitrary way , while preserving their relative order .
since @xmath372 , this can be done in @xmath373 ways , which leads to the first factor in .
[ rem : tree ] one way to visualize is to identify a random walk excursion with a planar rooted tree ( the random walk path traces out the `` external boundary '' of the tree ) . with this identification , the bond - crossing numbers @xmath374 represent the number of branches of the tree at level @xmath292 , and the number of such trees is given by @xmath375 .
the first step in our analysis of the free energy is to rewrite the annealed partition function as the partition function of the simple random walk weighted by a functional of its local times @xmath376 defined in . to that end
we define , for @xmath377 and @xmath48 , @xmath378.\ ] ] note that @xmath62 , so that @xmath379 .
[ lem : id_part_fct ] for @xmath42 and @xmath48 , @xmath380.\ ] ] rewrite as @xmath381 = { \ensuremath{\mathrm{e}}}\left [ \,\prod_{x\in { { \ensuremath{\mathbb z } } } } { { \ensuremath{\mathbb e } } } ^{\delta } \left [ e^ { -\beta \omega_{l_n(x)}^2 } \right ] \right],\ ] ] and use .
depending on which phase we are working in , it will be convenient to also use the function @xmath382 defined by @xmath383 which equals by ( recall and ) , and to rewrite lemma [ lem : id_part_fct ] as @xmath384 = { \ensuremath{\mathrm{e}}}\left[\exp\left\{\sum_{x\in{{\ensuremath{\mathbb z } } } } g^*_{\delta,\beta}(l_n(x))\right\}\right]\ ] ] with ( recall ) @xmath385 in appendix [ appa ] we collect some properties of @xmath60 that will be needed along the way .
* example : * if the marginal of @xmath24 is standard normal , then by direct computation @xmath386 * remark : * we close this section with the following observation . in section [ ss : model ] we argued that working with rather than as the interaction hamiltonian amounts to replacing @xmath45 by @xmath56 and adding a charge bias .
indeed , this is immediate from the relation @xmath387 for the annealed model , the last sum is not constant ( unless @xmath388 ) . to handle this , define @xmath389 as the product law with marginal given by @xmath390 where we need to assume that @xmath391 for all @xmath202 .
we put @xmath392,\ ] ] which is the same as but with @xmath393 instead of @xmath25 , and we define the partition function @xmath394.\ ] ] including the last sum in amounts to switching from @xmath395 to @xmath396 .
as mentioned in section [ ss : model ] , in this paper we work with the hamiltonian without the last sum .
the reader may check that @xmath397 has the same qualitative properties as @xmath60 , so that all the computations carried out below can be easily transferred . to compute the annealed free energy we use the generating function associated with the sequence of excess annealed partition functions , i.e. , @xmath398 where @xmath399 . the main result of this section is the following matrix representation of @xmath400 .
recall the matrices @xmath401 and @xmath402 defined in , and introduce an extra matrix @xmath403 [ prop : grcanspec ] for @xmath404 and @xmath48 , , @xmath328 , we define @xmath405 by putting @xmath406 , @xmath328 , with @xmath407 .
this is well - defined as a matrix with entries in @xmath408 $ ] , and if all the entries are finite , then @xmath405 is the inverse of @xmath409 ( where @xmath351 denotes the identity matrix ) and commutes with @xmath410 .
hence we can write @xmath411 without ambiguity . ]
@xmath412(0,0).\ ] ] to lighten the notation , we _ suppress the dependence on _ @xmath413 .
recalling , we can write @xmath414 \,,\ ] ] where @xmath415 accounts for the contribution of @xmath416 , which is the same as the contribution of @xmath286 . recalling , we can write @xmath417 next we apply theorem [ th : bra ] to rewrite the probability in the right - hand side of as a product of @xmath418 matrices , as in . in order to do this
, we must restrict to sequences @xmath419 satisfying the two conditions in .
thus , we define @xmath420 and restrict the sum over @xmath419 in to the set @xmath421 . at this point
we combine the definitions of the matrix @xmath418 , the function @xmath422 and the factor @xmath82 into the matrix @xmath423 defined in : @xmath424 we also introduce an extra matrix @xmath425 by @xmath426 which almost coincides with the matrix @xmath427 introduced in , the difference being that @xmath428 while @xmath429 .
altogether , we can rewrite as ( recall ) @xmath430 next we introduce the set of sequences @xmath338 that are _ eventually zero _ and satisfy @xmath372 : @xmath431 and observe that , for every fixed @xmath289 , @xmath432 the inclusion @xmath433 is obvious because sequences in @xmath434 are integer - valued with a finite sum and hence are eventually zero .
conversely , if @xmath338 are eventually zero , then @xmath435 for some @xmath436 .
this means that @xmath437 where the last line provides a convenient parametrization of the set @xmath438 : * sum over all possible values of @xmath439 ( when @xmath440 ) .
* denote by @xmath441 the smallest value of @xmath442 for which @xmath443 .
likewise denote by @xmath444 the smallest value of @xmath445 for which @xmath446 , so that by construction @xmath447 and @xmath448 can only take values in @xmath449 .
* it is implicit that @xmath443 for @xmath450 and @xmath451 for @xmath452 .
we now apply to .
we further split @xmath453 , where @xmath454 is the contribution of the single term @xmath352 , while @xmath455 is the sum @xmath456 . observing that @xmath457 by and
, we have ( with @xmath458 by convention ) @xmath459 and @xmath460 finally , we observe that we can replace @xmath461 by @xmath462 in , because for @xmath10 these two matrices coincide . after this replacement ,
the ranges @xmath463 and @xmath464 in the sums can be replaced by @xmath465 and @xmath466 , respectively , because @xmath467 .
this leads us to the desired matrix product representations : @xmath468(0,0 ) = \bigg [ \frac{1}{1- { \widetilde}a^\intercal } \ , \widehat a \ , \frac{1}{1 - { \widetilde}a } \bigg ] ( 0,0 ) \end{split}\ ] ] and @xmath469(0,0)\\ & = \bigg [ \frac{1}{1- { \widetilde}a^\intercal } \ , \frac{a}{1 - a}\ , \widehat a \ , \frac{1}{1 - { \widetilde}a } + \frac{1}{1- { \widetilde}a^\intercal } \ , \widehat a \ , \frac{a}{1 - a}\ , \frac{1}{1 - { \widetilde}a } \bigg ] ( 0,0)\\ & = \bigg [ \frac{1}{1- { \widetilde}a^\intercal } \ , \widehat a \ , \frac{2 a}{1 - a } \ , \frac{1}{1 - { \widetilde}a } \bigg ] ( 0,0 ) .
\end{split}\ ] ] summing these two formulas , we obtain the formula .
note that @xmath470 plays only a minor role in : the divergence of @xmath400 is controlled by @xmath70 and @xmath471 . for later use we state a version of proposition [ prop : grcanspec ] for bridges .
namely , let @xmath472 with @xmath473.\ ] ] [ lem : grcanspecloop ] for @xmath404 and @xmath48 , @xmath474(0,0).\ ] ] because of the bridge condition , the term with @xmath454 is absent , while in @xmath455 only the part with the @xmath475 s survives ( without @xmath476 ) .
in this section we prove some properties of the matrices @xmath70 in and @xmath471 in , viewed as operators from @xmath477 into itself .
these will be needed to exploit propositions [ prop : grcanspec][lem : grcanspecloop ] .
[ pr : prop.a.lambda.mu ] for all @xmath478 , the matrix @xmath479 has strictly positive entries , is symmetric , and is hilbert - schmidt .
it is obvious that @xmath70 has strictly positive entries and is symmetric .
the hilbert - schmidt norm of the operator @xmath70 is defined by @xmath480 ^ 2 \right)^{1/2}.\ ] ] by and in appendix [ appa ] , we have ( @xmath481 denotes a generic constant that may change from line to line ) @xmath482 observe that , by , @xmath483 where @xmath7 is a simple random walk .
we may therefore write @xmath484 ^ 2 \leq \frac{c}{i+j+1}\ , { \ensuremath{\mathrm{p}}}(s_{i+j } = i - j)^2 = \frac{c}{i+j+1}\ , { \ensuremath{\mathrm{p}}}^{\otimes 2}(s_{i+j } = s'_{i+j } = i - j),\ ] ] where @xmath485 is an independent copy of @xmath16 . using the change of variables @xmath486
, @xmath487 , we obtain @xmath488 ^ 2 & \leq c \sum_{u\in{\mathbb{n}}_0,\,v\in { \mathbb{z } } } \frac{1}{1+u}\ , { \ensuremath{\mathrm{p}}}^{\otimes 2}(s_u = s'_u = v)\\ & = c \sum_{u \in { \mathbb{n}}_0 } \frac{1}{1+u}\ , { \ensuremath{\mathrm{p}}}^{\otimes
2}(s_u = s'_u ) \leq c \sum_{u \in { \mathbb{n } } } u^{-3/2 }
< \infty , \end{aligned}\ ] ] which proves that @xmath489 .
thus , @xmath70 maps @xmath477 into itself . adapting the proof of proposition [ pr : prop.a.lambda.mu ]
, we see that @xmath490 and @xmath66 also map @xmath477 into itself .
let @xmath68 be the spectral radius of @xmath70 . by proposition [ pr : prop.a.lambda.mu ]
, @xmath70 is hilbert - schmidt , and hence is compact ( see dunford and schwartz ( * ? ? ?
* xi.6 , theorem 6 ) ) .
therefore @xmath68 is also the largest eigenvalue of @xmath70 ( see kato ( * ? ? ?
* v.2.3 . ) ) , and admits the rayleigh characterization ( see dunford and schwarz ( * ? ? ? * x.4 ) ) @xmath491 since the entries of @xmath70 are strictly positive , we have @xmath492 where @xmath493 means that all coordinates of @xmath494 are non - negative .
moreover , by the perron - frobenius theorem , there exists an eigenvector @xmath495 with @xmath496 , associated with @xmath68 with multiplicity one , whose coordinates are all strictly positive ( see baillon , clment , greven and den hollander ( * ? ? ?
* lemma 9 ) ) . we next state the regularity and the monotonicity of @xmath497 on @xmath498 .
[ prop : regl ] the following properties hold : + ( i ) @xmath499 is finite , jointly continuous and log - convex on @xmath498 , and analytic on @xmath500 .
+ ( ii ) for @xmath48 , @xmath72 is strictly decreasing and strictly log - convex on @xmath74 , with @xmath501 .
+ ( iii ) for @xmath502 , @xmath503 is strictly decreasing on @xmath81 .
\(i ) by proposition [ pr : prop.a.lambda.mu ] , @xmath70 has strictly positive entries , and as an operator on @xmath477 it is positive , irreducible and hilbert - schmidt , and hence compact .
consequently , we can apply the perron - frobenius theorem to obtain that @xmath68 is a simple eigenvalue of @xmath70 .
since each entry of @xmath70 is continuous and analytic in @xmath504 on @xmath498 , we can apply crandall and rabinowitz ( * ? ? ?
* lemma 1.3 ) to get the claimed continuity and analyticity of @xmath505 .
to get log - convexity , use the variational characterization of @xmath68 in and note that : ( a ) @xmath506 is log - convex for each @xmath507 , ( b ) the sum and the product of two log - convex functions is log - convex , ( c ) the supremum of convex functions is convex .
( use that a log - convex function can be written as a supremum over log - linear functions . ) + ( ii ) pick @xmath508 and set @xmath509 , @xmath510 . since @xmath511 has strictly positive entries and @xmath512 we have @xmath513 which yields the first claim .
the second claim follows from the fact that @xmath72 is analytic on @xmath81 and log - convex but not log - linear on @xmath74 .
the third claim follows from the estimate @xmath514 .
+ ( iii ) pick @xmath515 and set @xmath516 , @xmath510 . by proposition
[ pr : gbetadelta.monotonicity ] in appendix [ appa ] , we have @xmath517 and so we can repeat the argument in ( ii ) . two situations emerge , which correspond to the ballistic phase @xmath104 , respectively , the subballistic phase @xmath105 ( see figs . [ fig - spectral][fig - critcurve ] and recall ): * if @xmath518 , then there is a unique @xmath519 for which @xmath520 . * if @xmath521 , then there is no @xmath404 for which @xmath522 , and we set @xmath523 . for later use we state the following gap .
[ prop : gap ] @xmath524 for all @xmath525 .
let @xmath526 we prove the strict inequality by showing that @xmath527 the first inequality in is a consequence of the following relation : @xmath528 where @xmath529 is any bounded linear operator and @xmath530 is the spectral radius of @xmath529 ( dunford and schwartz ( * ? ? ?
* vii.3 ) ) .
we now prove the second inequality in . to that end , define @xmath531 for @xmath532 , we have @xmath533 since @xmath534 , this yields @xmath535 it therefore remains to show that @xmath536 . to that end , note that , since @xmath70 is compact and symmetric , there exist eigenvalues @xmath537 in @xmath2 and associated eigenvectors @xmath538 in @xmath477 such that @xmath539 ( see kato ( * ? ? ?
* theorem 6.38 , section 6.9 ) and zerner ( * ? ? ?
* theorem 1 ) ) .
let @xmath540 be such that @xmath541 .
write @xmath542 , with @xmath543 and @xmath544
. then @xmath545 if @xmath546 , then by the cauchy - schwarz inequality @xmath547 ^ 2 = 1-[\nu_0(0)]^2.\ ] ] since @xmath548 and @xmath549 , we get from
@xmath550 ^ 2 ) + |{\lambda}_1|^2 [ \nu_0(0)]^2 } < { \lambda}_0 = \|a_{\mu,\delta,\beta}\|_{\rm op}.\ ] ] finally , the last equality in follows immediately by the decomposition .
we show that @xmath400 in is finite . indeed , by the definition of @xmath553 and proposition [ prop : regl](ii ) , we know that @xmath554 , and therefore @xmath555 is invertible . moreover , since @xmath556 is symmetric , it is also normal ( i.e. , @xmath557 ) , and so the inverse of @xmath555 equals @xmath558 . thanks to proposition [ prop : gap ] we also have @xmath559
. therefore @xmath400 is finite .
suppose that @xmath561 ( otherwise there is nothing to prove ) .
we show that @xmath400 in is infinite .
indeed , by lemma [ lem : grcanspecloop ] , @xmath562 for @xmath563 , let @xmath564 and observe that @xmath565^x ( \check{a}^x_{\mu,{\delta},{\beta}})(0,0).\ ] ] since @xmath566 , the proof is complete once we show that the sequence of positive numbers @xmath567 is bounded away from @xmath41 . to that end , observe that @xmath568 is a transition matrix on @xmath569 with @xmath570 as invariant probability measure .
therefore , by the renewal theorem , @xmath571 which completes the proof .
we are finally ready to conclude the proof of theorem [ thm : free.energy ] . as shown above , @xmath572 is finite for @xmath552 and infinite for @xmath560
. therefore we have proven that @xmath573 below we show that @xmath574 combining and recalling , we get the spectral representation in in theorem [ thm : free.energy ] . since @xmath575 , we also get the lower bound in in theorem [ thm : free.energy ] .
the fact that @xmath576 is convex is immediate from .
consider the bridge version of the partition function defined in .
the sequence @xmath577 is super - additive because the concatenation of two bridges is again a bridge .
hence @xmath578 it follows from lemma [ lem : grcanspecloop ] that the limit equals @xmath83 . since @xmath579 , this settles
in section [ ss : critcurve ] we prove the qualitative properties of the excess free energy and the critical curve stated in theorem [ thm : critcurve ] . in section [ ss : jointldp ]
we prove the large deviation principle for the speed and the charge stated in theorem [ thm : jointldp ] , while in section [ ss : rfshape ] we show how the shape of the associated rate functions in figs .
[ fig - rfspeed][fig - rfcharge ] come about . in section [ cltproof ]
we prove the central limit theorem for the speed and the charge stated in theorem [ thm : cltspeedcharge ] . in section [ ss : lln ]
we prove the law of large numbers for the speed and the charge stated in theorems [ thm : speed][thm : charge ] . in this section
we prove theorem [ thm : critcurve ] .
fix @xmath99 .
clearly , @xmath580 is non - increasing and convex on @xmath81 ( see ) , and hence is continuous on @xmath81 . by theorem [ thm : free.energy ]
, we know that @xmath581 . since @xmath582 is non - increasing and continuous , there exists a @xmath583 such that @xmath584 when @xmath585 and @xmath586 when @xmath587 . since @xmath588 is convex on @xmath589 , the level set @xmath590 is convex , and it follows that @xmath591 ( which coincides with the boundary of this level set ) is also convex .
next we prove that @xmath592 for @xmath182 . by ,
@xmath593 for @xmath99 and @xmath594 .
hence , by the definition of the hilbert - schmidt norm of the operator @xmath70 in , we have @xmath595 .
but @xmath596 and , since @xmath70 is normal , @xmath597 , so that @xmath598 .
thus , @xmath599 for @xmath45 large enough , and so @xmath600 . also , observe that @xmath601 for @xmath182 , which yields that @xmath602 for @xmath603 . finally , since @xmath604 for @xmath202 ( recall ) , we get @xmath101 . the convexity of @xmath605 and the fact that @xmath606 for @xmath607 imply that @xmath605 is strictly increasing .
the continuity of @xmath605 follows from convexity and finiteness . from and
it follows that , for @xmath32 , @xmath608 coincides with the unique solution of @xmath609 .
it therefore follows from proposition [ prop : regl](i ) and the implicit function theorem that @xmath610 is analytic on @xmath81 .
finally , since @xmath611 , it follows from , proposition [ prop : regl](i ) and the implicit function theorem that @xmath612 is analytic on @xmath117 . in this section
we prove theorem [ thm : jointldp ] .
the proof comes in 6 steps . *
we begin by introducing the joint moment - generating function for the speed and the charge .
fix @xmath48 and @xmath613 .
let @xmath614\\ & = ( { { \ensuremath{\mathbb e } } } \times { \ensuremath{\mathrm{e}}})\big [ e^{\gamma s_n + ( { \delta}+ \gamma'){\omega}_n - { \beta}h^\omega_n(s ) } \big]\\ & = { \ensuremath{\mathrm{e}}}\big [ e^{\gamma s_n + \sum_{y\in { { \ensuremath{\mathbb z } } } } g^*_{{\delta}+\gamma',{\beta}}(l_n(y ) ) } \big ] .
\end{aligned}\ ] ] then @xmath615 = \frac{{{\ensuremath{\mathbb z } } } _ n^{*,\delta,\beta}(\gamma,\gamma')}{{{\ensuremath{\mathbb z } } } _ n^{*,\delta,\beta}(0,0)},\ ] ] where we recall that @xmath616 is the expectation w.r.t . the annealed polymer measure of length @xmath49 defined in ( [ eq : def.polmeasure][eq : def.annpartfunc ] ) .
next , let @xmath617 then @xmath618 has a spectral representation similar to the one in proposition [ prop : grcanspec ] .
indeed , the only difference is that @xmath70 must be replaced by @xmath619 and @xmath53 by @xmath620 .
the same is true for the bridge version of the moment - generating function , for which lemma [ lem : grcanspecloop ] holds with the same replacement ( see proposition [ prop : ldp.bridge ] ) . recall . repeating the argument in sections [ ss : spec][ss : concl ]
, we obtain that @xmath621 = \lim_{n\to\infty } \frac{1}{n } \log \frac{{{\ensuremath{\mathbb z } } } _ n^{*,\delta,\beta}(\gamma,\gamma')}{{{\ensuremath{\mathbb z } } } _ n^{*,\delta,\beta}(0,0)}\ ] ] is given by @xmath622 - \mu(\delta,\beta).\ ] ] here , the second term in the right - hand side comes from the denominator in , while the first term captures the crossover from a regime where the spectral radius of @xmath623 controls the blow up of the generating function to a regime where the spectral radius of @xmath624 does .
the result in allows us to apply the grtner - ellis theorem of large deviation theory ( den hollander ( * ? ? ?
* chapter v ) ) and obtain that the pair consisting of the empirical speed and the empirical charge satisfies the large deviation principle on @xmath74 , with the associated rate function @xmath625 given by the legendre transform of @xmath626 , i.e. , @xmath627.\ ] ] actually , the grtner - ellis theorem only gives us the large deviation principle in the regions where @xmath625 is `` exposed '' , i.e. , where @xmath628 is strictly convex . in the regions where @xmath628 is flat
, it only gives a lower bound on the rate function and so we need to provide a matching upper bound .
to prove the upper bound , we restrict ourselves to the marginal large deviation functions , which are obtained from @xmath629 by setting @xmath630 , respectively , @xmath631 . substituting into
, we get the formulas for @xmath138 and @xmath632 in , where for the former we use that @xmath633 .
* 3 . * before embarking on the proof of the matching upper bounds for the flat pieces in the rate functions , we state two auxiliary propositions whose proofs are deferred to the end of section [ ss : jointldp ] . recall and define the laws @xmath634 by @xmath635 we also need a bridge version of : @xmath636.\ ] ] [ prop : ldp.bridge ] ( 1 ) the limit @xmath637 exists and is given by @xmath638 ( 2 ) let @xmath639 be defined by @xmath640_{\mu = 0},\ ] ] which satisfies @xmath641 ( see fig .
[ fig - spectral ] and ) .
for every open set @xmath642 , @xmath643 where @xmath644 , \qquad \theta \in [ 0,\infty),\ ] ] and @xmath645 ( 3 ) for every open set @xmath646 , @xmath647 where @xmath648 , \qquad \theta ' \in [ 0,\infty).\ ] ] and @xmath649 ( with additional work , and can be turned into large deviation principles . )
define the partition function restricted to loops : @xmath650,\qquad n\in 2{\mathbb{n}}.\ ] ] [ prop : loop.part.func ] the sequence @xmath651 converges .
* 4 . flat piece of @xmath652 .
* we start with the matching upper bound for @xmath652 on @xmath653 ( see fig .
[ fig - rfspeed ] ) . for ease of notation
, we omit to write integer parts .
let @xmath654 , and let @xmath195 be small enough so that @xmath655 .
it is enough to show that @xmath656 for some @xmath657 that tends to zero as @xmath658 .
note that @xmath659 in order to prove , we adopt the following strategy : the polymer moves to the right ballistically at speed @xmath660 for a fraction of time @xmath661 and spends the rest of the time making a loop to the right . recall .
as the reader can easily check , the method explained in section [ freeenergy ] leads to the following representation of the grand - canonical partition function restricted to loops : @xmath662(0,0 ) , \qquad \mu\geq 0.\ ] ] note that @xmath663 when @xmath664 , while @xmath665 when @xmath666 . by repeating the same argument as in section [ ss : spec ]
, we deduce that @xmath667 moreover , by proposition [ prop : loop.part.func ] , the limsup is actually a lim .
recall that @xmath668 write @xmath669 and abbreviate @xmath670 .
then the strategy above translates into @xmath671\\ & \qquad \geq { \sum_{\substack{m_- \leq m \leq m_+ \\ n - m \in 2{\mathbb{n } } } } } \bigg(\frac{{{\ensuremath{\mathbb z } } } _ { m,\mathrm{bridge}}^{*,{\delta},{\beta}}(e_m)}{e^{m\mu({\delta},{\beta})[1+o(1 ) ] } } \bigg ) \bigg(\frac{{{\ensuremath{\mathbb z } } } _ { n - m , \mathrm{loop}}^{*,{\delta},{\beta}}}{e^{(n - m)\mu({\delta},{\beta})[1+o(1 ) ] } } \bigg)\\ & \qquad \geq { \sum_{\substack{m_- \leq m \leq m_+ \\ n - m \in 2{\mathbb{n } } } } } { \ensuremath{\mathrm{p}}}_{m,\mathrm{bridge}}^{{\delta},{\beta}}(e_m)\ , \frac{e^{(n - m){\widetilde}\mu({\delta},{\beta})[1+o(1)]}}{e^{(n - m)\mu({\delta},{\beta})[1+o(1)]}}\\ & \qquad \geq c\ , n \,\exp\big[n[1+o(1 ) ] \quad \big\{- \frac{\theta}{{\widetilde}v({\delta},{\beta } ) ( 1 - { \varepsilon } ) } i^v_{{\delta},{\beta}}\big({\widetilde}v(\delta,\beta)(1+{\varepsilon}^2)\big)\\ & \qquad\qquad\qquad\qquad + \big(1- \frac{\theta}{{\widetilde}v(\delta,\beta ) ( 1+{\varepsilon})}\big ) [ { \widetilde}\mu({\delta},{\beta } ) - \mu({\delta},{\beta})]\big\ } \big ] , \end{aligned}\ ] ] where @xmath481 is a positive constant that depends on @xmath672 , and @xmath673 is short - hand notation for the bridge partition function restricted to the event @xmath674 .
we use for the last inequality . by continuity of @xmath675 at @xmath676 and ,
this is the desired result .
note that the last inequality holds because of proposition [ prop : ldp.bridge ] . also note also that the @xmath677 s are uniform on @xmath678 because @xmath679)n$ ] .
* 5 . flat piece of @xmath680 .
* we now turn to the matching upper bound for @xmath680 on @xmath681 ( see fig . [ fig - rfcharge ] ) .
we call a path @xmath682 a _ half - bridge _ when @xmath683 for all @xmath684 , and we define @xmath685.\ ] ] let @xmath686 .
the strategy is similar as above : fix @xmath687 , let the first @xmath688 charges have an empirical average close to @xmath689 , and let the remaining @xmath690 charges have an empirical average close to @xmath41 . to be more precise ,
write @xmath691 and abbreviate @xmath692 and @xmath693 .
estimate @xmath694 thanks to and the grtner - ellis theorem , we know that @xmath695 below we will prove that @xmath696 recalling that @xmath697 and using the continuity of @xmath698 , we therefore obtain @xmath699 where @xmath700 as @xmath658 , which is the desired result . * 6 .
* it remains to prove . to that end , abbreviate @xmath701 and estimate @xmath702.\ ] ] the following strategy gives a lower bound .
fix @xmath703 and @xmath704 large , and define the events ( to ease the notation we omit writing integer parts ) @xmath705 \,\,\forall\,0\leq i \leq n\},\\ { { \ensuremath{\mathcal b } } } _ n & = \{\ell_n(x ) \leq c n^{1-{\alpha } } \,\,\forall\,x \in [ 0,n^{\alpha}]\},\\ { { \ensuremath{\mathcal c } } } _ n & = \big\ { \omega_n^x \in ( 0,m ] \,\,\forall\,x \in [ 0,n^{\alpha } ] \big\}. \end{aligned}\ ] ] on the event @xmath706 , we have @xmath707 therefore , for @xmath49 large enough , @xmath708.\ ] ] next , by the independence of the charges and the local central limit theorem , we have , for @xmath529 large enough , @xmath709 & \geq { \ensuremath{\mathrm{e}}}\left [ \big\{{{\ensuremath{\mathbb p } } } ( { \omega}^0_\ell \in ( 0,m ] \,\,\forall\ , \ell \leq cn^{1-{\alpha}}\big\}^{n^{\alpha}+1}\ , { { \sf 1}}_{\{{{\ensuremath{\mathcal a } } } _
n \cap { { \ensuremath{\mathcal b } } } _ n\}}\right ] \\ & \geq \left ( \frac{c_m}{\sqrt{cn^{1-{\alpha } } } } \right)^{n^{\alpha}+1 } { \ensuremath{\mathrm{p}}}({{\ensuremath{\mathcal a } } } _
\cap { { \ensuremath{\mathcal b } } } _ n ) , \end{aligned}\ ] ] where @xmath710 is a constant that depends on @xmath529 . in addition
, there exists a constant @xmath711 such that @xmath712 indeed , the probability that a simple random walk stays inside @xmath713 $ ] up to time @xmath49 is @xmath714 , \qquad n\to\infty,\ ] ] where @xmath715 is the principal dirichlet eigenvalue of the continuous laplacian on @xmath716 $ ] .
moreover , for @xmath481 large enough we have @xmath717 , because @xmath718}$ ] conditionally on @xmath719 converges in distribution to the square of the principal dirichlet eigenfunction . combining ( [ znanbncn][pnanbncn ] ) , we arrive at @xmath720,\ ] ] which yields .
the proof consists in a slight modification of the arguments at the beginning of this section , see . indeed , define @xmath721 a minor change in gives @xmath722(0,0),\ ] ] from which we deduce .
let @xmath723 as shown in fig .
[ fig - spectralalt ] .
observe that @xmath724 for @xmath725 , which yields .
finally , is an immediate consequence of @xmath726 . both and follow from the grtner - ellis theorem . for @xmath727 ,
let @xmath728 be the space of right - loops of length @xmath49 , that is @xmath729 for @xmath730 , define @xmath731 as the following subset of @xmath732 : @xmath733,\,s_j \geq s_a,\,\,\forall\,j \in [ a , a+m]\big\}. \end{aligned}\ ] ] note that : ( i ) there are at most @xmath734 ways of cutting a loop in @xmath732 to obtain two loops in @xmath735 and @xmath728 after gluing the two end - parts together ; ( ii ) the loops obtained in this way involve two independent sets of charges . therefore @xmath736
\geq \frac{1}{n+1 } { { \ensuremath{\mathbb z } } } _ { m,\mathrm{loop}}^{*,{\delta},{\beta } } { { \ensuremath{\mathbb z } } } _ { n,\mathrm{loop}}^{*,{\delta},{\beta}}\ ] ] and so @xmath737.\ ] ] hence the sequence @xmath738 is almost super - additive , and an application of hammersley ( * ? ? ?
* theorem 1 ) gives the result . in this section
we show how figs .
[ fig - rfspeed][fig - rfcharge ] come about .
recall from that @xmath739 when @xmath740 . differentiating this equation twice with respect to @xmath741
, we obtain @xmath742^{-1}_{\mu=\mu(\delta,\beta,\gamma)},\\ \frac{\partial^2\mu}{\partial\gamma^2}(\delta,\beta,\gamma ) & = \left[\frac{\partial^2}{\partial\mu^2}\log\lambda_{\delta,\beta}(\mu ) \right]_{\mu=\mu(\delta,\beta,\gamma ) } \left[\frac{\partial\mu}{\partial\gamma}(\delta,\beta,\gamma)\right]^3 . \end{aligned}\ ] ] from this and the strict convexity ( respectively , monotonicity ) of @xmath743 we see that @xmath744 is strictly convex ( respectively , increasing ) for @xmath740 ( recall fig .
[ fig - spectral ] ) .
moreover , since @xmath745 , we know that @xmath746 is strictly increasing and strictly convex as well .
( 12,12)(0,-4 ) ( 0,0)(1,0)12 ( 0,-6)(0,1)14 ( 0,6)(4,0)(12,-3 ) ( 0,1.5)(2,0)(10,-3.5 ) ( -1,-.5)@xmath41 ( 12.5,-0.2)@xmath82 ( 6,.7)@xmath83 ( .6,-1.5)@xmath84 ( 13,-3)@xmath747 ( 11,-4.5)@xmath748 ( -4.6,2.5)@xmath749 ( 0,6 ) ( 0,1.6 ) ( 6,0 ) ( 2.6,0 ) ( 2.6,0)(2.6,1.3)(2.6,2.7 ) ( 2.6,2.7)(1.6,2.7)(0,2.7 ) ( 12,12)(-4,-4.5 ) ( -7,0)(1,0)15 ( 0,-2)(0,1)10 ( -6,2)(-4,2)(-2,2 ) ( -2,2)(0,3)(3,6 ) ( -2,0)(-2,1)(-2,2 ) ( -2,2)(-1,2)(0,2 ) ( 0.5,-1)@xmath41 ( 8.5,-0.2)@xmath741 ( -5,8.7)@xmath750 ( -4.3,-1.2)@xmath751 ( 1,3.1)@xmath83 ( 1,1.7)@xmath84 ( -2,2 ) ( 0,3.3 ) recall . by and , @xmath752 when @xmath753 and @xmath754 when @xmath755 .
hence @xmath756 has the shape depicted in fig .
[ fig - maxmus ] .
the right slope of this function at @xmath757 is precisely the speed @xmath140 defined in .
hence @xmath758 , which is given by the legendre transform in , is linear on @xmath145 $ ] and strictly convex on @xmath759 . for @xmath760 the supremum over @xmath741 in
is taken at @xmath757 , so that @xmath761 = \big[\mu(\delta,\beta ) - { \widetilde}{\mu}(\delta,\beta)\big ] - \theta \log \lambda_{\delta,\beta}({\widetilde}{\mu}(\delta,\beta)).\ ] ] ( 12,12)(0,-2 ) ( 0,0)(12,0)12 ( 0,0)(0,8)8 ( 0,0)(5,0.5)(9,6.5 ) ( -1,-.5)@xmath41 ( 12.5,-0.2)@xmath53 ( -0.3,8.5)@xmath45 ( 0,0 ) ( 9,2)@xmath104 ( 3,4)@xmath105 ( 8.5,7.3)@xmath96 ( 4.7,-1.2)@xmath762 ( 7.8,-1.2)@xmath53 ( -1,2.7)@xmath45 ( 8,3 ) ( 8,3)(7,3)(6,3 ) ( 6,3)(6,1.5)(6,0 ) ( 8,3)(8,1.5)(8,0 ) ( 0,3)(4,3)(8,3 ) ( 12,12)(-4,-4 ) ( -7,0)(1,0)15 ( 0,-2)(0,1)10 ( -6,0)(-4,0)(-2,0 ) ( -2,0)(0,1)(2,5 ) ( .5,.5)@xmath41 ( 8.5,-0.2)@xmath763 ( -3,8.7)@xmath764 ( -4,-1.4)@xmath765 ( -2,0 ) let @xmath766 as shown in fig .
[ fig - critcurvealt ] , where we recall that @xmath767 is the inverse of the critical curve . by ,
@xmath768 when @xmath769 and @xmath770 when @xmath771 .
hence @xmath772 has the shape depicted in fig .
[ fig - mu ] .
the right slope of this function at @xmath773 is precisely @xmath144 defined in .
hence @xmath774 , which is given by the legendre transform in , is linear on @xmath146 $ ] and strictly convex on @xmath775 . for @xmath776 the supremum over @xmath763 in
is taken at @xmath777 , so that @xmath778 = \mu(\delta,\beta ) - \theta'[\delta-\delta_c(\beta)].\ ] ] in this section we prove theorem [ thm : cltspeedcharge ] for the speed .
the extension of the argument to the charge is given in appendix [ appb ] .
the following sketch of the proof for the speed is inspired by knig @xcite .
the proof comes in 6 steps . *
we begin with a probabilistic interpretation of @xmath779 , @xmath780 .
we recall that the kernels @xmath781 and @xmath782 are defined in , and , respectively .
let @xmath783 be the joint law of two independent positive recurrent markov chains on @xmath569 , denoted by @xmath784 and @xmath785 , where @xmath786 has transition kernel @xmath787 given in below , and @xmath784 and @xmath788 have transition kernel @xmath789 given in below .
the law of @xmath790 is the invariant law for @xmath787 , while the law of @xmath791 is the invariant law for @xmath789 restricted to @xmath792 and normalised by @xmath793 ( @xmath41 is an absorbing state ) . in particular , @xmath794 with @xmath795 defined below .
let @xmath796 and define @xmath797 [ lem : clt1 ] there exist explicit functions @xmath798 , which are given in below , such that @xmath799 .
\end{aligned}\ ] ] an adaptation of the arguments leading to gives @xmath800 where @xmath801 and @xmath802 . recall that the subscripts @xmath803 are suppressed from @xmath782 , @xmath423 and @xmath425 , and the same observation as made below allows us to replace @xmath425 by @xmath804 in the line above . from now on we choose @xmath805 .
then , the spectral radius of @xmath423 is @xmath351 .
writing @xmath806 for its normalized associated eigenvector ( right or left , by symmetry ) , we note that @xmath807 is a transition matrix with invariant probability distribution @xmath808 .
let us take a closer look at @xmath427 .
note first that is valid for @xmath427 and @xmath809 ( the restriction of @xmath427 to @xmath810 ) , since both matrices are hilbert - schmidt and therefore define bounded linear operators on @xmath477 , respectively , @xmath811 .
next , by proposition [ prop : gap ] , @xmath812 .
write @xmath813 and observe that @xmath814 . indeed , by , for all
@xmath42 we have @xmath815 and so we get the claim by letting @xmath816 .
therefore there exists a vector @xmath817 with strictly positive entries such that @xmath818 this defines a transition matrix @xmath789 on @xmath64 given by @xmath819 note that @xmath41 is an absorbing state for @xmath789 .
let @xmath820 be a left eigenvector of @xmath821 .
we may normalise @xmath822 and @xmath823 such that @xmath824 .
recalling , from we obtain @xmath825 , \end{split}\ ] ] which completes the proof of .
henceforth we denote by @xmath826 the law of the time - homogeneous markov chain @xmath827 with transition kernel @xmath787 .
we then have the following representation .
[ lem : clt3 ] there exists an explicit function @xmath494 such that , for all @xmath828 , @xmath829 define @xmath830 { \ensuremath{\mathrm{p}}}^{{\delta},{\beta}}(m_0^+ = a)\\ & \qquad\qquad \times { \ensuremath{\mathrm{e}}}^{x,{\delta},{\beta}}\left [ \frac{{\widetilde}\nu(m_x^+)}{\nu(m_x^+ ) } \frac{{\widetilde}a(m^+_{\tau -1},0)}{{\widetilde}\nu(m^+_{\tau-1 } ) } { { \sf 1}}_{\{y^+ = n_0^+\ } } ~\big|~ m^+_{x-1}=b \right ] .
\end{split}\ ] ] ( note that the right - hand side in the line above does not actually depend on @xmath55 , because @xmath831 is by definition a markov chain with kernel @xmath789 given in . ) by the markov property and lemma [ lem : clt1 ] , @xmath832 the increments of @xmath833 are strictly positive , which means that the events in the line above are disjoint for different values of @xmath55 . therefore ,
summing over @xmath55 we get the claim .
* 3 . * from lemma [
lem : clt3 ] we see that the fluctuations of @xmath213 are related to the fluctuations of @xmath834 .
the key ingredient of our proof is the following lemma .
[ lem : clt4 ] for every @xmath835 , the process @xmath834 under @xmath836 satisfies the clt with mean and variance @xmath837_{\mu = \mu({\delta},{\beta } ) } , \qquad \sigma_y^2 = \left[\frac{\partial^2}{\partial\mu^2 } \log\lambda_{\delta,\beta}(\mu)\right]_{\mu=\mu(\delta,\beta)},\ ] ] i.e. , @xmath838 in law as @xmath839 .
recall that under @xmath826 the process @xmath840 is a positive recurrent markov chain , with transition kernel @xmath787 and invariant probability measure @xmath841 . following ( * ? ? ?
* lemma 4.1 ) , we are going to apply the clt for markov chains ( see chung ( * ? ? ?
* theorem 16.1 ) ) .
the strategy is as follows . for @xmath308 ,
denote by @xmath842 the @xmath18-th return time to the state @xmath167 of @xmath840 ( with @xmath843 ) and set @xmath844 .
then , we can decompose @xmath845 where @xmath846 by the strong markov property , under @xmath836 the random variables @xmath847 are i.i.d . with finite expectation @xmath848 .
consequently , by the strong law of large numbers , @xmath849 a.s .
also the random variables @xmath850 are i.i.d .
, again by the strong markov property .
if we show that @xmath851_{\mu = \mu({\delta},{\beta } ) } = \frac{{\ensuremath{\mathrm{e}}}^{{\delta},{\beta}}(u_1|m_0^+=a)}{\mu_t } = \frac{{\ensuremath{\mathrm{e}}}^{{\delta},{\beta}}(u_1|m_0^+=a)}{{\ensuremath{\mathrm{e}}}^{{\delta},{\beta}}(t_1|m_0^+=a)},\ ] ] then the random variables @xmath852 are i.i.d . and are centered . if we pretend that we may replace the upper index @xmath853 by @xmath854 in , and if we ignore the contribution of the `` boundary term '' @xmath855 , then the standard clt yields @xmath838 with @xmath856 but this is indeed justified by chung ( * ? ? ?
* theorem 16.1 ) .
the boundary term @xmath857 in is harmless .
it only remains to show that @xmath858 and @xmath859 , as defined in the statement of lemma [ lem : clt4 ] , ( are finite and ) satisfy relations - .
recalling the definition of the transition kernel @xmath860 , we can write @xmath861 with @xmath805 . recalling the definition of the kernel @xmath556 ,
when we differentiate this relation with respect to @xmath82 term by term and set @xmath805 , we get precisely relation . differentiating it a second time ,
we only need an argument analogous to ( * ? ? ?
* eq . ( 4.6 ) ) to interchange differentiation and summation .
this follows from an adaptation of ( * ? ? ?
* eq . ( 4.7)(4.11 ) ) .
in particular , ( * ? ? ?
* eq . ( 4.8 ) ) is replaced in our context by @xmath862 for @xmath863 small enough , which uses the fact that @xmath864 is bounded from above and @xmath865 .
* 4 . * fix @xmath866 and set @xmath867 with @xmath868_{\mu = \mu({\delta},{\beta})}$ ] as in lemma [ lem : clt4 ] ( the second equality in follows from the definition in of the speed @xmath155 ) . let us next look at the probability in with @xmath869 . for simplicity
, we first forget about the constraint @xmath870 . recall .
we prove that , for fixed @xmath871 and @xmath835 , @xmath872}}{\sqrt{2\pi \sigma_v({\delta},{\beta})^2}}\,{\text{\rm d}}z,\ ] ] where @xmath873 is defined in .
indeed , @xmath874 ~\big|~ m_0^+ = a \big ) , \end{split}\ ] ] and so the claim follows from lemma [ lem : clt4 ] and the fact that @xmath875 , by .
adapting the argument below ( * ? ? ?
* eq . ( 4.20 ) ) , we show that , for fixed @xmath876 , @xmath877 and @xmath878 , @xmath879 for an explicit positive constant @xmath880 . indeed , by the standard ergodic theorem for positive recurrent aperiodic markov chains , @xmath881 as @xmath882 , which ensures in particular that @xmath883 is tight , since @xmath884 . as a consequence , the analogue of ( * ? ? ?
* eq . ( 4.20 ) ) holds , i.e. , @xmath885 .
* we may now conclude the proof by showing that @xmath886 first , note that @xmath887 so that , by lemma [ lem : clt3 ] , @xmath888 equation follows from and by letting @xmath882 .
if we set @xmath889 then the limit is justified by the following points : 1 .
[ it:1 ] @xmath890 for all @xmath891 , 2 .
[ it:2 ] @xmath892 .
note that allows us to truncate the sums in the numerator and denominator of to @xmath893 , after which allows us to apply dominated convergence . to see why and hold , we first note that , by the definition of @xmath427 , @xmath894 where the @xmath895 comes from @xmath896 , cf . , and we note that @xmath422 gives a negligible contribution , by proposition [ pr : asymptgstar ] . since , by lemma [ lem : nurate ] below , @xmath897 it follows that @xmath898 from which , recalling , we deduce that for some positive constant @xmath481 , @xmath899\ , { \ensuremath{\mathrm{e}}}^{x,{\delta},{\beta}}\big [ \frac{{\widetilde}\nu(m^+_x)}{\nu(m^+_x ) } \big| m^+_{x-1 } = b \big ] .
\end{aligned}\ ] ] we remind that the second expectation does not actually depend on @xmath55 , by the markov property .
moreover , we claim that for every @xmath900 there exists @xmath901 such that @xmath902 \le c e^{{\varepsilon}b } \
, , \qquad b \in { \mathbb{n}}_0 \,.\ ] ] in fact , recalling the definition of the transition kernel @xmath787 of @xmath883 , @xmath903 = \sum_{k\in { \mathbb{n}}_0 } \frac{{\widetilde}\nu(k)}{\nu(k ) } \ , a(b , k ) \ , \frac{\nu(k)}{\nu(b ) } = \frac{1}{\nu(b ) } \sum_{k\in { \mathbb{n}}_0 } { \widetilde}\nu(k ) \ ,
a(b , k ) \,.\ ] ] since @xmath904 ( recall ) , it suffices to show that @xmath905 . to this end , we show that @xmath906 recalling the definition of @xmath423 , we note that , since @xmath422 is bounded from above ( by proposition [ pr : asymptgstar ] ) , @xmath907 with @xmath213 the simple symmetric random walk on @xmath908 . since @xmath909 ( recall ) , setting @xmath910 and @xmath911 , we get @xmath912 = c \sum_{j\in{\mathbb{n}}_0 } e^{(\log \cosh(r-{\varepsilon}/2)-\mu ) j } < \infty , \end{split}\ ] ] because @xmath913 ( recall ) .
we have proved , and hence . to prove , recall and .
bound @xmath914 ( because @xmath915 and @xmath422 is bounded from above by proposition [ pr : asymptgstar ] ) , to obtain via that @xmath916 { \ensuremath{\mathrm{p}}}^{{\delta},{\beta}}(m_0^+ = a ) & \leq \sum_{a,\ell \in { \mathbb{n}}_0 } \frac{\widehat a(\ell , a)}{{\widetilde}\eta ( \ell ) \nu(a ) } \nu(a)^2 { \widetilde}\eta(\ell ) { \widetilde}\nu(\ell)\\ & \leq c \sum_{a,\ell \in { \mathbb{n}}_0 } e^{-\mu(a+\ell ) } \nu(a ) { \widetilde}\nu(\ell ) < \infty .
\end{aligned}\ ] ] to prove , we first note that , by cauchy - schwarz , @xmath917 because @xmath918 ( recall ) , and the relation @xmath919 proved below . recalling and , we can finally estimate @xmath920 \bigg ) \bigg ( \sum_{b > b } e^{{\varepsilon}b } \nu(b ) \bigg ) \ , .
\end{aligned}\ ] ] since @xmath921 ( recall ) , the sum in the second parenthesis converges , and vanishes as @xmath922 . to complete the proof of
, it suffices to show that the first parenthesis is finite .
similarly as above , recalling , and bounding @xmath923 ( recall ) , we get @xmath924 \leq c \sum_{a,\ell \in { \mathbb{n}}_0 } e^{-\mu(a+\ell)}{\widetilde}\nu(\ell ) < \infty.\ ] ] it only remains to prove .
if we define @xmath925 , then @xmath926 note that @xmath927 while , for all @xmath928 , @xmath929 if @xmath930 , then we have in particular that @xmath931 .
therefore @xmath932 from which we get for @xmath933 ( with the convention @xmath934 ) , @xmath935 because @xmath936 is stochastic . since @xmath937 , we get @xmath938 , for @xmath939 , so @xmath940 by fatou s lemma . by letting @xmath882 in
, we obtain @xmath941 for all @xmath942 . but
@xmath943 , which implies that @xmath944 , and there exists a constant @xmath711 such that @xmath945 .
necessarily , @xmath946 , so @xmath947 .
this completes the proof of , and hence the proof of the central limit theorem for the speed . in appendix [ appb ]
we list the key ingredients necessary to extend the above argument to prove the central limit theorem for the charge .
in this section we prove theorems [ thm : speed][thm : charge ] . if @xmath948 ( `` off the critical curve '' ) , then the following hold : * the rate function @xmath652 for the speed in has a unique zero at @xmath949 ( corresponding to @xmath631 ) , which equals . *
the rate function @xmath680 for the charge in has a unique zero at @xmath950 ( corresponding to @xmath630 ) , which equals .
hence the laws of large numbers follow from the large deviation principles .
( note : the condition @xmath951 in is put in to fix a direction for the speed : by the symmetry of the simple random walk the same large deviation principle holds to the left . )
if , on the other hand , @xmath952 ( `` on the critical curve '' ) , then the rate functions have a horizontal piece , and hence the laws of large numbers need a separate argument .
we first note that @xmath953 , as it appears from and lemma [ lem : slopecrit1 ] .
thanks to the strict convexity of @xmath954 at the right of @xmath955 , it suffices to prove that @xmath956 we will use the representation of @xmath957 , @xmath958 , as developed in the proof of theorem [ thm : cltspeedcharge ] . reproducing steps 16 with @xmath959 ,
we are left with controlling @xmath960 where @xmath961 is the stationary markov chain whose transition matrix is given by , for the choice of parameters @xmath962 and @xmath963 .
it will be shown in the proof of lemma [ lem : slopecrit1 ] below that @xmath964 , which means that the markov chain has finite mean . furthermore , @xmath965 . since @xmath966
, we only need that @xmath967 \geq { \varepsilon}n ) = 0,\ ] ] but this follows from the law of large numbers for stationary markov chains .
section [ ss : asympcritcurve ] contains the proof of the scaling of @xmath96 for @xmath968 stated in theorem [ thm : asymp.cc](2 ) .
section [ orderphtr ] explains why the phase transition is first order as claimed in theorem [ thm : orderphtr ] .
section [ ss : wil ] contains the proof of the scaling of @xmath96 for @xmath969 stated in theorem [ thm : asymp.cc](1 ) , and also deals with the weak interaction limit @xmath970 in theorem [ thm : fscalbeta ] . in this section
we give the proof of theorem [ thm : asymp.cc](2 ) .
we begin by stating a rough but helpful lemma .
recall : @xmath971 = \log { { \ensuremath{\mathbb e } } } \left[e^ { \delta \omega_\ell ( 1-\frac{\beta}{\delta } \omega_\ell)}\right].\ ] ] [ sbcond ] the following hold : + ( 1 ) if @xmath972 for all @xmath973 , then @xmath587 .
+ ( 2 ) if there exists @xmath974 such that @xmath975 , then @xmath976 .
\(1 ) this is immediate from and ( [ eq : identity*][eq : znzn*rel ] ) .
+ ( 2 ) let @xmath974 such that @xmath975 .
we restrict the partition function @xmath395 to the trajectory @xmath977 defined by @xmath978 and @xmath979 for @xmath980 , we can estimate @xmath981}=s_{[0,2a\ell]}).\ ] ] take @xmath982 on both sides of , let @xmath983 and recall , to obtain that @xmath984 . with the help of lemma [ sbcond ] the asymptotics of @xmath96 for @xmath968 is proved as follows .
_ lattice case _ : recall .
we prove that there exists a @xmath985 such that @xmath986 to prove the upper bound in , note that @xmath987 for all @xmath594 .
therefore , choosing @xmath988 in the right - hand side of , we obtain that @xmath989 for all @xmath973 and @xmath24-a.e .
consequently , @xmath990 for @xmath991 and @xmath594 which , by lemma [ sbcond](1 ) , implies that @xmath992 for @xmath991 . to prove the lower bound in , note that there exists an @xmath594 such that @xmath993 ( durrett ( * ? ? ?
* theorem 3.5.2 ) ) .
therefore , choosing @xmath994 in the right - hand side of , we obtain @xmath995\\ \nonumber & = \log { { \ensuremath{\mathbb p } } } ( \omega_\ell = t ) + \delta t\left(1-\tfrac{\beta(\delta , t)}{\delta } t\right ) = \log { { \ensuremath{\mathbb p } } } ( \omega_\ell = t)+c_1,\end{aligned}\ ] ] hence @xmath996 the right - hand side of is strictly positive for @xmath997 large enough , uniformly in @xmath53 .
therefore , by lemma [ sbcond](2 ) , we have @xmath998 , which completes the proof of . _ non - lattice case _ : we show that @xmath999 pick @xmath1000
. the proof of will be complete once we show that @xmath1001 for @xmath53 large enough . to that end
we note that for any @xmath1002 there exists an @xmath974 such that @xmath1003 ( by the non - lattice assumption ; recall with @xmath179 ) .
choosing @xmath1004 , we get @xmath1005 \\ & \geq \log { { \ensuremath{\mathbb p } } } \big(\omega_\ell \in ( 0,\tfrac{1}{c})\big ) + \delta \ , { { \ensuremath{\mathbb e } } } \big[\omega_\ell(1-c\omega_\ell ) \mid \omega_\ell \in ( 0,\tfrac{1}{c})\big ] . \end{split}\end{aligned}\ ] ] since @xmath1006 > 0 $ ] , the right - hand side tends to @xmath1007 as @xmath174 , and hence @xmath1008 for @xmath53 large enough .
this completes the proof of by lemma [ sbcond ] . _
non - lattice case with density _ : suppose that @xmath180 has a density @xmath1009 with respect to the lebesgue measure @xmath1010 .
then , for all @xmath594 , @xmath232 has a density @xmath1011 with respect to @xmath1010 .
[ lem : dc1 ] @xmath1012 . using
, we may write @xmath1013.\ ] ] suppose that @xmath1009 has a finite @xmath1014-norm
. then @xmath1011 is bounded and continuous for @xmath1015 , with @xmath1016 for @xmath231 large enough , by the local central limit theorem ( see feller ( * ? ? ? * theorem 2 in section xv.5 ) ) . by continuity , for every @xmath1017
there exists a @xmath1018 such that @xmath1019 for all @xmath1020 . if @xmath1021 , then we can write @xmath1022\\ & \geq \log[(1-{\varepsilon})g_\ell(0 ) ] + \log \big [ \int_{\mathbb{r}}e^{\frac{\delta^2}{\beta } s(1-s ) } \tfrac{\delta}{\beta } { { \sf 1}}_{\{s\in ( 0,1)\ } } { \text{\rm d}}s \big ]
. \end{aligned}\ ] ] choosing @xmath1023 , with @xmath1024 fixed and @xmath53 large enough , we get that the condition @xmath1025 is satisfied and @xmath1026 - \log \delta + \log \log \delta + \log[(1-{\varepsilon})g_\ell(0 ) ] - \log c.\ ] ] for every @xmath1017 , there exists a @xmath1027 such that @xmath1028 \geq \frac{1-{\varepsilon}}{4c } \log \delta + c_{\varepsilon}$ ] , and so @xmath1029 when @xmath1030 . since @xmath1017 is arbitrary , it follows from lemma [ sbcond ] that @xmath1031 for any @xmath1032 and @xmath53 large enough .
[ lem : dc2 ] @xmath1033 . here
we assume that the density @xmath1009 of @xmath180 is bounded , i.e. , @xmath1034 .
then @xmath1035 for all @xmath973 .
note that , for every @xmath900 , @xmath1036 & \le e^{-\delta { \varepsilon } } + { { \ensuremath{\mathbb p } } } ( \omega_\ell \in [ -{\varepsilon},0 ] \cup [ \tfrac{\delta}{\beta},\tfrac{\delta}{\beta}+{\varepsilon } ] ) \le e^{-\delta { \varepsilon } } + 2{\varepsilon}\|g_1\|_\infty , \end{split}\ ] ] because @xmath1037 when @xmath1038 or @xmath1039 , and @xmath1040 on the whole domain of integration @xmath1041 .
choosing @xmath1042 , we get @xmath1043 = 0\ ] ] uniformly in @xmath973 . on the other hand , since @xmath1044 , we can write @xmath1045 \le e^{\frac{\delta^2}{4\beta } } \ , { { \ensuremath{\mathbb p } } } ( \omega_\ell \in ( 0,\tfrac{\delta}{\beta } ) ) \le e^{\frac{\delta^2}{4\beta } } \ , \|g_1\|_\infty \ , \frac{\delta}{\beta}.\ ] ] we now choose @xmath1046 with @xmath1024 , to get @xmath1045 \le \|g_1\|_\infty \ , \delta^{\frac{1}{4c } } \ , \frac{\log \delta}{c \delta},\ ] ] and when @xmath1047 the right - hand side vanishes as @xmath968 uniformly in @xmath973 .
altogether , we have shown that @xmath1048 for any @xmath1047 and @xmath1049 , uniformly in @xmath973 .
it follows from lemma [ sbcond ] that @xmath1050 for any @xmath1051 and @xmath53 large enough . in this section
we give the proof of theorem [ thm : orderphtr ] .
it follows from that @xmath1052 since @xmath1053 by , and @xmath1054 , lemmas [ lem : slopecrit1][lem : slopecrit2 ] below imply in theorem [ thm : orderphtr ] .
[ lem : slopecrit1 ] for all @xmath182 , @xmath1055_{\mu=0 } \in ( 0,\infty).\ ] ] for @xmath1056 , @xmath1057_{\mu=0 } = ( i+j+1 ) a_{0,\delta,\beta}(i , j).\ ] ] in the following , @xmath53 is kept fixed and @xmath952 , which is the value of @xmath45 for which @xmath100 .
let @xmath1058 be the right eigenvector of @xmath1059 associated with the eigenvalue @xmath351 .
recall proposition [ pr : prop.a.lambda.mu ] .
since @xmath1059 is hilbert - schmidt and symmetric , @xmath1060 and @xmath1061 is a left eigenvector as well .
we choose @xmath1061 such that @xmath1062 .
our starting point is the relation @xmath1063_{\mu=0 } = \sum_{i , j \in { \mathbb{n}}_0 } \chi(i)\ , ( i+j+1 ) a_{0,\delta,\beta_c({\delta})}(i , j ) \ , \chi(j),\ ] ] which follows by and by @xmath1064 . from proposition [ pr : asymptgstar ] in appendix [ appa ] , we know that @xmath1065 for some constant @xmath481 . therefore , recalling and , we need to show that @xmath1066 this is done in 5 steps . in steps
14 we derive successively stronger tail estimates on @xmath1061 . in step 5
we use these to prove . in what follows
, @xmath481 is a constant that may change from line to line ( and depend on other choices of constants ) .
. * estimate @xmath1067^{1/2 } \left [ \sum_{i \geq k } \left ( \sum_{j\in{\mathbb{n}}_0 } \frac{1}{\sqrt{i+j}}\,{\ensuremath{\mathrm{p}}}(s_{i+j}= i - j)\ , \chi(j ) \right)^2\ , \right]^{1/2 } \end{aligned}\ ] ] and @xmath1068 combining , we get @xmath1069 abbreviating @xmath1070 , we find that , for any @xmath703 , @xmath1071\\ & \qquad = \sum_{i\in{\mathbb{n } } } [ i^{\alpha}- ( i-1)^{\alpha } ] \sigma_2(i ) \leq c \sum_{i\in{\mathbb{n } } } i^{{\alpha}-1 } \sigma_2(i ) \leq c \sum_{i\in{\mathbb{n } } } i^{\alpha - \tfrac32 } < \infty .
\end{aligned}\ ] ] * 2 . *
next we use to prove that @xmath1072 for all @xmath1073 .
indeed , let @xmath1074 and @xmath1075 , and use hlder s inequality to estimate @xmath1076 with @xmath1077 and @xmath1078 .
since @xmath1079 , @xmath1080 and @xmath1081 , we can use to get that both sums in the right - hand side are finite . *
* next we prove that @xmath1082 for all @xmath1083 .
indeed , since @xmath1084 we can use hlder s inequality with @xmath1085 and @xmath1086 to estimate @xmath1087 hence , using that the second sum is finite as shown in , we get @xmath1088 where in the next - to - last inequality we use that @xmath1089 * 4 . *
next we prove that @xmath1090 .
let @xmath1091 .
by hlder s inequality with @xmath1092 and @xmath1093 , @xmath1094 the first term converges .
abbreviating @xmath1095 , we find that @xmath1096 \sigma_{4-{\varepsilon}}(i ) \leq c \sum_{i\in{\mathbb{n } } } i \sigma_{4-{\varepsilon}}(i ) < c \sum_{i\in{\mathbb{n } } } i^{-\tfrac32+{\varepsilon } } < \infty,\ ] ] where we use the estimate in .
* we can now give the proof of .
the change of variables @xmath1097 and @xmath1098 yields @xmath1099 .
\end{aligned}\ ] ] by the cauchy - schwarz inequality and the symmetry of the simple random walk , the right - hand side is bounded from above by @xmath1100 = \sum_{u \in { \mathbb{n}}_0 } \sqrt{u+1 } \sum_{v\in { { \ensuremath{\mathbb z } } } } \chi\big(\frac{u+v}{2}\big)^2 { \ensuremath{\mathrm{p}}}(s_u = v).\ ] ] split the sum into two parts : @xmath1101 and @xmath1102 . the first part can be estimated by ( recall that @xmath1103 ) @xmath1104 the second part can be estimated by @xmath1105 where we use in the last equality .
[ lem : slopecrit2 ] for all @xmath182 , @xmath1106_{\beta = \beta_c(\delta ) } \in ( 0,\infty).\ ] ] since we fix @xmath1107 , we have @xmath1108 , hence @xmath1109_{\beta = \beta_c(\delta ) } = \sum_{i , j \in { \mathbb{n}}_0 }
\chi(i ) \chi(j ) \left [ \frac{\partial}{\partial \beta}g^*_{\delta,\beta}(i+j-1 ) \right ] a_{0,\beta_c(\delta),\delta}(i , j).\ ] ] note that , for @xmath1110 , @xmath1111 } { { { \ensuremath{\mathbb e } } } [ e^{\delta { \omega}_\ell -\beta\omega_\ell^2}]}.\ ] ] use lemma [ lem : der.beta.g ] below to conclude that the sum in the numerator is finite .
[ lem : der.beta.g ] @xmath1112 .
recall above .
an argument similar to gives @xmath1113 \geq c\ , \ell^{-1/2},\ ] ] and an argument similar to gives @xmath1114 \leq c\ , \big(\sum_{k\in{\mathbb{n } } } k^2e^{{\delta}k - { \beta}k^2}\big)\ , \ell^{-1/2},\ ] ] which ends the proof .
in this section we give the proof of theorems [ thm : asymp.cc](1 ) and [ thm : fscalbeta ] .
recall and .
the proof comes in 10 steps . *
1 . * to study limits of variational formulas , we make use of the notion of _ epi - convergence _ ( which was used in the derivation of scaling limits for weakly self - avoiding walks as well ; see van der hofstad and den hollander @xcite ) .
[ def : epi ] let @xmath1115 be a metrizable topological space , and let @xmath1116 be dense in @xmath1117 .
given @xmath1118 with @xmath202 and @xmath1119 , the family @xmath1120 is said to be epi - convergent to @xmath1121 on @xmath1122 , written @xmath1123 when the following properties hold : @xmath1124 the importance of the notion of epi - convergence is contained in the following proposition , for which we refer to attouch ( * ? ? ?
* theorem 1.10 and proposition 1.14 ) .
[ prop : epi ] suppose that * @xmath1125 on @xmath1122 .
* for all @xmath202 , @xmath1126 is continuous on @xmath1117 and has a unique maximiser @xmath1127 .
* there exists a @xmath1128 such that @xmath1129 is @xmath1130-relatively compact in @xmath1117 , @xmath1121 has a unique maximizer @xmath1131 , and there exists a sequence @xmath1132 in @xmath1133 such that @xmath1134 and @xmath1135 as @xmath1136 .
then , as @xmath1136 , @xmath1137 below we will apply proposition [ prop : epi ] with the following choices : @xmath1138 where @xmath1139 is the set defined in , @xmath1140 are the functions defined in below , @xmath1141 is the functional defined in below , @xmath1142 ( or @xmath1143 ) for theorem [ thm : asymp.cc](1 ) ( or theorem [ thm : fscalbeta ] ) , while @xmath711 is a constant chosen large enough so that @xmath1144 .
the constants @xmath167 and @xmath891 are chosen as we go along . * 2 . * from the rayleigh formula , we have @xmath1145 we begin by showing that the supremum can actually be taken over functions in @xmath1146 . indeed , for @xmath1147 write @xmath1148 let @xmath1149 with @xmath1150 . define the piecewise constant function @xmath1151 then @xmath1152 which , by jensen s inequality , is smaller than or equal to @xmath1153 .
next , denote the renormalised version of @xmath1154 by @xmath1155 .
then @xmath1156 therefore we may write @xmath1157 with @xmath1158 * 3 . * next , recall . using that , for all @xmath55 , @xmath1159 we decompose the variational formula in as @xmath1160 where @xmath1161,\\ i^2_{{\delta},{\beta}}(f ) & = \frac12 \int_0^\infty { \text{\rm d}}x \int_0^\infty { \text{\rm d}}y\ , [ f(x ) - f(y)]^2 \frac{1}{{\beta}^\eta } a_{\mu,\delta,\beta}\big(\left\lfloor\frac{x}{{\beta}^\eta}\right\rfloor , \left\lfloor\frac{y}{{\beta}^\eta}\right\lfloor\big ) .
\end{aligned}\ ] ] we are interested in the behaviour of the quantity in as @xmath1136 . define @xmath1162 ^ 2,\ ] ] with @xmath1163 when @xmath1164 does not exist everywhere .
our key observation is the following lemma .
parts ( 1 ) and ( 2 ) settle requirement in ( i ) in proposition [ prop : epi ] , part ( 3 ) settles requirement ( iii ) , while requirement ( ii ) follows from the fact that has a unique maximiser and hence so does .
[ lem : epiconvergence ] ( 1 ) pick @xmath1165 and put @xmath1166 , @xmath1167 , @xmath1142 . then @xmath1168 & { \stackrel{\mathrm{(epi)}}{= } } -u_3(f ) .
\end{aligned}\ ] ] ( 2 ) pick @xmath1169 and put @xmath1170 , @xmath1171 . then @xmath1172 & { \stackrel{\mathrm{(epi)}}{= } } - u_3(f ) , \end{aligned}\ ] ] where @xmath1173 $ ] . + ( 3 ) fix @xmath1174 and
let @xmath1175 .
let @xmath1176 be the unique maximizer of @xmath1177 in @xmath1117 defined in .
then there exist @xmath1178 , @xmath202 , such that @xmath1179 the same holds for @xmath1180 satisfying @xmath1181 , @xmath1182 .
we give a brief sketch , the details of which will be worked out in steps 79 below . using proposition [ pr : g.conv - to - brownian ] in appendix [ appa ] we get , for @xmath1136 , @xmath1183 \frac{1}{{\beta}^\eta } q\big(\left\lfloor\frac{x}{{\beta}^\eta}\right\rfloor + 1 , \left\lfloor\frac{y}{{\beta}^\eta}\right\rfloor\big)\\ & \qquad { \stackrel{\mathrm{(epi)}}{\sim } } \big[(\tfrac12{\delta}^2 - { \beta}- \mu ) { \beta}^{-\eta}(x+y ) - { \beta}^{1 - 2\eta}{\delta}^2 ( x+y)^2 \big ] \frac{1}{{\beta}^\eta } q\big(\left\lfloor\frac{x}{{\beta}^\eta}\right\rfloor + 1 , \left\lfloor\frac{y}{{\beta}^\eta}\right\rfloor\big ) .
\end{aligned}\ ] ] inserting into the first line of and using proposition [ lem : asymptp ] in appendix [ appa ] , we find the first lines of . inserting into the second line of and using proposition [ lem : asymptp ] in appendix [ appa ] , we find the second lines of . part ( 3 ) will be achieved by taking for @xmath1184 the linear interpolation of @xmath1185 . * 4 .
proof of theorem [ thm : asymp.cc](1 ) .
* we look at the scaling of @xmath1186 as @xmath1187 . put @xmath1188 then , proposition [ prop : epi ] and lemma [ lem : epiconvergence](1,3 ) imply that @xmath1189 = \chi(c - b,1),\ ] ] where @xmath1190 - \tfrac12 f'(x)^2 ( 2x ) \big ] \right\ } , \qquad a\in{{\ensuremath{\mathbb r } } } .\ ] ] with @xmath1139 the set defined in . via integration by parts , the variational formula in can be rewritten as @xmath1191 f(x),\ ] ] which is the variational representation of the largest eigenvalue @xmath1192 of the sturm - liouville operator @xmath1193 introduced in .
pick @xmath1194 in and use that @xmath1195 changes sign at @xmath1196 according to , to obtain that yields the scaling for the critical curve given in .
proof of theorem [ thm : fscalbeta](2 ) .
* this a consequence of and the shape of @xmath1197 ( recall fig . [ fig - chiab ] ) .
indeed , let @xmath45 be as in .
then @xmath1198(\tfrac12{\delta}^2)^{\tfrac43 } , \qquad { \delta},{\beta}\downarrow 0.\ ] ] the right - hand side is equivalent to @xmath1199 as soon as @xmath1200 . *
proof of theorem [ thm : fscalbeta](1 ) . * set @xmath1201 .
then , proposition [ prop : epi ] and lemma [ lem : epiconvergence](2,3 ) imply that @xmath1202 = \chi(b,\rho_{\delta}),\ ] ] from which we get @xmath1203 $ ] . because @xmath1204 , this proves the asymptotics for the free energy in the first part of . as to the asymptotics for the speed , recall , which reads @xmath1205^{-1}_{\mu = \mu(\delta,\beta)}\ ] ] because @xmath1206 .
we have just proven that @xmath1207 , \qquad \beta \downarrow 0.\ ] ] by convexity , we may take the derivative of with respect to @xmath410 , to get @xmath1208_{\mu = -f({\delta } ) - b{\beta}^{\tfrac23 } } = \left[\frac{\partial}{\partial b}\chi(b,\rho_{\delta})\right ] { \beta}^{\tfrac13 } [ 1+o(1)].\ ] ] using monotonicity of @xmath1209 and continuity of @xmath1210 , we in turn deduce that @xmath1211_{\mu = \mu({\delta},{\beta } ) } & = -\left[\frac{\partial}{\partial b}\chi(b,\rho_{\delta } ) \right]_{b = a^*(\rho_{\delta } ) } { \beta}^{-\tfrac13 } [ 1+o(1 ) ] , \end{aligned}\ ] ] which via proves the second part of . to get the asymptotics for the charge we note that @xmath1212 = \frac{\partial}{\partial{\delta } } f(\delta,\beta).\ ] ] we know that @xmath1213 as @xmath1136 .
hence we get the third part of , provided we show that the differentiation w.r.t .
@xmath53 and the limit @xmath1136 may be interchanged .
this can be justified as follows .
fix @xmath182 . for @xmath1214 ,
estimate @xmath1215\big| & \leq { \beta}^{-\tfrac23 } \big|\frac{\partial}{\partial{\delta}}f({\delta},{\beta } ) - \frac{f({\delta}_1,{\beta})- f({\delta}_2,{\beta})}{{\delta}_1 - { \delta}_2}\big|\\ & \qquad + \big| \frac{[{\beta}^{-\tfrac23}f({\delta}_1,{\beta } ) + a^*(\rho_{{\delta}_1 } ) ] - [ { \beta}^{-\tfrac23 } f({\delta}_2,{\beta } ) + a^*(\rho_{{\delta}_2})]}{{\delta}_1 - { \delta}_2}\big|\\ & \qquad + \big|\frac{a^*(\rho_{{\delta}_2})-a^*(\rho_{{\delta}_1})}{{\delta}_1 - { \delta}_2 } - \frac{{\text{\rm d}}}{{\text{\rm d}}{\delta}}[-a^*(\rho_{\delta})]\big| . \end{aligned}\ ] ] the second term in the right - hand side tends to zero as @xmath1136 for every choice of @xmath1216 .
the third term does not depend on @xmath45 , and tends to zero as @xmath1217 . to control the first term it is enough to prove that for some @xmath1218 , @xmath1219 a sufficient condition for the latter is that there exist @xmath1220 such that @xmath1221 finally , from , @xmath1222 .
* 7 . * in the remaining steps we prove lemma [ lem : epiconvergence ] . along the way we need two technical lemmas .
[ lem : lim_moment ] as @xmath1223 , @xmath1224 = c_m\ \ell^m [ 1+o(1 ) ] \mbox { with } c_m= 1 \times 3 \times \ldots \times ( 2m-1),\\[0.2 cm ] & \forall m\in{\mathbb{n}}\colon & { { \ensuremath{\mathbb e } } } [ { \omega}_\ell^{2m+1 } ] \leq c'_m\ \ell^m \mbox { for some constant $ c'_m$ depending on $ m$. } \end{array}\ ] ] the computations are straightforward and are left to the reader .
[ lem : momentsqij ] for every @xmath1225 there exists a constant @xmath1226 such that @xmath1227 note that @xmath1228 , and use to show that @xmath1229 for some constant @xmath481 .
* lemmas [ lem : epi.one.sup][lem : epi.two.inf ] below prove the epi - convergence claimed in lemma [ lem : epiconvergence](1,2 ) , which is requirement ( i ) in proposition [ prop : epi ] .
[ lem : epi.one.sup ] let @xmath1230 and @xmath1166 .
then , for all @xmath1231 in @xmath1139 , @xmath1232 for simplicity , we start with the case @xmath1194 . for @xmath974 ,
let @xmath1233 .
note that @xmath1234 , \qquad { \beta}\downarrow 0.\ ] ] we expand @xmath1235 to fifth order in @xmath261 , so that the expansion includes the limiting term .
there exists a constant @xmath711 such that @xmath1236 keep in mind that @xmath231 will later be replaced by @xmath1237 , where @xmath1238 .
reordering the terms of the expansion according to @xmath261 , we get @xmath1239 \leq \big ( \tfrac12 { \delta}^2 - { \beta}\big ) { { \ensuremath{\mathbb e } } } [ { \omega}_\ell^2 ] + \big ( \tfrac12 { \beta}^2 - \tfrac12 { \delta}^2 { \beta}+ \tfrac{1}{24 } { \delta}^4 \big ) { { \ensuremath{\mathbb e } } } [ { \omega}_\ell^4 ] + r_{{\delta},{\beta}}(\ell),\ ] ] where @xmath1240 is a remainder term given by @xmath1241 + \big(\tfrac12 { \delta}{\beta}^2 - \tfrac16 { \delta}^3 { \beta}\big){{\ensuremath{\mathbb e } } } [ { \omega}_\ell^5 ] + \tfrac14 { \delta}^2 { \beta}^2 { { \ensuremath{\mathbb e } } } [ { \omega}_\ell^6 ] - \tfrac16 { \delta}{\beta}^3 { { \ensuremath{\mathbb e } } } [ { \omega}_\ell^7]\\ + \tfrac{1}{24 } { \beta}^4 { { \ensuremath{\mathbb e } } } [ { \omega}_\ell^8 ] + { \delta}^5 { { \ensuremath{\mathbb e } } } [ { \omega}_\ell^5 ] + 10 { \delta}^3 { \beta}^2 { { \ensuremath{\mathbb e } } } [ { \omega}_\ell^7 ] + 5 { \delta}{\beta}^4 { { \ensuremath{\mathbb e } } } [ { \omega}_\ell^9 ] .
\end{split}\ ] ] recall the first line of , and set @xmath1142 . estimate @xmath1242 , where @xmath1243 + \big(\tfrac12 { \beta}^2-\tfrac12 { \delta}^2 { \beta}+ \tfrac{1}{24}{\delta}^4\big ) { { \ensuremath{\mathbb e } } } \left[{\omega}_{\left\lfloor x{\beta}^{-2/3}\right\rfloor + \left\lfloor y{\beta}^{-2/3}\right\rfloor } ^4 \right ] \right\ } \end{split}\ ] ] and @xmath1244 let us first deal with @xmath1245 .
we cut the integrals over @xmath55 and @xmath314 at @xmath195 . using lemmas [ lem : lim_moment][lem : momentsqij ]
we get that , for all @xmath195 small enough , there exists a constant @xmath711 such that @xmath1246\,{\beta}^{-\tfrac23 } \int_{\varepsilon}^\infty { \text{\rm d}}x \int_{\varepsilon}^\infty { \text{\rm d}}y\ , f_{\beta}^2(x)\ , q\left(\left\lfloor x{\beta}^{-2/3}\right\rfloor + 1 , \left\lfloor y{\beta}^{-2/3}\right\rfloor\right)\\ & \qquad \qquad \qquad \times [ c(x+y ) - ( x+y)^2]\,{\beta}^{\frac23}. \end{aligned}\ ] ] therefore , @xmath1247 . \end{aligned}\ ] ] recall . making the change of variables @xmath1248 and @xmath1249 , we obtain @xmath1250\\ & \leq \tfrac12 { \beta}^{-\tfrac23 } \int_0^\infty { \text{\rm d}}u \int_{-\infty}^\infty { \text{\rm d}}v\ , f_{\beta}^2\,\big(\frac{u+v}{2}\big)\,(cu - u^2)\,{\ensuremath{\mathrm{p}}}\big(s_{\lfloor u{\beta}^{-2/3}\rfloor } = \lfloor v{\beta}^{-\tfrac23}\rfloor\big)\\ & = \int_0^\infty { \text{\rm d}}\left(\frac{u}{2}\right)\ , ( cu - u^2)\ , { \ensuremath{\mathrm{e}}}\big [ f^2_{\beta}\big ( \frac{u + { \beta}^{2/3}s_{\lfloor u{\beta}^{-2/3}\rfloor}}{2 } \big ) \big ] \\ & \stackrel{{\beta}\downarrow 0}{\longrightarrow } \int_0^\infty { \text{\rm d}}\left(\frac{u}{2}\right ) \,(cu - u^2)\,f\left(\frac{u}{2}\right)^2 . \end{aligned}\ ] ] we next deal with the remainder term @xmath1251 and show it is @xmath1252 .
for brevity we deal with the first term of @xmath1240 only , and leave the reader to check that the other terms in @xmath1240 can be handled in the same way . using lemmas [ lem : lim_moment][lem : momentsqij ] ,
we get @xmath1253\\ & \leq c { \beta}^{3/2 } \int_0^\infty { \text{\rm d}}x\ , f_{\beta}^2(x ) \int_0^\infty { \text{\rm d}}({\beta}^{-2/3}y)\ , q\left(\left\lfloor x { \beta}^{-2/3}\right\rfloor + 1,\left\lfloor y{\beta}^{-2/3}\right\rfloor\right)\ , \big[(x+y){\beta}^{-2/3}\big]\\ & \leq c { \beta}^{3/2 } \int_0^\infty { \text{\rm d}}x\ , f_{\beta}^2(x)\,\big[x{\beta}^{-2/3}\big ] \stackrel{{\beta}\downarrow 0}{\sim } c { \beta}^{5/6 } \int_0^\infty { \text{\rm d}}x\ , xf^2(x ) . \end{aligned}\ ] ] we next indicate how to deal with the case @xmath1254 .
the left - hand side of has to be replaced by @xmath1255 .
since @xmath1256 is of order @xmath1257 , the first term in the right - hand side of becomes @xmath1258 ( recall that @xmath1259 ) .
moreover , @xmath1260 is equivalent to @xmath1261 , so we may repeat the computations for the case @xmath1194 after replacing @xmath481 by @xmath1262 .
[ lem : epi.one.inf ] let @xmath1230 and @xmath1166 .
then , for all @xmath1263 , @xmath1264 instead of , use @xmath1265 the analysis for small @xmath45 in lemma [ lem : epi.one.sup ] essentially carries over .
[ lem : epi.two.sup ] let @xmath1266 and @xmath1166 .
then , for all @xmath1267 , @xmath1268 we need a lower bound for @xmath1269 ^ 2\\ & \qquad\qquad \times { \beta}^{-2/3}\ , e^{g^*_{{\delta},{\beta}}(\lfloor(x+y){\beta}^{-2/3}\rfloor)-\mu(\lfloor(x+y){\beta}^{-2/3}\rfloor)}\ , q\left(\left\lfloor x{\beta}^{-2/3}\right\rfloor+1,\left\lfloor y{\beta}^{-2/3}\right\rfloor\right ) .
\end{aligned}\ ] ] fix @xmath1270 and observe that @xmath1271 fix @xmath1272 . then , restricting the integral over @xmath55 to the interval @xmath1273 and the integral over @xmath314 to @xmath1274 , we obtain for @xmath863 and @xmath261 small enough @xmath1275 ^ 2\ , q\left(\left\lfloor x{\beta}^{-2/3}\right\rfloor + 1 , \left\lfloor y{\beta}^{-2/3}\right\rfloor\right ) . \end{aligned}\ ] ] for fixed @xmath1276 , the exponential term goes to @xmath351 and the infimum tends to @xmath1277 as @xmath1278 . put @xmath1279 and use , so that the integral becomes @xmath1280 \int_{\varepsilon}^n { \text{\rm d}}x\ , \int_{-n}^n\ , { \text{\rm d}}{\omega}\ , { \beta}^{-2/3}\ , [ f_{\beta}(x)-f_{\beta}(x+{\beta}^{1/3}{\omega})]^2\ , \frac{1}{\sqrt{2\pi ( 2x)}}e^{-\frac{{\omega}^2}{2(2x)}}.\ ] ] we are now in the same situation as in the proof of ( * ? ? ?
* lemma 7 , eq.(2.14 ) ) .
we refer to ( * ? ? ?
( 2.17)(2.26 ) ) to show that the limit of this integral as @xmath1281 is the integral with @xmath1282 ^ 2 $ ] in place of @xmath1283 ^ 2 $ ] in the integrand . letting @xmath658 and @xmath1284 , we get the desired result .
[ lem : epi.two.inf ] let @xmath1285 and @xmath1166 .
then , for all @xmath1263 , @xmath1286 since @xmath1287 , a first upper bound on @xmath1288 is @xmath1289 ^ 2\\ & \qquad \qquad \times e^{g^*_{{\delta},{\beta}}((x+y){\beta}^{-2/3})}\ , q\left(\left\lfloor x{\beta}^{-2/3}\right\rfloor + 1,\left\lfloor y{\beta}^{-2/3}\right\rfloor\right ) .
\end{aligned}\ ] ] recall that @xmath1290 \leq e^{{\delta}^2/4{\beta } } = \sqrt{e}[1+o(1)],\quad \mbox{as } { \beta}\downarrow 0,\ ] ] and that the maximum of @xmath1291 is achieved at @xmath1292 , as @xmath1293 . as in the proof of ( * ? ? ?
* lemma 8) , we split the integral into three parts ( note that @xmath261 there is @xmath1294 here ) .
fix @xmath1295 .
part 1 corresponds to @xmath1296 or @xmath1297 .
we may use and ( * ? ? ?
( 2.28)(2.29 ) ) to show that this part is negligible .
in part 2 we integrate over @xmath1298 and @xmath1299 . again , and ( * ? ? ? * eq .
( 2.32 ) ) are enough to conclude . finally ,
in part 3 we integrate over @xmath1298 and @xmath1300 .
we only need to prove that the factor @xmath1301 is harmless .
indeed , let @xmath1302 .
abbreviating @xmath1303 , we get @xmath1304 \leq e^{-\kappa^2 + \kappa\sqrt{2}},\\ & { { \ensuremath{\mathbb e } } } \big[e^{-{\beta}{\omega}_\ell^2 + { \delta}{\omega}_\ell } { { \sf 1}}_{\left\{{\omega}_\ell > \kappa/\sqrt{{\beta}}\right\ } } \big ] \leq [ \sqrt{e } + o(1)]\ , { { \ensuremath{\mathbb p } } } \big({\omega}_\ell > ( \kappa/\sqrt{2c } ) \sqrt{\ell } \big ) .
\end{aligned}\ ] ] therefore we get the result by first letting @xmath1281 ( see ( * ? ? ?
* eq . ( 2.34)(2.43 ) ) ) , then @xmath1305 , and finally @xmath1306 .
the analogues of lemmas [ lem : epi.one.sup][lem : epi.two.inf ] for @xmath1281 and @xmath607 fixed are proved in the same manner as in steps 8 and 9 .
the details are left to the reader .
thus we have completed the proof of lemmas [ lem : epiconvergence](1,2 ) .
the proof of lemma [ lem : epiconvergence](3 ) , which is requirement ( iii ) in proposition [ prop : epi ] , is given in appendix [ appc ] .
we prove three properties of the function @xmath60 defined in that were used in sections [ subsec : eigenvalue ] , [ ss : critcurve ] and [ ss : wil ] .
[ pr : asymptgstar ] for @xmath1307 , @xmath1308 recall that @xmath1309 . with the help of the local limit theorem we can estimate @xmath1310\}}\right ] \\ &
\geq c(\delta,\beta ) + \log { { \ensuremath{\mathbb p } } } \big(\omega_\ell \in [ 0,1]\big ) \sim -\tfrac12 \log \ell , \qquad \ell\to\infty , \end{aligned}\ ] ] and @xmath1311\big)\right ] \leq \log(c \ell^{-1/2}),\ ] ] from which the claim follows .
[ pr : g.conv - to - brownian ] let @xmath1312 .
for @xmath1313 and @xmath1314 , @xmath1315 abbreviate @xmath1316 , @xmath1317 and @xmath1318 . note that @xmath1319 , while @xmath1320 = o(\ell^{-1/2 } ) \to 0 $ ] and @xmath1321\to 3 $ ] . by expanding the exponential below to fourth order and keeping only the terms that are of order lower than or equal to @xmath1257 , we obtain @xmath1322 which is the desired result .
[ pr : gbetadelta.monotonicity ] for @xmath973 , @xmath1323 is strictly decreasing on @xmath74 and @xmath1324 is strictly convex on @xmath74 . for @xmath202 ,
compute @xmath1325 } { { { \ensuremath{\mathbb e } } } \big [ e^{\delta \omega_\ell-\beta \omega_\ell^2 } \big]},\ ] ] which is strictly negative because @xmath1326 is non - degenerate . for @xmath182 , compute @xmath1327 } { { { \ensuremath{\mathbb e } } } [ e^{\delta \omega_\ell-\beta \omega_\ell^2}]}.\ ] ] therefore @xmath1324 is stricly convex on @xmath81 .
( van der hofstad and den hollander ( * ? ? ?
* lemma 3 ) ) [ lem : asymptp ] @xmath1328 for @xmath1329 with @xmath1330 .
in section [ cltproof ] we gave the proof of the central limit theorem for the speed . in this appendix
we list the key ingredients necessary to extend the argument to the charge .
this comes in 3 steps . *
write , for @xmath1331 ( recall ) , @xmath1332/\sqrt{n } } \big ] = e^{-{\lambda}\rho({\delta},{\beta})\sqrt{n}}\,\frac{{{\ensuremath{\mathbb z } } } _ n^{*,{\delta}+ ( { \lambda}/\sqrt{n}),\beta}}{{{\ensuremath{\mathbb z } } } _ n^{*,{\delta},\beta}}\\ & \qquad = e^{-{\lambda}\rho({\delta},{\beta})\sqrt{n}+\big[\mu({\delta}+({\lambda}/\sqrt{n}),{\beta } ) - \mu({\delta},{\beta})\big]n } \ , \frac{e^{-\mu({\delta}+({\lambda}/\sqrt{n}),{\beta})n } \,{{\ensuremath{\mathbb z } } } _ n^{*,{\delta}+({\lambda}/\sqrt{n}),\beta}}{e^{-\mu({\delta},{\beta})n}{{\ensuremath{\mathbb z } } } _ n^{*,{\delta},\beta}}. \end{aligned}\ ] ] by and the analyticity of @xmath1333 , as @xmath882 the first factor converges to @xmath1334,\ ] ] which is the desired limit , see .
therefore we need to prove that the ratio in the second factor converges to @xmath351 .
an adaptation of lemma [ lem : clt3 ] gives @xmath1335 recalling points - on page , it is enough to show that the probabilities @xmath1336 both converge to @xmath880 as @xmath882 ( recall ) .
the difficulty is that in the second probability one of the parameters depends on @xmath49 . to handle this , we need the following uniform renewal theorem . [
lem : appr.ren.thm ] suppose that @xmath1337 is a sequence of inter - arrival time distributions , each with finite mean , converging to an inter - arrival time distribution @xmath1338 , also with finite mean .
suppose further that there exists a constant @xmath1339 such that @xmath1340 then , @xmath1341 denotes the sequence of arrival times . by and dominated convergence
, we have @xmath1342 = e[\tau_1]$ ] . according to ney @xcite
, there exists a constant @xmath1343 such that @xmath1344^{-1}| \leq c_0\ , p^{(n)}\left ( \sum_{i=1}^n z_i > n \right),\ ] ] where the @xmath1345 s are i.i.d . with @xmath1346 , and @xmath1347 for all @xmath1348 with @xmath1349 .
moreover , condition ensures that the constants @xmath1350 are uniform in @xmath42 , and that @xmath239 is bounded away from @xmath41 uniformly in @xmath42 .
* 2 . * abbreviate @xmath1351 . apply lemma [ lem : appr.ren.thm ] to the renewal process whose inter - arrival times have law @xmath1352 delayed by a first inter - arrival time with law @xmath1353 we can now explain why holds . for simplicity , we restrict to the case without delay , i.e. , @xmath1354 .
let @xmath1355 be such that @xmath1356 , and put @xmath1357 .
define @xmath1358 .
we rely on the following lemma , whose proof is given in step 3 below . [
lem : rfss ] there exists a @xmath1295 ( depending on @xmath1129 ) such that @xmath1359 this implies that there exists a @xmath1360 such that @xmath1361 define , recursively , @xmath1362 and @xmath1363 , @xmath10 , as well as the random processes @xmath1364 note that the pair @xmath1365 constitutes a markov renewal process , and that @xmath1366 we can now derive an exponential upper bound on @xmath1367 . indeed , write @xmath1368 split the sum according to @xmath1369 and @xmath1370 , with @xmath1371 a constant to be determined later .
we have @xmath1372 where @xmath1373 . using lemma [ lem : rfss ] , we know that @xmath1374 \big),\ ] ] which is finite for @xmath1375 small enough . therefore ,
choosing @xmath741 small enough , we find a @xmath1295 such that @xmath1376 * 3 .
* we conclude by giving the proof of lemma [ lem : rfss ] . for the moment , ignore the dependence on @xmath49 and write @xmath19 , @xmath20 instead of @xmath1377 , @xmath1378 ( see the comment at the end of the proof ) . as we will see
, the difficulty lies in the lack of uniform exponential decay for the one - step transition probability of @xmath1379 w.r.t .
the initial state .
instead , we will analyse the exponential decay of @xmath1380 as @xmath1381 , and prove that the maximum is attained at @xmath1382 ( step i below ) . consequently , even when the markov chain starts at a large initial state , it _ quickly
_ returns to a predetermined finite subset of its state space ( step ii below ) .
* step i. * by and ( recall also ) @xmath1383 where @xmath1384 was defined in , we denote by @xmath1385 the simple symmetric random walk on @xmath908 , and we suppress the dependence on @xmath239 and @xmath261 on @xmath422 and @xmath806 .
we first need to control the exponential decay of @xmath806 .
this is the content of the following lemma .
[ lem : nurate ] the limit @xmath1386 exists and is the positive solution of @xmath1387 .
the same holds when @xmath806 is replaced by @xmath822 .
let @xmath1388 let @xmath195 .
since @xmath422 is bounded , by proposition [ pr : asymptgstar ] , by we get , for a constant @xmath1389 , @xmath1390 therefore , using the change of variables @xmath1391 and @xmath1392 , we obtain @xmath1393,\\ & \leq c_{\varepsilon}\sum_{k\in{\mathbb{n } } } \exp\big\{k\big[\frac{x}{2 } - \mu - \frac{r-{\varepsilon}}{2 } + \log \cosh \big(\frac{x+r-{\varepsilon}}{2}\big)\big]\big\ } , \end{aligned}\ ] ] from which we deduce , by evaluating at @xmath1394 , that @xmath1395 .
letting @xmath658 , we get @xmath1396 let us next prove that @xmath1397 is a limit .
define @xmath1398 by a standard large deviations estimate , @xmath1399 where @xmath1400 from we get ( henceforth we assume that @xmath1401 , etc . for notational convenience ) @xmath1402 where @xmath711 is a constant .
therefore , using proposition [ pr : asymptgstar ] , we get @xmath1403 setting @xmath1404 we obtain , in particular , that @xmath1405 .
we can rewrite the previous relation as @xmath1406 let us recall that the fenchel - legendre transform of the rate function @xmath1407 is the log - moment generating function of the the simple random walk : @xmath1408 . by direct computation ,
the @xmath1409 in is uniquely attained at @xmath1410 , so that @xmath1411 and in order to have @xmath1412 we must have @xmath1413 . since by definition @xmath1414 , and hence @xmath1415 , it follows from that @xmath1416 .
only a few minor modifications lead to the same result for @xmath822 .
note that @xmath1417 @xmath1418 is defined in with @xmath1419 .
setting @xmath1420 , we know that @xmath1421 , while @xmath1422 for @xmath1423
. henceforth we fix @xmath900 so that @xmath1424 .
again by , we have @xmath1425 , hence @xmath1426 .
we next choose @xmath1002 small , so that @xmath1427 let us reinforce .
we fix @xmath1428 such that @xmath1429 for @xmath1430 .
since @xmath422 is bounded , and we can replace @xmath1431 by @xmath1432 in by super - additivity , we get the following upper bound from , for @xmath1433 : @xmath1434 where @xmath481 is an absolute constant . in particular , recalling and setting @xmath1435
, we get @xmath1436 in the next step we complete the proof of lemma [ lem : rfss ] , choosing @xmath1437 .
* step ii . *
define @xmath1438 for some @xmath1439 to be determined later .
define the filtration @xmath1440 , @xmath1441 .
then @xmath1442 so the process defined by @xmath1443 is a martingale with respect to @xmath1444 .
( even though we do not know yet if @xmath1445 is integrable , we know by that @xmath1446 is well - defined with integral @xmath351 . )
since @xmath1447 is a stopping time , we get @xmath1448 .
we next provide an upper bound on @xmath1449 for @xmath1450 .
let @xmath195 be such that @xmath1451 .
we know from step i that typically @xmath1452 is at most @xmath1453 .
we split accordingly : @xmath1454 recalling , for @xmath1455 , @xmath1456 which is finite for @xmath1457 . combining with , we have ( possibly enlarging @xmath1129 ) @xmath1458 in the sequel , we redefine @xmath1459 as @xmath1460 , which we may safely assume to be @xmath1461 .
we get , since @xmath1462 a.s . , @xmath1463 \geq { \ensuremath{\mathrm{e}}}(e^{r_{\varepsilon}(1-a_0^+ ) \sum_{i=1}^{n \wedge \sigma } m^+_i - r_{\varepsilon}a_0^+ k}).\ ] ] note that @xmath1464 is a.s .
finite , because the markov chain @xmath1465 is positive recurrent .
therefore , by fatou s lemma , @xmath1466 which is the desired result .
recall that @xmath19 is originally @xmath1467 . to deal with uniformity in @xmath49 , it is enough to note that @xmath1468 is continuous , so the limit @xmath1397 in lemma [ lem : nurate ] and the function @xmath1469 in are continuous in @xmath239 , and hence in @xmath49 .
the details are left to the reader .
in this appendix we prove lemma [ lem : epiconvergence](3 ) .
the proof strategy is that of ( * ? ? ?
* proposition 4 ) . for the sake of conciseness
, we only write the proof in the regime @xmath1470 the regime @xmath1136 with @xmath239 fixed follows the same line of argument and is left to the reader . in what follows
we use @xmath1471 and @xmath1472 as shorthand notation for @xmath556 and @xmath1473 with @xmath1474 as in .
we choose for @xmath1475 the linear and renormalised interpolation of the solution of the discrete variational problem .
namely , if @xmath1476 is the normed eigenvector of @xmath1471 associated with @xmath1472 , then we define @xmath1477 \big\ } , \quad ( i-1){\beta}^{2/3 } < u \leq i{\beta}^{2/3 } , \quad i\in{{\ensuremath{\mathbb n } } } , \ ] ] and we pick @xmath1478 as an approximate maximiser . mimicking ( * ? ? ?
* lemmas 910 ) , we see that it is enough to prove the following adaptation of ( * ? ? ?
* lemma 11 ) : for @xmath261 small enough , @xmath1479 where @xmath1480 , @xmath10 , and the @xmath1481 s are constants whose precise values are irrelevant .
the estimates in are proved in two steps : gaussian disorder ( section [ gaussdisorder ] below ) and general disorder ( section [ generaldisorder ] below ) . the first allows for explicit formulas , the second makes use of gaussian approximations and taylor expansions . to avoid a lengthy proof
, we only indicate the necessary modifications to the proof in @xcite .
recall and . in
what follows we use the short - hand notation @xmath1482 with @xmath1483 * 1 . * recall .
we have @xmath1484 .
\end{split}\ ] ] recall . a taylor expansion as @xmath1485 ( uniformly over @xmath45 , as long as @xmath1486 )
yields @xmath1487 and hence @xmath1488
henceforth we fix @xmath900 such that @xmath1489 for all @xmath1490 . in this way we get @xmath1491 also note that @xmath1492 as @xmath1136 , and hence @xmath1493 for @xmath1494 small enough .
we show in the proof of lemma [ th : sup ] below that @xmath1495 attains its global maximum at @xmath1496 , and that @xmath1497 is decreasing for @xmath1498 . since @xmath1499 for @xmath1494 small ,
we have shown that @xmath1500 note that is the analogue of ( * ? ? ?
* eq . ( 2.5 ) ( i ) ) in our context .
the analogue of ( * ? ? ? * eq .
( 2.5 ) ( ii ) ) is given by the next lemma , where we estimate the global maximum of @xmath1501 .
[ th : sup ] @xmath1502 note that @xmath1503 setting @xmath1504 , we have @xmath1505 if and only if @xmath1506 , whose positive solution ( if any ) is @xmath1507 recalling , we have @xmath1508 , and hence @xmath1509 since @xmath1510 , we get @xmath1511 recalling , we get @xmath1512 which is the claim . before continuing with the main line of the proof , we provide two estimates that are the analogue of ( * ? ? ?
* lemma 13 ) ( see - below ) .
note that @xmath1513 by lemma [ th : sup ] . arguing as in ( * ? ? ?
* eq . ( 4.16 ) ) , we get @xmath1514 because @xmath1515 and @xmath1516 .
consequently , @xmath1517 for an analogous lower bound , arguing as in ( * ? ? ?
* eq . ( 4.17)-(4.20 ) ) , we get @xmath1518 recall that we want to prove ( i ) in . we start by proving the analogue of ( * ? ? ?
* steps 1 and 2 , pages 419 - 420 ) . [ th:12 ] the following relations hold ( where @xmath900 is fixed so that is in force ) : @xmath1519 in analogy with ( * ? ? ?
* eq . ( 3.15 ) ) , from the trivial inequality @xmath1520 ^ 2 a(i , j ) \geq 0\ ] ] we get @xmath1521 p(i , j ) \geq 0.\ ] ] note that there exists a @xmath1522 such that @xmath1523 for all @xmath1524 $ ] . by lemma
[ th : sup ] , @xmath1525 provided @xmath45 is small enough .
therefore @xmath1526 p(i , j ) \\ & \le 1-\lambda(\beta ) + \sum_{i\in{\mathbb{n}}_0 } \tau_\beta(i)^2 \sum_{j\in{\mathbb{n}}_0 } h(i+j+1 ) p(i , j ) + o(\beta^{4/3 } ) , \end{split}\ ] ] where the second inequality follows again by lemma [ th : sup ] and the fact that @xmath1527 . by , we can write @xmath1528 on the other hand , for @xmath1529 we can bound @xmath1530 because for @xmath1531 we have @xmath1532 , and consequently @xmath1533 again by .
having thus restricted the range of @xmath358 , we have @xmath1534 and we can apply : @xmath1535 with @xmath1536 .
the sums have been evaluated in ( * ? ? ?
* eq . ( 1.17 ) ) when @xmath358 runs over all of @xmath569 . here
we have the restriction @xmath1537 , which is harmless .
in fact , since @xmath1529 , the range of summation for @xmath358 includes @xmath1538 , and we know that the mass of @xmath1539 is concentrated around the diagonal , and decays exponentially in @xmath1540 when @xmath1541 for any @xmath1542 .
consequently , for some @xmath1543 we have @xmath1544 substituting and into , we get @xmath1545 let us abbreviate @xmath1546 and @xmath1547 .
note that @xmath1548 by . by cauchy - schwarz , @xmath1549 , since @xmath1550 .
hence yields @xmath1551 setting @xmath1552 and dividing by @xmath1553 , we obtain @xmath1554 since @xmath1555 , we can drop it from the left - hand side . setting @xmath1556 ,
se see that the inequality becomes @xmath1557 , which can only hold when @xmath55 is bounded from above , say @xmath1558 .
we have thus shown that @xmath1559 , i.e. , the first relation in .
next , we can drop @xmath1560 from the left - hand side of , getting @xmath1561 , which shows that also @xmath1562 is bounded from above , say @xmath1563 .
this _ almost _ proves the second relation in , in the sense that we have proved it with @xmath1564 instead of @xmath1565 .
however , recalling , we see that , and still hold when @xmath1564 is replaced by @xmath1565 .
consequently , writing with @xmath1564 replaced by @xmath1565 , we complete the proof of .
we can now complete the proof of ( i ) . arguing as in ( * ? ? ?
( 3.24 ) ) , by , for every @xmath1542 ( which will be fixed later ) there is @xmath1566 such that @xmath1567 where the last inequality holds by cauchy - schwarz . indeed ,
@xmath1568 and hence @xmath1569 and @xmath1570 .
since @xmath1571 as @xmath1572 , we get @xmath1573 \\ & \le \frac{(1+\delta)^2}{(1-\delta)^2 } e^{-\frac{{\varepsilon}^2}{5 } } \left\ { \sum_{j > ( 1-\delta ) \frac{{\varepsilon}}{4\beta } } j^2 \ , \tau_\beta(j)^2 \sum_{i < \frac{j}{1-\delta } } p(i , j ) \right\ } [ 1 + o(1)]\\ & \le \frac{(1+\delta)^2}{(1-\delta)^2 } e^{-\frac{{\varepsilon}^2}{5 } } \left\ { \sum_{j > ( 1-\delta ) \frac{{\varepsilon}}{4\beta } } j^2 \ , \tau_\beta(j)^2 \right\ } [ 1 + o(1 ) ] , \end{split}\ ] ] because @xmath1574 . consequently , @xmath1575\\ & \le \frac{(1+\delta)^2}{(1-\delta)^2 } e^{-\frac{{\varepsilon}^2}{5 } } \left\ { \sum_{i > \frac{{\varepsilon}}{4\beta } } i^2 \ , \tau_\beta(i)^2 + \frac{{\varepsilon}^2}{16\beta^2 } \sum_{i > ( 1-\delta)\frac{{\varepsilon}}{4\beta } } \tau_\beta(i)^2 \right\ } [ 1 + o(1)]\\ & \le \frac{(1+\delta)^2}{(1-\delta)^2 } e^{-\frac{{\varepsilon}^2}{5 } } \left\ { \sum_{i > \frac{{\varepsilon}}{4\beta } } i^2 \ , \tau_\beta(i)^2 + \frac{{\varepsilon}^2}{16\beta^2 } c_7 \ , \beta^{2/3 } \right\ } [ 1 + o(1 ) ] , \end{split}\ ] ] where in the last inequality we use the second relation in , provided we choose @xmath1576 so that @xmath1577 .
fixing @xmath1542 small enough so that @xmath1578 , we get @xmath1579 recalling the first relation in , we have completed the proof of ( i ) in .
the proof works similarly as in @xcite .
the only estimate we need is an upper bound on @xmath1580 ( see ( * ? ? ? * eq .
( 3.32 ) ) ) . recall .
explicit calculation gives @xmath1581 using that @xmath1582 and , we get , for some constant @xmath711 , @xmath1583- b { \beta}^{4/3}\\ & \geq c{\beta}^{4/3 } - 2{\beta}^2\ell .
\end{aligned}\ ] ] inserting this estimate into the analogue of ( * ? ? ?
* eq . ( 3.32 ) ) , we get @xmath1584 for some constant @xmath1585 .
an application of cauchy - schwarz and ( i ) in gives the result .
the proof is the same as in @xcite .
again , retracing the proof in @xcite , we see that an upper bound on @xmath1586 is needed .
using that @xmath1587 for @xmath55 small enough , we get @xmath1588 which is less than a constant times @xmath1589 , by . here , the function @xmath1495 defined in is denoted by @xmath1590 when @xmath1591 .
we use the proofs of ( i)-(iv ) for gaussian disorder as a reference frame and focus on the necessary modifications only .
below @xmath711 denotes a positive and finite constant that may change from line to line .
the first step is to prove for general disorder via a taylor expansion .
since @xmath1592 is bounded from above , we may write @xmath1593 we expand and take the expectation , keeping in mind that @xmath1594 will be chosen small and that @xmath1595 , by .
this gives @xmath1596 by discarding all the terms that are @xmath1597 , we obtain @xmath1598({\beta}\ell)^2 + o(\{{\beta}\ell\}^2 ) , \qquad \ell\to\infty.\ ] ] therefore there exists @xmath1599 and @xmath1600 such that , for @xmath261 small enough , @xmath1601 the second step is to extend to general disorder .
this will be a consequence of inequality in lemma [ lem : gapprox ] below . finally , @xmath1602 .
indeed , we have that @xmath1603 when @xmath1604 , whereas @xmath1605 when @xmath1606 and @xmath1607 .
moreover , the inequality is clearly satisfied for the finite set of values @xmath1608 .
[ lem : gapprox ] for every @xmath195 there exists a @xmath1609 such that , for @xmath45 small enough and for @xmath1610 , @xmath1611 and latexmath:[\[\label{firstineq } in the proof we use the additional assumption on the charge distribution stated in .
we give the proof only for case ( b ) , namely , @xmath180 is continuous with a density that is in @xmath1613 for some @xmath1614 .
the proof requires that the fifth moment of @xmath180 is finite ( which is amply guaranteed by ) and that there exists a @xmath1615 such that the function @xmath1616 is integrable on @xmath2 ( which is guaranteed by ( b ) ) .
the proof for case ( a ) is analogous and is omitted .
we refer to petrov ( * ? ? ?
* theorem 13 , chapter vii ) for the necessary details on local limit theorems .
we recall and note that in the proof we may remove the term @xmath1617 from the definition of @xmath1495 because we are only considering differences of @xmath1495-functions
. we will use @xmath1618 to denote a strictly positive constant that depends on @xmath863 only and whose value may change from line to line .
we begin with the proof of .
to prove , we use an edgeworth expansion for the density @xmath1619 of @xmath1620 ( see ( * ? ? ?
* theorem 2 , chapter xvi , section 2 ) ) .
let @xmath1621 denote the standard normal density .
then there are three polynomials @xmath1622 such that @xmath1623\big| = o\big(\ell^{-3/2}\big ) } , \qquad \ell \to \infty,\ ] ] where , for @xmath1624 , @xmath1625 with @xmath1626 for each @xmath1627 .
we recall , and from deduce that @xmath1628 a similar computation allows us to write , for @xmath1629 and @xmath1117 a standard normal random variable , @xmath1630 combining and , we get @xmath1631 , \quad \ell \to \infty,\,\beta \ell \geq { \varepsilon},\ ] ] where @xmath1632 in the remainder term in the right - hand side of , the factor @xmath1633 comes from the integral @xmath1634 , while the factor @xmath1635 comes from the factorisation of @xmath1636 ( recall ) and the fact that @xmath1637 is bounded from above and below by two strictly positive constants depending on @xmath863 , uniformly in @xmath1610
. the product of these factors is harmless when @xmath1638 .
also note that the term between brackets in the right - hand side of converges to @xmath351 as @xmath1223 , uniformly in @xmath1639 .
we next take the logarithm of for @xmath231 and @xmath1640 , and use the fact that @xmath1641 is lipschitz on @xmath1642 , so that there exists a @xmath1295 such that @xmath1643_{t=\ell}^{t=\ell+1}\bigg | + o\big(\tfrac{(1 + 2\beta \ell)^{1/2}}{({\beta}\ell)^{1/2}\ell^{3/2}}\big),\\ & \ell \to \infty,\,\beta \ell
\geq { \varepsilon } , \end{aligned}\ ] ] where @xmath1644 differentiating @xmath1645 , we get @xmath1646 using , we have , for @xmath1624 , @xmath1627 and @xmath1647 , @xmath1648_{\ell}^{\ell+1}\bigg |\leq \max_{t\in [ \ell,\ell+1 ] } \big|\frac{\partial}{\partial t } \mathfrak{t}_{k , s , j}(\beta , t)\big| \leq c_{{\varepsilon } } \ , \beta^{3/2 } , \qquad \ell\geq { \varepsilon}/\beta.\ ] ] combining and , we get .
to prove , we again use the edgeworth expansion introduced in but until @xmath1649 only , i.e. , @xmath1650\big| = o\big(\tfrac{1}{\sqrt{\ell}}\big ) } , \qquad \ell \to \infty,\ ] ] where @xmath1651 with @xmath1652 .
we compute @xmath1653 , and combine and with the observation that @xmath1654 , to obtain @xmath1655 , \quad \ell \to \infty,\,\beta \ell \geq { \varepsilon}.\ ] ] since @xmath1637 is bounded by strictly positive constants uniformly in @xmath1656 , it follows from that @xmath1657 is bounded for all @xmath1610 .
thus , we can rewrite in the form @xmath1658\\ & \leq e^{h_g(\ell)}\ , \bigg[1+c_{\varepsilon}\frac{1}{\sqrt{\beta } \ell } + o\big(\tfrac{1}{\sqrt{\ell}}\big ) \bigg ] , \qquad \ell \to \infty,\,\beta \ell \geq { \varepsilon}. \end{aligned}\ ] ] the reverse inequality holds with @xmath1618 replaced by @xmath1659 . taking the logarithm
, we conclude that @xmath1660 \leq h_g(\ell ) + \frac{c_{\varepsilon}}{\sqrt{\ell } } , \qquad \ell \to \infty,\,\beta \ell \geq { \varepsilon},\ ] ] and similarly @xmath1661 , which completes the proof of .
we need an upper bound on @xmath1662 ( see ( * ? ? ?
* ( 3.27 ) ) ) .
write @xmath1663 [ [ case - ell - leq - varepsilonbeta . ] ] case @xmath1664 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + define @xmath1665 which is a function interpolating between @xmath1666 and @xmath1667 .
rewrite @xmath1668 where @xmath1669 } { { { \ensuremath{\mathbb e } } } \big[e^{-{\beta}{\omega}_\ell^2+{\delta}{\omega}_\ell}\big]}.\ ] ] note that ( the derivatives are w.r.t .
the parameter @xmath441 ) @xmath1670 a taylor expansion to fifth order gives @xmath1671 moreover , @xmath1672 e^{h_t } \big),\quad { { \ensuremath{\mathcal g } } }
^{(3)}(t ) = { \widetilde}{{\ensuremath{\mathbb e } } } \big ( [ 3h'_th''_t + h'^3_t ] e^{h_t } \big),\\ & { { \ensuremath{\mathcal g } } } ^{(4)}(t ) = { \widetilde}{{\ensuremath{\mathbb e } } } \big ( [ 3h''^2_t + 6 h'^2_t h''_t + h'^4_t ] e^{h_t } \big),\quad { { \ensuremath{\mathcal g } } } ^{(5)}(t ) = { \widetilde}{{\ensuremath{\mathbb e } } } \big ( [ 15 h'_t h''^2_t + 10 h'^3_t h''_t + h'^5_t ] e^{h_t } \big ) \end{aligned}\ ] ] and @xmath1673 the terms arising from @xmath1674 in can be shown to be of negligible order .
therefore we obtain , for @xmath45 small enough , @xmath1675.\ ] ] it remains to provide upper bounds on @xmath1676 for @xmath1677 . to that end , let us write @xmath1678 on the one hand , by a first order expansion of the exponential we get @xmath1679 \leq \sqrt{2 } { \beta}^{1/2}\ell ( 1+c{\varepsilon}^{3/2 } + c{\varepsilon}^{1/2}).\ ] ] on the other hand , since @xmath1680 we get @xmath1681 therefore @xmath1682 with the same method of proof we obtain @xmath1683 inserting and into , we arrive at @xmath1684 which is the desired estimate , cf . .
[ [ case - ell - varepsilonbeta . ] ] case @xmath1685 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + decompose @xmath1686 we already know that @xmath1687 .
moreover , by lemma [ lem : gapprox ] , @xmath1688 from which we get @xmath1689 which is the desired estimate , cf . .
the proof is the same as in @xcite .
it is enough to prove that @xmath1690 ( see ( * ? ? ?
* ( 3.35 ) ) ) . in the analogue of (
* ( 3.35 ) ) , we decompose the sum in two parts : ( a ) @xmath1691 ; ( b ) @xmath1692 .
for ( a ) we use a gaussian approximation .
write @xmath1693 we already know that @xmath1694 . moreover , by lemma [ lem : gapprox ] , @xmath1695 which is the desired estimate .
for ( b ) we write @xmath1696 where @xmath1697 . using @xmath1698 we get @xmath1699 inserting this estimate into the analogue of ( * ? ? ? * ( 3.35 ) ) , and noting that @xmath1700 which we know from ( i ) in , we get the claim .
as promised in section [ ss : motivation ] , in this appendix we prove two modest results for the quenched version of the model , which has path measure @xmath1701 defined by ( recall ( [ eq : def.polmeasure][eq : def.hamiltonian ] ) ) @xmath1702 where @xmath40 is the set of nearest - neighbour paths starting at @xmath41 and @xmath1703 is the _ quenched partition function of length _
@xmath49 .
let @xmath1710 be the one - sided path that takes right - steps only , i.e. , @xmath1711 for @xmath308 . recall and estimate @xmath1712 = ( \tfrac{1}{2})^n\,e^{-\beta\sum_{i=1}^n \omega_i^2 } = ( \tfrac{1}{2})^n\,e^{-\beta n + o(n)}.\ ] ] moreover , by jensen s inequality we have ( recall ) @xmath1713\\ \label{eq:3 } & \geq r_n(s ) \left(\frac{1}{r_n(s ) } \sum _ { { x\in{\mathbb{z}}\colon } \atop { l_n(s , x)>0 } } \sum_{i=1}^n \omega_i { { \sf 1}}_{\{s_i = x\}}\right)^2 = \frac{1}{r_n(s)}\,\omega_n^2 .
\end{aligned}\ ] ] combining ( [ eq:2][eq:3 ] ) , we obtain @xmath1714\\ & \leq \exp\left\{-\beta n \left[\frac{1}{c_1n^2}\,\omega_n^2 -1-\tfrac{\log 2}{\beta}\right]\right\}. \end{aligned}\ ] ] by the strong law of large numbers for @xmath22 , we have @xmath1715 for @xmath29-a.a .
@xmath22 , and so the term between square brackets equals @xmath1716 $ ] with @xmath1717 .
therefore , by choosing @xmath985 small enough so that @xmath1718 , we get ( [ eq:1 ] ) with @xmath1719 .
fix @xmath202 .
pick @xmath1722 such that @xmath1723 if @xmath1724 , then we can choose @xmath1725 in proposition [ pr:1 ] and use the inequality @xmath1726 to conclude that a positive fraction of the sites is visited precisely once . consequently
, if the polymer chain chooses to go to the right , then @xmath1727 has a strictly positive @xmath1728 . | this paper studies an undirected polymer chain living on the one - dimensional integer lattice and carrying i.i.d
. random charges .
each self - intersection of the polymer chain contributes to the interaction hamiltonian an energy that is equal to the product of the charges of the two monomers that meet . the joint probability distribution for the polymer chain and the charges is given by the gibbs distribution associated with the interaction hamiltonian .
the focus is on the _ annealed free energy _ per monomer in the limit as the length of the polymer chain tends to infinity .
we derive a _ spectral representation _ for the free energy and use this to prove that there is a critical curve in the parameter plane of charge bias versus inverse temperature separating a _ ballistic phase _ from a _
subballistic phase_. we show that the phase transition is _
first order_. we prove large deviation principles for the laws of the empirical speed and the empirical charge , and derive a spectral representation for the associated rate functions .
interestingly , in both phases both rate functions exhibit _ flat pieces _ , which correspond to an inhomogeneous strategy for the polymer to realise a large deviation . the large deviation principles in turn
lead to laws of large numbers and central limit theorems .
we identify the scaling behaviour of the critical curve for small and for large charge bias .
in addition , we identify the scaling behaviour of the free energy for small charge bias and small inverse temperature . both are linked to an associated _ sturm - liouville eigenvalue problem_. a key tool in our analysis is the ray - knight formula for the local times of the one - dimensional simple random walk .
this formula is exploited to derive a _ closed form expression _ for the generating function of the annealed partition function , and for several related quantities .
this expression in turn serves as the starting point for the derivation of the spectral representation for the free energy , and for the scaling theorems .
what happens for the _ quenched free energy
_ per monomer remains open .
we state two modest results and raise a few questions . |
the importance of quantum entanglement as a physical resource for performing informational tasks which would be classically impossible @xcite has spurred the study of its dynamical behavior in many different systems .
the time - evolution of most of these is reversible and generated by a hamiltonian ; however , it is important for quantum entanglement to be used as an efficient physical resource that its temporal behavior should also be studied when systems are driven by noisy environments and their dynamics is irreversible . in the following , we will consider _ open quantum systems _
@xcite i.e. systems where the interactions between the subsystem @xmath0 and the external environment @xmath1 , though weak , can not be neglected . in this case , a standard way of obtaining a manageable dissipative time - evolution of the density matrix @xmath2 describing the state of @xmath0 at time @xmath3 is to construct it as the solution of a liouville - type master equation @xmath4 $ ] , where the generator @xmath5 of lindblad type @xcite takes care of the effects of the environment through a characteristic matrix of coefficients known as _
kossakowski matrix_. this can be done by tracing away the environment degrees of freedom and by performing a markovian approximation , i.e. by studying the evolution on a slow time - scale and neglecting fast decaying memory effects .
then , the irreversible reduced dynamics of @xmath0 is described by one - parameter semigroups of linear maps , called _ quantum dynamical semigroups _
, obtained by exponentiation : @xmath6 , @xmath7 , such that @xmath8 $ ] . in order to guarantee full physical consistency , namely that @xmath9 be positivity preserving on all states of the compound system @xmath10 for any inert ancilla @xmath11 , @xmath12 must be completely positive @xcite .
the formalism of open quantum systems has been used to describe the tendency to thermal equilibrium of a small system in weak interaction with a large heat bath at a certain temperature .
the main tool in this thermodynamical picture is the quantum relative entropy @xcite ; it is related to the difference between the free energy of the irreversibly evolving open quantum system and that of its equilibrium asymptotic state : this difference monotonically decreases in time because so does the quantum relative entropy with respect to completely positive maps @xcite , as quantum dynamical semigroups are .
namely , the time derivative of the quantum relative entropy , called _ entropy rate _ , has a definite sign .
the quantum relative entropy has also been used as a possible measure of the entanglement content of a quantum state : the so - called _ relative entropy of entanglement _ provides a pseudo - distance between a state and the closed convex set of separable states @xcite . in @xcite the natural question
was raised whether the entropy production due to thermodynamical tendency to equilibrium is somewhat related to the _ entanglement rate _ , that is to the speed of variation of the relative entropy of entanglement .
a conjecture was put forward that for systems immersed in an external bath without a direct source of entanglement due to hamiltonian interactions , the absolute value of the entanglement production is always smaller than the entropy production .
typically , a system @xmath0 immersed in a large environment @xmath1 is subjected to decoherence ; therefore , one expects quantum entanglement to be generically depleted by a dissipative and noisy time - evolution .
the conjecture mentioned above is motivated by the fact that , if a quantum open system tends to a separable equilibrium state , then , in a suitable neighborhood of the latter , the entanglement production is zero while the entropy production is not .
indeed , in @xcite a concrete example that validates the conjecture is offered of a 2-qubit system in which only one of them evolves as a quantum open system .
in such a case an initial maximally entangled state evolves towards a separable steady state with an entropy production always larger than the speed with which entanglement is dissipated
. however , in certain specific situations , an environment affecting both parties of a bipartite system may even build quantum correlations between the subsystems which compose @xmath0 ( see , e.g. , @xcite ) .
in particular , in @xcite this possibility is shown to depend on the specific form of the generator of the reduced dynamics . in @xcite an inequality was found , involving the entries of such a matrix which , if fulfilled , is sufficient to ensure that a specific initial separable pure state of two qubits gets entangled . further , in @xcite this inequality was proven to be a necessary and sufficient condition for environment - induced entanglement in an initially separable pure state of two qubits . even more interestingly ,
starting from an initially separable state , the entanglement generated at small times can persist asymptotically ; also , starting from an initially entangled state , its entanglement content can asymptotically increase .
this work is organized as follows : in section 2 , we consider the open dynamics of two qubits with a generator that depends on a parameter which allows to range over all the above mentioned cases , analytically solving the master equation ; then , in section 3 , we introduce the notions of entropy and entanglement rates and the conjecture from @xcite ; finally , in section 4 , numerically studying the time - behaviour of the entanglement and entropy rates for various initial states , we show that , whenever there is asymptotic entanglement the conjecture in @xcite is violated , while it holds if there is no asymptotic entanglement .
let a bipartite system composed of two qubits be immersed in an external environment in such a way that , via standard weak - coupling limit techniques @xcite , one describes their reduced , irreversible dynamics by means of the master equation @xmath13 = -i\,\frac{\omega}{2}\big[\sigma_3\,,\,\varrho_t\big]\ , + \
, \sum_{i , j=1}^3 a_{ij}\big(\sigma_i \varrho_t \sigma_j - \frac{1}{2}\{\sigma_j\sigma_i , \varrho_t\}\big)\ , \ ] ] where @xmath14 is the system frequency , @xmath15 , @xmath16 is the @xmath17 identity matrix , @xmath18 , @xmath19 are the pauli matrices and the matrix @xmath20= \begin{pmatrix } 1 & i \alpha & 0 \\ -i \alpha & 1 & 0 \\ 0 & 0 & 1 \end{pmatrix}\ , \qquad \alpha\in\mathbb{r}\ , \
\alpha^2\leq 1\ , \ ] ] is positive semi - definite .
this latter request ensures that the semigroup generated by ( [ master - eq ] ) consist of completely positive maps @xmath12 for all @xmath21 @xcite . *
remark 1*by means of the single qubit pauli matrices @xmath22 and @xmath23 one writes the purely dissipative contribution to the generator as @xcite @xmath24=\sum_{i , j=1}^3a_{ij}\sum_{a , b=1}^2\bigl(\sigma^{(a)}_i\,\rho_t\,\sigma^{(b)}_j\,-\ , \frac{1}{2}\left\{\sigma_j^{(b)}\sigma_i^{(a)}\,,\,\rho_t\right\}\bigr)\ .\ ] ] in this way there are six kraus operators @xmath25 , @xmath26 , @xmath19 and the @xmath27 kossakowski matrix reads @xmath28=\begin{pmatrix } k^{(11)}&k^{(12)}\cr k^{(21)}&k^{(22 ) } \end{pmatrix}=\begin{pmatrix}a&a\cr a&a\end{pmatrix}\ .\ ] ] from the theory of open quantum systems @xcite one knows that the coefficients @xmath29 in the kossakowski matrix relative to the @xmath30-th pauli matrix of the @xmath31-th qubit , respectively the @xmath32-th pauli matrix of the @xmath33-th qubit , @xmath34 , @xmath35 , are determined by the fourier transforms of the two - point time - correlation functions with respect to an environment equilibrium state @xmath36 , @xmath37 , of the environment operators @xmath38 appearing in the system - environment interaction @xmath39 . the symmetric form of ( [ diss - gen ] ) thus results when both qubits are linearly coupled to bath operators such that : @xmath40 and @xmath41 . * * remark 2**considering two qubits weakly interacting with a thermal bath modeled as a collection of spinless , massless scalar fields ( see , e.g. , @xcite ) at very high temperature @xmath42 , the parameter @xmath43 in the kossakowski matrix is related to @xmath44 , i.e. @xmath45 , where @xmath14 is the system frequency when isolated from the environment . correspondingly , in the case of one qubit immersed in such a thermal bath at high temperature ( @xmath46 ) , any initial state is driven to the thermal asymptotic state @xmath47 the master equation ( [ master - eq ] ) is explicitly integrated in appendix a ; in the following we will mainly focus upon the time - evolution of initial states of the form @xmath48 diagonal with respect to the orthonormal vectors @xmath49 where @xmath50 , @xmath51 , and @xmath52 , @xmath53 , @xmath54 , @xmath55 form the so - called standard basis in @xmath56 , with respect to which the states ( [ in - state ] ) are represented by @xmath57 from equations ( [ r00])([r33 ] ) and ( [ a2 ] ) in appendix a , it turns out that these initial states evolve at time @xmath21 into states of the same form @xmath58 where @xmath59 and @xmath60 with @xmath61 and @xmath62 since @xmath63 , the asymptotic states resulting from the initial states ( [ in - state ] ) are @xmath64 there is thus a one - parameter family @xmath65 of asymptotic states such that all initial states of the form ( [ in - state ] ) with the same @xmath66 go into the same @xmath67 . in order to study the asymptotic entanglement generation capability of the present model
, we shall measure the entanglement of 2-qubit states @xmath68 by the concurrence @xcite : @xmath69 where @xmath70 are the square roots of the positive eigenvalues of @xmath71 with @xmath72 , @xmath73 denoting the complex conjugated matrix . for 2-qubit states of the form ( [ matform ] ) , @xmath74 and one easily computes @xmath75 this latter matrix has positive eigenvalues @xmath76 ( twice degenerate ) , @xmath77 , @xmath78 ; then , their square roots @xmath79 in decreasing order yield @xmath80 * * remark 3**as already emphasized in the introduction , despite decoherence , the presence of an environment need not have only destructive effects in relation to entanglement : entanglement can even be asymptotically increased with respect to the initial amount .
this can happen in the present case and the entanglement generation capability of the environment is entirely due to the non - hamiltonian contribution ( [ diss - gen ] ) to the generator in ( [ master - eq ] ) .
indeed , the 2-qubit hamiltonian does not contain coupling terms and can not be a source of entanglement ; instead , this can be true for ( [ diss - gen ] ) because the off - diagonal contributions in the kossakowski matrix ( [ kossmat ] ) couple the two qubits .
of course , this is only necessary , but not sufficient to ensure entanglement generation and its asymptotic persistence .
they indeed depend on a trade - off between the off - diagonal couplings and the purely decohering diagonal terms in ( [ kossmat ] ) .
in this section we shall introduce the notions of entropy and entanglement rates ; for sake of simplicity , we shall consider finite @xmath81-level systems whose states are described by normalized , positive , @xmath82 density matrices @xmath83
. given two such density matrices , their quantum relative entropy is defined by @xcite @xmath84 consider a quantum system with hamiltonian @xmath85 ; if in contact with a heat bath at temperature @xmath42 ( with the boltzmann constant @xmath86 ) , it is expected to be driven asymptotically into the thermal ( gibbs ) state @xmath87 , where @xmath88 $ ] .
suppose that under an irreversible time - evolution @xmath89 , an initial state @xmath90 is driven into thermal equilibrium , that is @xmath91 ; then , @xmath92 where @xmath93 is the von neumann entropy of the state @xmath68 .
since the second term corresponds to the system s internal energy , the first two contributions give the system s free energy corresponding to the time - evolving state @xmath94 @xcite : @xmath95 finally , @xmath96 implies that the quantum relative entropy is related to the difference of free energies @xmath97 because of the second law of thermodynamics , the above quantity should be positive and its time - derivative non - positive .
the first property is guaranteed by the properties of the quantum relative entropy @xcite , while the second one holds true when the irreversible time - evolution is given by a markovian semigroup , that is when @xmath98 $ ] and @xmath99 for all @xmath100 . indeed , since @xmath101=\varrho_t$ ]
, one derives @xmath102\,||\,\gamma_t[\varrho_t]\big)= s\big(\gamma_{t - s}\circ\gamma_s[\varrho]\,||\,\gamma_{t - s}\circ\gamma_s[\varrho_t]\big)\\ & \leq & s(\gamma_s[\varrho]\,||\,\gamma_s[\varrho_t])=s(\varrho_s\,||\,\varrho_t)\quad\forall\ 0\leq s\leq t\ , \end{aligned}\ ] ] where the last inequality follows from the fact that the quantum relative entropy decreases under the action of completely positive trace - preserving maps @xcite . based on the previous thermodynamical arguments
, one may consider generic open quantum dynamics @xmath103=\varrho_t$ ] with asymptotic states @xmath104 , that are not necessarily thermal ones .
the speed of convergence to such stationary states starting from an initial state @xmath90 will then be measured by the _ entropy rate _
@xmath105 = -\frac{d}{dt}s(\varrho_t||\varrho_\infty ) = { \rm tr}\big(\dot{\varrho}_t(\log\varrho_\infty\,-\,\log\varrho_t)\big)\ .\ ] ] the entropy production that accompanies the tendency to equilibrium of the states of the form ( [ rho - tilde ] ) is easily computed ; indeed , being the states @xmath2 and @xmath106 diagonal with respect to the same orthonormal basis , the entropy rate ( [ entroprod0 ] ) has the analytic expression @xmath107=\dot{a}_t\,\log\frac{(1-\alpha)^2(1-c)}{a_t(3+\alpha^2)}\,+\ , \dot{b}_t\,\log\frac{(1-\alpha^2)(1-c)}{b_t(3+\alpha^2)}\ , + \,\dot{d}_t\,\log\frac{(1+\alpha)^2(1-c)}{d_t(3+\alpha^2)}\ .\ ] ] when the density matrix @xmath68 is the state of , say , a bipartite quantum system , it makes sense to introduce the _ relative entropy of entanglement _ , @xmath108= \inf_{\varrho_{sep } } s(\varrho||\varrho_{sep})\ , \ ] ] as a measure of the entanglement content of @xmath68 .
indeed , the above quantity vanishes if and only if @xmath68 is separable and can be used to measure the distance of @xmath68 from the convex set of separable states ; furthermore , it can not increase , but at most remain constant , under the action of local operations , described by trace - preserving completely positive maps acting independently on the two parties @xcite .
analogously to what was done for the entropy production , one may look at the _ entanglement rate _ when the system evolves , i.e. at the time - derivative of the pseudo - distance @xmath109 = \frac{d}{dt}e[\varrho_t ] \ .\ ] ] in @xcite it was argued that @xmath110\big| \leq \sigma[\varrho_t]\\ ] ] always holds in absence of direct entangling interactions between the parties .
the argument on which the conjecture is based is that decoherence is expected to deplete entanglement before reaching the asymptotic state and thus before the entropy production vanishes . such asymptotic intuition
is then extrapolated at all times .
now we will illustrate the various possibilities offered by the reduced dynamics discussed in the previous section ; in particular , we will compare the entropy and entanglement rates , thus checking the validity of the conjecture ( [ vedralconj ] ) .
we will first derive an explicit expression for the relative entropy of entanglement ( [ ent - entr0 ] ) in the case of states of the form ( [ rho - tilde ] ) and then compute numerically the behaviour of its time - derivative ( [ ent - prod ] ) . let us first rewrite ( [ ent - entr0 ] ) as follows : @xmath111 where @xmath112 is the von neumann entropy of the time - evolving state .
the following result , which we prove in appendix b , helps to explicitly solve the above maximization problem . * * proposition**in the case of states as in ( [ rho - tilde ] ) , the supremum in ( [ constr0 ] ) is achieved for separable states of the form @xmath113 are real and such that @xmath114 this leads to the following maximization problem @xmath115 which can be analytically solved .
the above maximization problem is explicitly solved in appendix b thus permitting to calculate numerically the entanglement rate ( [ ent - prod ] ) , to compare it with the entropy rate ( [ entroprod0 ] ) and to check the conjecture ( [ vedralconj ] ) .
we shall do this in a number of cases that cover all possible initial and asymptotic entanglement properties for which we plot the behaviors of the relative entropy and of the relative entropy of entanglement , separately , while the entropy rate and the entanglement rate are plotted together for direct comparison . in the following ,
the choice of the range of values for the plots axes was made only for graphic reasons to make the plots clearer .
moreover , in the first four cases , we took the parameter in the matrix @xmath116 to be @xmath117 as it makes the plots easier to read ; changing @xmath43 does not alter the results . in the last example , instead , we show two different behaviors of the entanglement of the initial state depending on the choice of the parameter @xmath43 . *
* case 1.**an initial pure separable state ( [ state1 ] ) goes into a mixed separable state ; the dissipative time - evolution is not able to generate entanglement at any time , as shown by the second and the third plot below where the entropy of entanglement and the entanglement rate are both zero . in this case
the conjecture ( [ vedralconj ] ) holds .
@xmath118 ; left : @xmath112 ; right : @xmath119,title="fig:",width=226 ] ; left : @xmath112 ; right : @xmath119,title="fig:",width=226 ] ; @xmath120 $ ] dashed line , @xmath121|$ ] continuous line , width=226 ] * * case 2.**an initial mixed separable state ( [ state2 ] ) goes into a mixed entangled state and the conjecture ( [ vedralconj ] ) is violated after some time .
@xmath122\ .\end{aligned}\ ] ] ; left : @xmath123 ; right : @xmath124$],title="fig:",width=226 ] ; left : @xmath123 ; right : @xmath124$],title="fig:",width=226 ] ; @xmath120 $ ] dashed line , @xmath121|$ ] continuous line , width=226 ] * * case 3.**an initial mixed entangled state ( [ state3 ] ) goes into an asymptotic mixed state which is more or equally entangled and the conjecture ( [ vedralconj ] ) is violated after some time .
@xmath125\ .\end{aligned}\ ] ] ; left : @xmath123 ; right : @xmath124$],title="fig:",width=226 ] ; left : @xmath123 ; right : @xmath124$],title="fig:",width=226 ] ; @xmath120 $ ] dashed line , @xmath121|$ ] continuous line , width=226 ] * * case 4.**an initial mixed entangled state ( [ state5 ] ) goes into a state with less entanglement @xmath126 with the choice @xmath117 , the dissipative time - evolution shows a sudden death of entanglement , that is the concurrence , instead of plotting the concurrence as defined in ( [ concurrence1 ] ) , we have plotted the difference @xmath127 : this simply means that , as soon as this difference becomes negative , the state is separable . ] ( [ concurrence1 ] ) vanishes at finite time . the conjecture ( [ vedralconj ] ) always holds . ; @xmath128,width=226 ] ; left : @xmath123 ; right : @xmath124$],title="fig:",width=226 ] ; left : @xmath123 ; right : @xmath124$],title="fig:",width=226 ] ; @xmath120 $ ] dashed line , @xmath121|$ ] continuous line , width=226 ] * * case 5.**an initial mixed entangled state ( [ state4 ] ) goes into a mixed entangled state with more or less entanglement depending on the choice of the parameter @xmath43 . @xmath129 the concurrence of the asymptotic state @xmath130 can be larger or smaller than @xmath131 depending on the value of the parameter @xmath43 .
if , for instance , we take @xmath117 , then @xmath132 , i.e. the asymptotic state has less entanglement than the initial state , and from the plot of the entanglement rate vs. the entropy rate ( figure 11 ) we can see that the conjecture ( [ vedralconj ] ) is always violated . ; left : @xmath133 ; right : @xmath124$],title="fig:",width=226 ] ; left : @xmath133 ; right : @xmath124$],title="fig:",width=226 ] ; @xmath120 $ ] dashed line , @xmath121|$ ] continuous line , width=226 ] if , instead , we take for example @xmath134 , then the initial entanglement first diminishes and then increases again , leading to an asymptotic state with more entanglement than the initial one , as can be seen from the plot of the entropy of entanglement as a function of time ( figure 12 ) . from the corresponding plot of the entanglement rate vs. the entropy rate ( figure 13 )
we can see that in this case the conjecture ( [ vedralconj ] ) is violated after some time . ; left : @xmath123 ; right : @xmath124$],title="fig:",width=226 ] ; left : @xmath123 ; right : @xmath124$],title="fig:",width=226 ] ; @xmath120 $ ] dashed line , @xmath121|$ ] continuous line , width=226 ] * remark 4*in the last plot of the entanglement rate vs. the entropy rate , the cusp is due to the change of sign in the entropy of entanglement @xmath123 and to the fact that in the conjecture ( [ vedralconj ] ) the absolute value of the entanglement rate @xmath135 $ ] is considered . on the other hand , all the other plots of the entanglement rate present a continuous behavior which reflects the fact that the entropy of entanglement does not change sign , i.e. it either increases or decreases monotonically .
finally , the plots of the relative entropy show its monotonic behavior under the action of completely positive trace - preserving maps ; and to this corresponds a monotonic decreasing behavior for the entropy rate .
the time - derivative of the quantum relative entropy serves as a measure of how fast an open quantum system tends to equilibrium dissipating free energy under a quantum dynamical semigroup of completely positive maps generated by a lindblad - type master equation . on the other hand , via a variational formulation , the relative entropy may be used as a pseudo - distance of an entangled bipartite state from the convex subset of separable states ( relative entropy of entanglement ) ; therefore , its time derivative can be interpreted as the speed with which a time - evolving state moves toward , or away from , becoming separable .
based on the expectation that the entanglement content of dissipatively - driven bipartite systems disappears asymptotically due to decoherence effects , in @xcite a conjecture was put forward , namely that the entropy rate , measured by ( minus ) the time - derivative of the relative entropy of a dissipatively evolving state and its asymptotic state , should always be larger than the absolute value of the time - derivative of the relative entropy of entanglement .
however , beside being a source of decoherence , an environment can in some cases build quantum correlations that can even persist asymptotically : in this paper we have studied the fate of the above conjecture in the case of a lindblad - type master equation that presents a rich manifold of asymptotic states that may be more or less entangled with respect to the initial states they emerge from .
the entropy and entanglement rates have been explicitly calculated and numerically plotted for a class of initial states .
it turns out that , when the asymptotic state is entangled , the conjecture is violated either at all times or after a finite time ; instead , the conjecture is confirmed in all cases when the asymptotic state is separable .
the conjecture put forward in @xcite should be thus reformulated as follows : @xmath136\big| \leq \sigma[\rho_t]\quad\forall \rho_t\ \hbox{with}\ \varrho_\infty\ \hbox{separable}\ .\ ] ] while the asymptotic predominance of the entropy rate over the entanglement rate in the latter case has already been explained in @xcite based on the fact that there are no entangled states in a suitable neighborhood of the separable asymptotic state , the truly remarkable fact about ( [ vedralconjref ] ) is its validity at all @xmath21 in all the cases that have been checked .
in order to explicitly solve the master equation ( [ diag - noise ] ) , one first writes the matrix @xmath116 in diagonal form : @xmath137 and recasts the dissipative term in ( [ diss - gen ] ) in the form @xmath138&=&2(1+\alpha)\big(\sigma_-
\varrho_t \sigma_+ - \frac{1}{2}\{\sigma_+\sigma_- , \varrho_t\}\,+\ , 2(1-\alpha)\big(\sigma_+ \varrho_t \sigma_- - \frac{1}{2}\{\sigma_-\sigma_+ , \varrho_t\}\big ) \nonumber \\ & + & \sigma_3 \varrho_t \sigma_3 - \frac{1}{2}\{\sigma_3\sigma_3 , \varrho_t\}\ , \end{aligned}\ ] ] with @xmath139 .
since @xmath140 , it follows that @xmath141 .
thus , setting @xmath142 , in the interaction picture the master equation ( [ diag - noise ] ) becomes @xmath143e^{-iht } \nonumber \\ & = & 2(1+\alpha)\big(\sigma_- \tilde{\varrho}_t \sigma_+ - \frac{1}{2}\{\sigma_+\sigma_- , \tilde{\varrho}_t\}\big)\,+\ , 2(1-\alpha)\big(\sigma_+ \tilde{\varrho}_t\sigma_- - \frac{1}{2}\{\sigma_-\sigma_+ , \tilde{\varrho}_t\}\big ) \nonumber \\ & + & \sigma_3 \tilde{\varrho}_t\sigma_3 - \frac{1}{2}\{\sigma_3\sigma_3 , \tilde{\varrho}_t\}\ .\end{aligned}\ ] ] in order to solve it , it proves convenient to represent @xmath144 with respect to the orthonormal basis ( [ onb ] ) . indeed , using that @xmath145 one derives from ( [ tilde - master - eq ] ) the following equations : @xmath146 plus the complex conjugated equations for @xmath147 , @xmath148 ; whence @xmath149 of the remaining equations , two of them couple the off - diagonal terms @xmath150 and @xmath151 , yielding @xmath152 while the other three solutions couple the diagonal entries : @xmath153 where @xmath154 is a constant of the motion and @xmath155 are quantities which decay asymptotically with @xmath156 . the remaining entries @xmath157 follow from complex conjugation . by returning to the schrdinger representation , using ( [ a1 ] )
the explicit solution of ( [ diag - noise ] ) reads @xmath158
in order to prove the proposition in section 3 , let us consider the spectral decompositions @xmath159 ( see ( [ rho - tilde ] ) ) and @xmath160 .
we have : @xmath161\big)\ , \end{aligned}\ ] ] where the inequality follows from the convexity of @xmath162 , @xmath163 and @xmath164 . also , we have introduced the completely positive map @xmath165:=\sum_{i=1}^4|i\rangle\langle i|\varrho|i\rangle\langle i|\ , \ ] ] on the 2-qubit density matrices @xmath166 that diagonalizes its argument with respect to the orthonormal basis ( [ onb ] ) .
this map has the following property which allows one to analytically solve the variational problem ( [ log ] ) . *
* proof:**given the density matrix of an arbitrary @xmath168-qubit state in the standard basis @xmath169 the action of the map @xmath170 transforms it into a density matrix of the form @xmath171=\begin{pmatrix } 0 & 0 & 0 & \varrho_{11,11 } \end{pmatrix}\ .\ ] ] by partial transpostion @xcite , @xmath172 $ ] is entangled if and only if @xmath173 .
but then , the partially transposed @xmath90 ( with respect to the second qubit ) , @xmath174 can not be positive semi - definite , for @xmath175 in the sub - matrix @xmath176 .
therefore , if @xmath90 is separable , then also @xmath172 $ ] must be so .
@xmath177 observe that ( [ log ] ) implies @xmath178\big)$ ] ; on the other hand , since @xmath170 maps separable states into separable states , @xmath179\big ) \leq
\sup_{\varrho_{sep}}{\rm tr}\big(\varrho_t\log\varrho_{sep}\big)\ .\ ] ] thus , the maximum in ( [ log ] ) is attained on the subset @xmath180 of separable qubit states that are diagonal with respect to the orthonormal basis , namely of the form ( [ constr1 ] ) with the second bound on the real parameters @xmath181 in ( [ param ] ) coming from the condition of positivity under partial transposition of matrices of the form ( [ matform ] ) , which is necessary and sufficient for separability
. it thus follows from ( [ constr0 ] ) and ( [ log ] ) that for @xmath2 as in ( [ rho - tilde ] ) the relative entropy of entanglement can be reduced to the computation of ( [ max ] ) . in order to explicitly solve such a variational problem
, we seek the stationary points of a function of the form @xmath182 with given @xmath183 such that @xmath184 , relative to variations of the parameters @xmath181 over values achieving separable states of the form ( [ constr1 ] ) .
stationarity implies @xmath185 whence @xmath186 and @xmath187 .
however , this can be the required solution only if the state @xmath90 in @xmath188 $ ] is separable so that @xmath188=0 $ ] .
otherwise , the solution must lie on the border of the subset of separable states of the form ( [ constr1 ] ) , where the inequality in ( [ param ] ) is saturated . from @xmath189 and @xmath190 ,
one gets @xmath191 so that the function to be maximized becomes @xmath192 stationarity with respect to @xmath193 leads to a system of two equations for the two unknowns @xmath194 and @xmath195 in terms of the coefficients @xmath196 . from setting @xmath197 and from the condition
@xmath184 it follows that @xmath198 and that @xmath199^{1/2}\big\}.\end{aligned}\ ] ] by inserting into it the values ( [ sa])([sc ] ) , this expression yields the separable state of the form ( [ constr1 ] ) which is closest to an evolving entangled state of the form ( [ rho - tilde ] ) .
though cumbersome , the resulting entanglement rate ( [ ent - prod ] ) is amenable to numerical inspection . | we study a recent conjecture about the behavior of the quantum relative entropy compared to the relative entropy of entanglement in open bipartite systems .
the conjecture states that , under a dissipative time - evolution , the positive rate of change of the relative entropy will always be larger than that of the relative entropy of entanglement .
after explicitly solving a two - qubit master equation of lindblad - type with separable and entangled stationary states , we show that the conjecture can be violated for initial states with an entangled asymptotic state , while it appears to be confirmed when the asymptotic states are separable . |
considering the system rotating along the @xmath0-axial in the counter - clockwise direction , we have @xcite @xmath172 here , @xmath173 is the coordinates in the inertial lab frame , while @xmath174 is in the co - rotating frame .
we obtain the general coordinate transformation matrix as @xmath175=[\frac{\partial x^{\alpha}}{\partial x^{\beta}}]=\left[\begin{array}{cccc } 1\\ \omega y / c & \cos\omega t & \sin\omega t\\ -\omega x / c & -\sin\omega t & \cos\omega t\\ & & & 1 \end{array}\right].\ ] ] the spacetime metric @xmath13 in the co - rotating frame , as a covariant tensor , can be calculated by @xmath176=[\frac{\partial x^{\mu}}{\partial x^{\alpha}}\cdot\eta^{\alpha\beta}\cdot\frac{\partial x^{\nu}}{\partial x^{\beta}}]\nonumber \\ & = \left[\begin{array}{cccc } -1 & -\frac{\omega y}{c } & \frac{\omega x}{c}\\ -\frac{\omega y}{c } & 1-\frac{\omega^{2}y^{2}}{c^{2 } } & \frac{\omega^{2}xy}{c^{2}}\\ \frac{\omega x}{c } & \frac{\omega^{2}xy}{c^{2 } } & 1-\frac{\omega^{2}x^{2}}{c^{2}}\\ & & & 1 \end{array}\right],\end{aligned}\ ] ] where @xmath177 is the metric in the inertial lab frame .
and we also have @xmath178^{-1}\nonumber \\ & = \left[\begin{array}{cccc } -1+\frac{\omega^{2}}{c^{2}}(x^{2}+y^{2 } ) & -\frac{\omega y}{c } & \frac{\omega x}{c}\\ -\frac{\omega y}{c } & 1\\ \frac{\omega x}{c } & & 1\\ & & & 1 \end{array}\right].\end{aligned}\ ] ]
in the canonical quantization of the rotating em field , it is crucial to find a proper orthogonal relation of the field operator @xmath58 .
this relation can be obtained with the help of a density - flow relation derived from the equation of motion . from the equation of @xmath53 @xmath30\mathbf{a}_{i}=0,\ ] ] we obtain the following density - flow relation @xmath179=\partial_{s}[\mathbf{a}^{*}{\cal j}\mathbf{a}],\label{eq : den - flow}\ ] ] where @xmath180 integrating eq.([eq : den - flow ] ) over the whole space , we obtain @xmath181\big)=0,\ ] ] where @xmath182 is the integral over the cross section and gives a constant area @xmath57 .
then we know that this integral is a conserved constant independent of time ( but still not determined yet )
. then we can find some orthogonal relations .
for the eigen solution @xmath183 we have the following orthogonal relation @xmath184 & = -\frac{2i|\overline{\mathrm{z}}_{k}|^{2}\mathsf{l}}{c}(\omega_{k}+v_{r}k)\,\delta_{kq}\delta_{\lambda\sigma},\nonumber \\
\int_{0}^{\mathsf{l}}ds[\boldsymbol{a}_{q\sigma}{\cal d}\boldsymbol{a}_{k\lambda } ] & = \int_{0}^{l}ds[\boldsymbol{a}_{q\sigma}^{*}{\cal d}\boldsymbol{a}_{k\lambda}^{*}]=0.\label{eq : orthog}\end{aligned}\ ] ] notice that no matter @xmath185 and @xmath186 , we both have @xmath187 . with the help of the above orthogonal relation , for the field operator @xmath59,\ ] ] we have @xmath188 & = -2i|\overline{\mathrm{z}}_{k}|^{2}\mathsf{l}|k|\cdot\hat{a}_{k\lambda},\nonumber \\
\int_{0}^{l}ds[\boldsymbol{a}_{k\lambda}{\cal d}\hat{\mathbf{a } } ] & = 2i|\overline{\mathrm{z}}_{k}|^{2}\mathsf{l}|k|\cdot\hat{a}_{k\lambda}^{\dagger}.\label{eq : a_k}\end{aligned}\ ] ] notice that indeed the above terms in the integrals can be written as @xmath189 then , together with the canonical quantization conditions @xmath190=i\hbar\delta_{\lambda}^{\sigma}\delta(s - s')\cdot\mathsf{a}^{-1}$ ] and @xmath191=[\hat{{\cal e}}^{\lambda}(s , t),\,\hat{{\cal e}}^{\sigma}(s',t)]=0 $ ] , we can calculate @xmath192 $ ] from eq.([eq : a_k ] ) as follows @xmath193\\ = & \iint dsds'\big[\boldsymbol{a}_{k\lambda}^{*}(s , t){\cal d}\hat{\mathbf{a}}(s , t),\,\boldsymbol{a}_{q\sigma}(s',t){\cal d}\hat{\mathbf{a}}(s',t)\big]\\ = & \iint dsds'\big[\frac{\mathsf{a}_{k}^{*}\hat{{\cal e}}^{\lambda}(s)}{c\varepsilon_{0}}-i|k|\mathsf{a}_{k}^{*}\hat{\mathbf{a}}_{\lambda}(s),\,\frac{\mathsf{a}_{q}\hat{{\cal e}}^{\sigma}(s')}{c\varepsilon_{0}}+i|q|\mathsf{a}_{q}\hat{\mathbf{a}}_{\sigma}(s')\big]\\ = & \frac{2\hbar\mathsf{l}|k|}{c\varepsilon_{0}\mathsf{s}}\cdot|\overline{\mathrm{z}}_{k}|^{2}\cdot\delta_{\lambda}^{\sigma}\delta_{kq}.\end{aligned}\ ] ] therefore , we choose the normalization constant @xmath39 to be @xmath194^{\frac{1}{2}}=[\frac{\hbar}{2\varepsilon_{0}vc|k|}]^{\frac{1}{2}},\ ] ] where @xmath195 is the volume of the quasi-1d ring , so that we obtain the bosonic commutation relation @xmath192=\delta_{\lambda\sigma}\delta_{kq}$ ] . with the same method
, we can also check @xmath196=[\hat{a}_{k\lambda}^{\dagger},\,\hat{a}_{q\sigma}^{\dagger}]=0 $ ] .
after we have got the normalization constant @xmath39 , it is straightforward to verify that the hamiltonian of the rotating em field is @xmath197\\ & = \sum_{k,\lambda}\frac{1}{2}\hbar\omega_{k}(\hat{a}_{k\lambda}^{\dagger}\hat{a}_{k\lambda}+\hat{a}_{k\lambda}\hat{a}_{k\lambda}^{\dagger}).\end{aligned}\ ] ]
the hamiltonian equation can be derived from the following variation , @xmath198 where @xmath199 $ ] is the functional of independent variables @xmath74 and @xmath75 .
the above variation gives @xmath200\big)dt\\ = & p_{i}\delta x^{i}\big|_{a}^{b}+\delta\int(\frac{dx^{i}}{dt}-\frac{\partial h}{\partial p_{i}})\delta p_{i}+(\frac{dp_{i}}{dt}+\frac{\partial h}{\partial x^{i}})\delta x^{i}.\nonumber \end{aligned}\ ] ] @xmath201 and @xmath202 are the initial and final points of the trajectories of @xmath74 which are fixed , and thus the first term is zero . since @xmath203 and @xmath204 are independent variables , the above variation leads to the hamilton equations @xmath205 here we have utilized the following relation @xmath206dt\nonumber \\ & = \int\big[v^{i}\delta p_{i}+p_{i}\frac{d}{dt}(\delta x^{i})\big]dt\\ & = p_{i}\delta x^{i}\big|_{a}^{b}+\int\big[\frac{d}{dt}x^{i}\cdot\delta p_{i}-\frac{d}{dt}p_{i}\cdot\delta x^{i}\big]dt.\nonumber \end{aligned}\ ] ] therefore , for a charged particle in the co - rotating em field , we have @xmath207 here we denote @xmath69 , @xmath208 . and
we denote @xmath77 as the mechanical momentum . here
we take @xmath209 and @xmath210 , thus we have @xmath211 .
further , we still need to replace @xmath212 by the canonical momentum @xmath213 in the hamiltonian @xmath214 .
we emphasize that here we only have got the relation between @xmath215 and @xmath78 from eq.([eq : ham - e ] ) , but we did not have the definition of @xmath216 , so we can not just naively replace @xmath212 as @xmath217 . to find the relation between @xmath212 and @xmath215 , we should notice the following two relations , @xmath218 therefore ,
we obtain an equation about @xmath212 , i.e. , @xmath219 which leads to the solution ( @xmath212 should be negative , so the other positive solution is invalid ) @xmath220 therefore , we have @xmath82-ev^{0}a_{0},\ ] ] where @xmath221 .
for the jc - model for a two - level atom in a rotating ring cavity @xmath222 we can directly verify that the ground state is @xmath223 and the eigen energy is @xmath130 . when @xmath134 , the above eigen equations can be diagonalized exactly and we can obtain the eigen energy states in the single excitation subspace .
the eigen energies are @xmath226 and @xmath227 , where @xmath133 , and the eigenstates are @xmath228,\nonumber \end{aligned}\ ] ] where @xmath229 and @xmath230 are normalization constants .
when we use an external laser to drive the @xmath159 mode , the master equation of the system is @xmath144+\sum_{\alpha=+,-}\frac{\gamma}{2}\big(2\hat{a}_{\alpha}\rho\hat{a}_{\alpha}^{\dagger}-\{\rho,\hat{a}_{\alpha}^{\dagger}\hat{a}_{\alpha}\}\big),\ ] ] where @xmath231 is the driving term .
we make a unitary transformation by @xmath232 $ ] , and the equation becomes time independent . then we obtain equations for observable expectations as follows @xmath233 here we denote @xmath234 and @xmath235 , @xmath149 . in this equation
we have omitted terms of higher orders , like @xmath236 , so the above linear equations become complete .
this approximation requires that the driving strength is weak thus the excitation is very low . from this set of linear equations ,
we obtain the solution of @xmath237 , and then we obtain @xmath238 as follows @xmath239}{(fd+4{\cal e}^{2}g)^{2}},\label{eq : n+-}\\ \overline{n}_{-}\simeq & |\alpha_{-}|^{2}=\frac{16{\cal e}^{2}g^{4}\cdot f^{2}d}{(fd+4{\cal e}^{2}g)^{2}},\nonumber \end{aligned}\ ] ] where @xmath240\\ & + 16\tilde{\omega}^{2}[2g^{2}-\tilde{\omega}^{2}+\delta^{2}]^{2},\\ g:= & 16\delta^{2}\big[(2g^{2}-\tilde{\omega}^{2})\tilde{\omega}_{-}^{2}+2g^{2}\tilde{\omega}\delta\big]+\gamma^{4}(2g^{2}-\tilde{\omega}^{2})\\ & + 4\gamma^{2}\big[(2g^{2}-\tilde{\omega}^{2})\delta^{2}+(2g^{2}-\tilde{\omega}^{2})\tilde{\omega}_{-}^{2}+2g^{2}\tilde{\omega}\delta\big].\end{aligned}\ ] ] when the driving strength @xmath146 is weak , we omit terms of @xmath241 in eq.([eq : n+- ] ) and obtain @xmath242 for the case that the atom is resonant with the resonant frequency , i.e. , @xmath134 , we have @xmath243 , and we obtain @xmath148^{2}+\gamma^{2}\tilde{\omega}^{2},\\ f:= & \gamma^{4}\tilde{\omega}^{2}+8\gamma^{2}[2g^{4}+\tilde{\omega}^{4}-2g^{2}\tilde{\omega}^{2}+\delta^{2}\tilde{\omega}^{2}]\\ & + 16\tilde{\omega}^{2}[2g^{2}-\tilde{\omega}^{2}+\delta^{2}]^{2}\end{aligned}\ ] ] this is what we have shown in the text . the above master equation can be also solved numerically by setting a cutoff on the hilbert dimension of the cavity modes .
we find that when the driving strength @xmath146 is weak , the above analytical results of the steady photon number show well accordance with the numerical calculations . | we study quantum optics with the atoms coupled to the quantized electromagnetic ( em ) field in a non - inertial reference frame by making use of quantum field theory in curved spacetime .
we rigorously establish the microscopic model for a two - level atom interacting with the quantized em field in a rotating ring cavity by deriving a jaynes - cummings ( jc ) type hamiltonian . due to the two fold degeneracy of the ring cavity modes , the rabi splitting exhibits three rather than two resonant frequency peaks .
we find that the heights of the two side peaks show a sensitive linear dependence on the rotating velocity .
this high sensitivity can be utilized to detect the angular velocity of the whole system . *
introduction * _ _ the interference of of two light beams in a rotating ring can be utilized to measure the rotating velocity .
this is well known as the sagnac effect , which bases some optical gyroscope schemes @xcite . for the sagnac effect in the medium with linear dispersion ,
there have been a lot of sophisticated studies based on classical optics .
if we want to make the optical gyroscopes to an extremely high precision , it is necessary to consider the quantum fluctuations in this rotating optical system .
we notice that a rigorous quantum theory about the microscopic model about the interaction between the atoms and the quantized em field in a rotating reference frame is still not well established , but it is obviously essential for the study of quantum and nonlinear effects in a rotating optical system . in this letter
, we ascribe the effects of rotation to the `` curved '' spacetime metric according to the generic principles in relativity .
starting from the classical lagrangians of the em field and a charged particle based on the the principle of least action , we obtain the covariant motion equations in the rotating non - inertial reference frame ( nirf ) .
the variation is carried out in the rotating reference frame , and all the non - inertial physical effects rooted in the rotation are included in the `` curved '' spacetime metric .
then we obtain the quantized hamiltonian for quantum optics in nirf through the canonical quantization .
we derive a microscopic model of an atom interacting with the quantized em field in a rotating reference frame , which then gives the jaynes - cummings ( jc ) model for a rotating ring cavity coupled with a two - level atom . for a ring cavity in the inertial rest frame ,
the clockwise ( cw ) and counter - clockwise ( ccw ) propagating optical modes are always exactly degenerated .
thus , in the jc model of a ring cavity , the atom couples with the two degenerate modes simultaneously . due to the existence of two optical modes , the rabi splitting of this system exhibits three rather than two resonant frequency peaks .
more importantly , we find that the rotation of the ring would induce a detuning between the original degenerated modes in the non - inertial frame , and the heights of the two side peaks would change with the rotating speed due to this rotation induced detuning . at lower speed ,
the heights of the side peaks depend linearly on the rotating velocity .
this sensitivity can be utilized to detect the rotation of the whole system , which can be regarded as a quantum sagnac effect . * em field in the rotating reference frame * _ _ we consider the em field rotating around the @xmath0-axial in the ccw direction with angular speed @xmath1 .
physics laws must have the same covariant mathematical form in all reference frames ( including the non - inertial rotating frame that we are studying ) .
thus , in the rotating reference frame , the lagrangian density of the em field is @xmath2 here @xmath3 is the magnetic constant , @xmath4 and @xmath5 . in the above definitions , @xmath6 , where @xmath7 is the 4-dimensional coordinate , and @xmath8 ( @xmath9 ) is the electromagnetic 4-potential .
@xmath10 is the covariant derivation defined by @xmath11 here @xmath12 is the christoffel symbol and it can be calculated from the spacetime metric @xmath13 @xcite .
all the physical effects due to the rotation are included in the `` curved '' spacetime metric @xmath13 ( see ref.@xcite or appendix [ sub : space - time - metric ] ) , i.e. , @xmath14.\label{eq : g^uv}\ ] ] from the variation of the action @xmath15 we obtain the euler - lagrangian equation as @xmath16=\frac{\partial\tilde{{\cal l}}}{\partial a_{\nu}}\rightarrow\frac{1}{\sqrt{-\mathfrak{g}}}\partial_{\mu}[\sqrt{-\mathfrak{g}}f^{\mu\nu}]=\nabla_{\mu}f^{\mu\nu}=0,\end{gathered}\ ] ] where @xmath17 . here
@xmath18 is a functional of @xmath19 and @xmath20 .
this is the covariant form of the maxwell equation , which applies to all general reference frames ( both inertial and non - inertial ones ) @xcite . substituting the spacetime metric eq.([eq : g^uv ] ) into the above covariant equation , and taking the coulomb gauge @xmath21
, we obtain the dalembert equation @xmath22 as @xmath23\mathbf{a}_{i}=0.\ ] ] here @xmath24 , @xmath25 , and @xmath26 is the linear velocity .
next we reduce the problem into a quasi-1d ring configuration .
we assume that @xmath27 is homogenous in the transverse direction , and we have @xmath28 @xcite , where @xmath29 is the direction along the ring
. then we have @xmath30\mathbf{a}_{i}=0.\label{eq : motion}\ ] ] here @xmath31 and @xmath32 is the linear speed .
the general solution of the above equation is @xmath33,\ ] ] where @xmath34 , and @xmath35 is the length of the ring;^{-1}>2\pi r$ ] . here
we omit this small change when @xmath36 , because it is @xmath37 .
] @xmath38 is the polarization directions in the transverse section , and @xmath39 is a normalization constant .
the eigenvalue equation of eq.([eq : motion ] ) , @xmath40 , gives rise to the following dispersion relation @xmath41 therefore , the rotation makes the dispersion relation anisotropic , i.e. , for the two modes @xmath42 and @xmath43 , their frequencies @xmath44 no longer equal .
* quantization of the em field * * _ _ * under the coulomb gauge we used above , the canonical momentums are @xmath45 and @xmath46 where @xmath47 is the electric constant . the hamiltonian density @xmath48 is obtained as @xmath49 restricted in the quasi-1d ring configuration , respectively they become @xmath50 notice that since @xmath28 , the coulomb gauge @xmath51 leads to @xmath52 .
namely , @xmath53 only has two transverse directions , and so does the canonical momentum @xmath54 .
we apply the following canonical quantization condition @xmath55 & = i\hbar\delta_{\lambda}^{\sigma}\delta(s - s')\cdot\mathsf{s}^{-1},\label{eq : quant}\\ { } [ \hat{\mathbf{a}}_{\lambda}(s , t),\,\hat{\mathbf{a}}_{\sigma}(s',t ) ] & = [ \hat{{\cal e}}^{\lambda}(s , t),\,\hat{{\cal e}}^{\sigma}(s',t)]=0.\nonumber \end{aligned}\ ] ] here @xmath56 mean the two transverse directions , and @xmath57 is the cross - sectional area of the ring . since we have reduced the problem into 1-dimension the above quantization condition is consistent with the coulomb gauge condition @xmath51 automatically .
now we write down the field operator @xmath58 as @xmath59.\label{eq : a(s , t)}\ ] ] with the help of the above canonical quantization condition ( [ eq : quant ] ) , we can prove the following bosonic commutation relations ( see appendix [ sec : canonical - quantization - of ] ) , @xmath60 & = [ \hat{a}_{k\lambda}^{\dagger},\,\hat{a}_{q\sigma}^{\dagger}]=0,\nonumber \\ { } [ \hat{a}_{k\lambda},\,\hat{a}_{q\sigma}^{\dagger } ] & = \delta_{kq}\delta_{\lambda\sigma}\cdot\frac{\hbar\cdot|\overline{\mathrm{z}}_{k}|^{-2}}{2\epsilon_{0}vc|k|}.\end{aligned}\ ] ] thus , the normalization constant is taken as @xmath61 , where @xmath62 is the effective volume of the ring , so that @xmath63=\delta_{kq}\delta_{\lambda\sigma}$ ] .
then we obtain the quantized hamiltonian of the em field as @xmath64 this hamiltonian has the same form as that of the em field in the inertial frame , but the dispersion relation @xmath65 is changed due to the rotation [ see eq.([eq : disperse ] ) ] .
* jc - model in the rotating frame * next we study the motion of a charged particle in the em field in the rotating reference frame . to this end
, we start with the invariant lagrangian of a charged particle in the effective curved spacetime @xcite @xmath66^{\frac{1}{2}}+ea_{\mu}\frac{dx^{\mu}}{d\tau},\ ] ] where @xmath67 is the proper time . the hamiltonian description is obtained from the following variation ( see ref.@xcite or appendix [ sec : hamiltonian - description ] ) @xmath68dt.\ ] ] here @xmath69 , @xmath70 , @xmath71 and @xmath72 as a conservative system , the hamiltonian @xmath73 is the functional of @xmath74 and @xmath75 ( @xmath9 ) , and @xmath76 here @xmath77 is the mechanical momentum . replacing @xmath78 and @xmath79 by the canonical momentum @xmath80 in @xmath81 , the hamiltonian of the charged particle is obtained as ( see appendix [ sec : hamiltonian - description ] ) @xmath82-ev^{0}a_{0}.\ ] ] substituting in the metric @xmath13 [ eq.([eq : g^uv ] ) ] , the above hamiltonian becomes @xmath83 where @xmath84 . in the quasi-1d ring
, @xmath85 is always in the transverse direction and perpendicular to the linear velocity @xmath86 , thus @xmath87 .
we consider an atom with the nucleus fixed at a certain position inside the ring cavity . around the nucleus
, the electron is trapped by the central force potential @xmath88 .
thus , in the expansion of the above equation , @xmath89 is exactly the hamiltonian of a hydrogen - like atom plus a perturbation term @xmath90 , which comes from the non - inertial effect of rotation , and the term @xmath91 describes the coupling between the atom and the em field .
treating @xmath92 and @xmath93 as operators , we obtain the energy levels of the hydrogen - like atom .
then we focus on two energy levels @xmath94 and @xmath95 with a dipole transition . neglecting the @xmath96 term , with the help of eq.([eq : a(s , t ) ] )
, the total hamiltonian @xmath97 reads , @xmath98+\sum_{k,\lambda}\hbar\omega_{k}\hat{a}_{k\lambda}^{\dagger}\hat{a}_{k\lambda},\ ] ] where @xmath99 , @xmath100 and @xmath101 $ ] .
the coefficients @xmath102 and @xmath103 are explicitly given as @xmath104^{\frac{1}{2}}e^{iks_{0}},\nonumber \end{aligned}\ ] ] where @xmath105 is the dipole moment @xcite . here
the dipole approximation is applied , and @xmath106 is the position of the atom .
we choose @xmath107 to cancel the phase factors . comparing with the inertial case @xmath108
, we see that @xmath1 and @xmath103 is unchanged , and the contribution of rotation appears in the correction term @xmath109 and the dispersion relation @xmath65 .
we set the two polarization directions to be parallel and vertical to the projection of @xmath110 in the transversal section respectively .
then the vertical optical modes are decoupled with the atom [ see eq.([eq : coupling ] ) ] .
when the ring is not too long , the frequencies of different eigen modes of the em field are well separated from each other , so we only consider the modes nearly resonant with the atom .
but we should notice that , different from the fabry - prot type , the @xmath42 and @xmath43 modes are always nearly degenerate in ring cavity ( unless @xmath111 ) . when @xmath108 , they are exactly degenerated @xmath112 .
that means , the @xmath113 and @xmath114 modes must be considered together .
therefore , by omitting the double creation and annihilation terms , we establish the jc - model of two - level atom in a rotating ring cavity with the jc hamiltonian @xmath115 where the coupling strength @xmath116 is @xmath117^{\frac{1}{2}}.\ ] ] here @xmath118 is the annihilation operator for the modes @xmath119 , and their frequencies are @xmath120 , where we define @xmath121 as the rest frequency of the ring cavity and @xmath122 as the rotation detuning . for simplicity
, hereafter we absorb the phase of @xmath116 into the operators to make @xmath123 and set @xmath124 .
* rabi splitting in the rotating frame * we consider the case that the dipole moment is parallel to the transverse direction , thus @xmath125 and we can choose the polarization direction to satisfy @xmath126 . in this case , the excitation number @xmath127 is conserved , i.e. , [ @xmath128=0 $ ] . the ground state is @xmath129 and the eigen energy is @xmath130 .
but generally we can not give an analytical solution for the whole energy spectrum .
when @xmath131 , we can obtain the eigen energy of the first three excited levels [ fig.[fig-3peak](d ) ] , and they are @xmath132 where @xmath133 ( see appendix [ sec : steady - photon - number ] or ref.@xcite ) . of the two cavity modes .
the two - level atom is resonant with the rest frequency @xmath134 , and we set @xmath135 as the unit .
the other parameters are @xmath136 , @xmath137 , @xmath138 , and ( a ) @xmath139 ( b ) @xmath140 .
( c ) the height of the side peak of @xmath141 at @xmath142 changes with the rotation detuning @xmath143 .
( d ) demonstration of the ground state and the first three excited states . ] therefore , if we use an external input to drive the ring cavity weakly , we predict to see rabi splitting with three resonant peaks @xcite . we consider a probing setup as demonstrated in fig.[fig - setup ] .
an external driving laser is input to drive the @xmath113-mode of the ring cavity .
the photons in the ring cavity can leak into the output fiber , and then be probed by photon detectors .
we use the following master equation to describe the system , @xmath144+\sum_{\alpha=+,-}\frac{\gamma}{2}\big(2\hat{a}_{\alpha}\rho\hat{a}_{\alpha}^{\dagger}-\{\rho,\hat{a}_{\alpha}^{\dagger}\hat{a}_{\alpha}\}\big),\ ] ] where @xmath145 is the driving term . here
we consider the case @xmath134 .
when the driving strength @xmath146 is weak , the cavity modes will not be excited to states with large photon numbers , and we obtain ( see appendix [ sec : steady - photon - number ] ) @xmath147 where @xmath148^{2}+\gamma^{2}\tilde{\omega}^{2},\label{eq : n+-ps}\\ f:= & \gamma^{4}\tilde{\omega}^{2}+8\gamma^{2}[2g^{4}+\tilde{\omega}^{4}-2g^{2}\tilde{\omega}^{2}+\delta^{2}\tilde{\omega}^{2}]\nonumber \\ & + 16\tilde{\omega}^{2}[2g^{2}-\tilde{\omega}^{2}+\delta^{2}]^{2},\nonumber \end{aligned}\ ] ] and @xmath149 .
when @xmath150 , the denominator @xmath151 has three minima around @xmath152 and @xmath153 , which give rise to three peaks in @xmath154 corresponding to the first three excited energy eq.([eq : eigene ] ) .
we can also explicitly see that @xmath151 is symmetric for @xmath155 , i.e. , @xmath156 , but @xmath157 is not . therefore , @xmath158 is symmetric for @xmath155 , but @xmath141 is not .
we plot the steady photon number @xmath154 of the cavity modes in fig.[fig-3peak ] . when there is no rotation , @xmath108 , the steady photon number of the two modes @xmath154 are both symmetric with respect to the driving detuning @xmath155 [ fig.[fig-3peak](a ) ] .
it is worth noticing that when there is a small rotating velocity , the heights of the side peaks of @xmath141 , which correspond to the mode @xmath159 being driven , change sensitively with rotating velocity [ fig.[fig-3peak](b ) ] .
since the positions of the side peaks are @xmath160 , we plot the height of @xmath161 with respect to the rotation detuning @xmath162 around @xmath163 [ fig.[fig-3peak](c ) ] .
when @xmath143 is quite small ( in this example , we have @xmath164 ) , @xmath165 depend linearly on @xmath162 around @xmath139 . from eqs.([eq : n+- ] , [ eq : n+-ps ] ) , we obtain the slope of @xmath166 around @xmath139 as @xmath167 from these results , we see that the sensitivity of this height change with respect to the rotating velocity can be controlled by the coupling strength @xmath116 and the decay rate @xmath168 .
a ring cavity with high quality promises a sensitive measurement . in experiments
, the steady photon number @xmath169 can be measured by the average output photon current . in a probe setup as shown in fig.[fig - setup ] , we have @xmath170 , and the average output photon current is @xmath171 @xcite .
this photon current directly characterizes the steady photon numbers of the cavity modes , and can be measured by the photon detectors .
* summary * in conclusion , we have generally studied the quantum optics for the interaction between light and atom in a non - inertial reference frame by the approach of quantum field theory in curved spacetime . for the two - level atom interacting with the quantized em field in a rotating ring cavity ,
our approach built a microscopic model described by a two - mode jc hamiltonian . based on this generalized jc model ,
our study predicts that that the heights of the side peaks in the rabi splitting show a sensitive linear dependence on the rotating velocity at low speed .
therefore , this model can not only be utilized for hybrid optical gyroscope design , but also provide the future development of quantum gyroscope with a solid physical base to consider the effect of quantum fluctuation .
this work was supported by the national 973-program ( grants nos.2014cb921403 , 2012cb922104 and 2012cb922103 ) , the national natural science foundation of china ( grants nos.11421063 , 11447609 , 11404021 , 11374095 , 11422540 , 11434011 ) , and postdoctoral science foundation of china no.2013m530516 .
s .- w .
li wants to thank prof .
y. li and dr .
l. ge for helpful discussions .
13ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/revmodphys.39.475 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.57.61 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.92.253201 [ * * , ( ) ] link:\doibase 10.1103/physreve.71.056621 [ * * , ( ) ] link:\doibase 10.1103/physreva.75.053807 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.075402 [ * * , ( ) ] @noop _
_ ( , ) @noop _ _ ( , ) link:\doibase 10.1103/physreva.8.2103 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1088/0305 - 4470/5/1/002 [ * * , ( ) ] link:\doibase 10.1103/physreva.42.6873 [ * * , ( ) ] @noop _ _ , ed .
( , ) |
the recent results from rhic ( see e.g. @xcite ) have attracted renewed interest in better understanding the dynamics of particle production , not only in nuclear collisions .
quite different approaches have been used to describe the particle spectra from the nuclear collisions @xcite . the model in ref.@xcite with an educated guess for ugd describes surprisingly well the whole charged particle rapidity distribution by means of gluonic mechanisms only .
such a gluonic mechanism would lead to the identical production of positively and negatively charged hadrons .
the recent results of the brahms experiment @xcite put into question the successful description of ref.@xcite . in the light of this experiment , it becomes obvious that the large rapidity regions have more complicated flavour structure .
i discuss the relation between ugd s in hadrons and the inclusive momentum distribution of particles produced in hadronic collisions .
the results obtained with different ugd s @xcite are shown and compared .
i present first results obtained with unintegrated parton distributions from the solution of the so - called ccfm equation .
at sufficiently high energy the cross section for inclusive gluon production in @xmath2 can be written in terms of the ugd s `` in '' both colliding hadrons @xcite @xmath3 above @xmath4 and @xmath5 are ugd s in hadron @xmath6 and @xmath7 , respectively .
the longitudinal momentum fractions are fixed by kinematics : @xmath8 .
the argument of the running coupling constant is taken as @xmath9 . here
i shall not discuss the distributions of `` produced '' gluons , which can be found in @xcite .
instead i shall discuss what are typical values of @xmath10 and @xmath11 in the jet ( particle ) production .
average value @xmath12 and @xmath13 , shown in fig.[fig : x1_x2 ] , only weakly depend on the model of ugd . for @xmath14 0 at the rhic energy w = 200 gev one tests ugd s at @xmath15 = 10@xmath16 - 10@xmath17 .
when @xmath18 grows one tests more and more asymmetric ( in @xmath10 and @xmath11 ) configurations . for large @xmath18 either @xmath10 is extremely small ( @xmath19 10@xmath20 ) and @xmath21 1 or @xmath22 1 and @xmath11 is extremely small ( @xmath23 10@xmath20 ) .
these are regions of gluon momentum fraction where the ugd s is rather poorly known . the approximation used in obtaining ugd s are valid certainly only for @xmath24 0.1 . in order to extrapolate the gluon distribution to @xmath25 1
i multiply the gluon distributions from the previous section by a factor @xmath26 , where n = 5 - 7 . and @xmath13 for @xmath27 0.5 gev and at w = 200 gev .
[ fig : x1_x2 ] , width=226 ]
in ref.@xcite it was assumed , based on the concept of local parton - hadron duality , that the rapidity distribution of particles is identical to the rapidity distribution of gluons . in the present approach i follow a different approach which makes use of phenomenological fragmentation functions ( ff s ) . for our present exploratory study
it seems sufficient to assume @xmath28 .
this is equivalent to @xmath29 , where @xmath30 and @xmath31 are hadron and gluon pseudorapitity , respectively .
then @xmath32 where the transverse mass @xmath33 . in order to introduce phenomenological
ff s one has to define a new kinematical variable . in accord with @xmath34 and @xmath35 collisions
i define a quantity @xmath36 by the equation @xmath37 .
this leads to the relation @xmath38 where @xmath39 is given in ref.@xcite .
now we can write the single particle distribution in terms of the gluon distribution as follows @xmath40 in the present calculation i shall use only lo ff s from @xcite or @xcite
. 0.2 gev .
the experimental data of the ua5 collaboration are taken from @xcite .
[ fig : eta_glue ] , width=226 ] let us analyze now how the results for pseudorapidity distributions depend on the choice of the ugd . in fig.[fig :
eta_glue ] i compare pseudorapidity distribution of charged pions for different models of ugd s . in this calculation ff from @xcite has been used .
in contrast to ref.@xcite , where the whole pseudorapidity distribution , including fragmentation regions , has been well described in an approach similar to the one presented here , in the present approach pions produced from the fragmentation of gluons in the @xmath41 mechanism populate only midrapidity region , leaving room for other mechanisms in the fragmentation regions .
these mechanisms involve quark / antiquark degrees of freedom or leading protons among others .
this strongly suggests that the agreement of the result of the @xmath41 approach with the phobos distributions @xcite in ref.@xcite in the true fragmentation region is rather due to approximations made in @xcite than due to correctness of the reaction mechanism . in principle , this can be verified experimentally at rhic by measuring the @xmath42 ratio in proton - proton scattering as a function of ( pseudo)rapidity in possibly broad range .
the brahms experiment can do it even with the existing apparatus . in fig.[fig
: pt_glue ] i compare the theoretical transverse momentum distributions of charged pions obtained with different gluon distributions with the ua1 collaboration data @xcite .
the best agreement is obtained with the karzeev - levin gluon distribution .
the distribution with the gbw model is much too steep in comparison to experimental data .
this is probably due to neglecting qcd evolution in @xcite .
, width=226 ]
many unintegrated gluon distributions in the literature are ad hoc parametrizations of different sets of experimental data rather than derived from qcd .
recently jan kwieciski and collaborators @xcite have shown how to solve so - called ccfm equations by introducing unintegrated parton distributions in the space conjugated to the transverse momenta @xcite @xmath43 the approach proposed is very convenient to introduce the nonperturbative effects like internal ( nonperturbative ) transverse momentum distributions of partons in nucleons .
it seems reasonable to include the nonperturbative effects in the factorizable way @xmath44 in the following , for simplicity , i shall use a flavour and @xmath45-independent form factor @xmath46 which describes the nonperturbative effects . in fig.[fig : ccfm ] i show the distribution in pseudorapidity of charged pions calculated with the help of the ccfm parton distributions @xcite and the gaussian form factor ( [ formfactor ] ) with @xmath47 = 0.5 gev@xmath48 , adjusted to roughly describe the ua5 collaboration data . now both gluon - gluon and ( anti)quark - gluon and gluon-(anti)quark fussion processes can be included in one consistent framework .
, width=226 ] as anticipated the missing up to now terms are more important in the fragmentation region , although its contribution in the central rapidity region is not negligible .
more details concerning the calculation will be presented elsewhere @xcite .
i have calculated the inclusive distributions of gluons and associated charged @xmath1 s in the nn collisions through the @xmath49 mechanism in the @xmath50-factorization approach .
the results for several ugd s proposed recently have been compared .
the results , especially @xmath51 distributions , obtained with different models of ugd s differ considerably .
contrary to a recent claim in ref.@xcite , i have found that the gluonic mechanism discussed does not describe the inclusive spectra of charged particles in the fragmentation region , i.e. in the region of large ( pseudo)rapidities for any ugd from the literature .
i think that at presently available energies the gluonic mechanism is not the only one .
some of the missing mechanisms have been estimated in the approach based on parton distributions originating from the solution of the ccfm equation .
it was found that the dominant missing mechanisms are ( anti)quark - gluon and gluon-(anti)quark fusion processes .
the existing ugd s lead to the contributions which almost exhaust the strength at midrapidities and leave room for other mechanisms in the fragmentation regions .
it seems that a measurement at rhic of @xmath0 distributions of particles for different rapidities should be helpful to test unintegrated parton distributions . | the inclusive distributions of gluons and pions for high - energy nn collisions are calculated .
the results for several unintegrated gluon distributions ( ugd s ) from the literature are compared .
we find huge differences in both rapidity and @xmath0 of gluons and @xmath1 s in nn collisions for different models of ugd s .
the karzeev - levin ugd gives good description of momentum distribution of charged hadrons at midrapidities .
we find , however , that the gluonic mechanism discussed does not describe the inclusive spectra of charged particles in the fragmentation region .
some of the missing mechanisms are calculated with the help of unintegrated parton distributions from the solution of the ccfm equation .
pacs : 12.38.bx , 13.85.hd , 13.85.ni |
in the following , we derive hamiltonian ( 1 ) of the main text , which describes the spectrum of a quantum dot formed in a straight , suspended carbon nanotube , in the absence of phonons ( i.e. in a static tube ) . as usual
, we start from a tight - binding description of a graphene sheet , which is then rolled up with the condition of periodic boundary conditions . using the conventions as in weiss _
@xcite , this gives the following hamiltonian for the longitudinal degree of freedom @xmath98 here @xmath8 and @xmath99 represent the coordinate and unit vector in the direction of the tube , @xmath100 is the fermi velocity , @xmath101,@xmath4 , and @xmath5 are pauli matrices in sublattice , valley and spin spaces , respectively .
note that to translate between the convention used here and that used in e.g. ref .
, @xmath102 must be replaced by @xmath103 .
the energy gap between the valence band and the conduction band is @xmath104 , where @xmath105 with @xmath106 being the curvature induced minigap , which is typically of order 10 mev , but proportional to @xmath107 , where @xmath108 is the chiral angle of the tube . for nominally metallic tubes , @xmath109 .
the spin - orbit interaction has two terms , one that is diagonal in sublattice ( the @xmath110 term ) and one which is off - diagonal ( the @xmath111 term ) .
the spin - orbit interaction connects the spin projection along the tube axis to the @xmath112 ( valley flavor ) quantum number through the prefactor @xmath113 .
the microscopic derivation of this hamiltonian can be found in refs . .
finally , the term @xmath114 describes the confining potential of the quantum dot .
it is assumed to be smooth on the atomic scale and hence has no sublattice structure .
furthermore , an applied magnetic field couples to both the spin and orbital degrees of freedom .
the coupling to the orbital degree of freedom appears through the aharonov - bohm flux , which modifies the boundary condition of the circumferential wave vector and hence changes @xmath115 . in total , the hamiltonian due to magnetic field is @xmath116 where @xmath117 .
there are several ways to arrive at the hamiltonian in eq .
( 1 ) in the main text . here
we will assume a hierarchy of energy scales , typical of many experimental realizations of nanotube quantum dots , namely @xmath118 where @xmath119 is the level spacing due to longitudinal quantization and @xmath120 , @xmath121 , and @xmath122 are the energy changes due to the external magnetic field , spin - orbit coupling , and valley mixing , respectively .
this allows us to first solve for the dot wavefunction in absence of these three contributions and then project onto a single longitudinal mode .
in passing we note that in order to get more information about dependence of the orbital magnetic moment and spin - orbit coupling on the number of electrons in the quantum dot , one has to be more precise and use a specific form of the confining potential , e.g. assuming a square well potential , as in refs . .
this was done in refs . , where a method to experimentally extract the two spin - orbit parameters @xmath110 and @xmath111 , as well as @xmath123 , was shown .
now imagine that one has solved for the case without magnetic field , spin - orbit coupling , and mixing between @xmath124 and @xmath125 .
this gives a set of longitudinal wavefunctions , each one four - fold degenerate due to the spin and valley degrees of freedom .
the energy splitting between these shells is @xmath126 .
we label these states by the valley and spin quantum numbers @xmath127 and @xmath128 , respectively , which indicate corresponding eigenvalues under @xmath129 and @xmath130 .
projected onto eigenstates of the spatial coordinates @xmath8 and @xmath131 ( the circumferential coordinate ) , the wave function in the envelope - function representation has the form@xcite @xmath132 where @xmath133 is the wave vector associated with the gap , @xmath134 is a pseudospinor describing the sublattice degrees of freedom , @xmath135 and @xmath136 describe the spin and valley degrees of freedom , respectively , @xmath137 and @xmath138 , and the precise form of the envelope wave function @xmath139 depends on the confining potential .
we can now take matrix elements with respect to @xmath140 and @xmath141 and also a term describing the microscopic disorder that couples valleys : @xmath142 , where @xmath143 is a short - range disorder potential depending on the longitudinal and circumferential coordinate operators .
( note that short - range disorder can be systematically incorporated into the envelope - function description , see e.g. , refs . .
) this procedure will produce a hamiltonian of the form in eq .
( 1 ) in the main text , @xmath144 with @xmath145 here the kets @xmath146 stand for the orbital states with @xmath127 . note that in general , the valley - mixing term can include both @xmath147 and @xmath148 , but an appropriate unitary transformation in the valley space can be used to put it into the form above with real @xmath149 .
next we look at how the vibrations couple to the four states of the quantum dot .
the amplitude of the tube is in terms of the harmonic oscillator raising / lowering operators given by @xmath150 where we focus on a single vibrational mode .
the coupling to the spin is via the change of the tangent direction given by @xmath151 where @xmath152 is perpendicular to the tube and in the plane of the vibration .
the interaction hamiltonian then becomes@xmath153 as above , we now project onto a single longitudinal mode , thus taking matrix elements of @xmath154 in the basis @xmath155 .
such matrix elements involve form factors like @xmath156 at this point we note that the coupling is small for a symmetric dots and even harmonics , because the @xmath157 factors then tend to cancel , see discussion in the main text .
the effective hamiltonian for coupling between the four states of the quantum dot and the vibration now becomes @xmath158 we see from eq .
that the coupling of the vibrations to quantum dot states have different form factors from what one would get by simply setting into eq .
( 1 ) of the main paper .
however , when the energy scales are clearly separated as in the eigenstates @xmath155 are eigenstates of @xmath102 ( which can be seen from ) and therefore @xmath159 ( for the conduction band ) . therefore
, we do not need to take into account the different form factors in , which simplifies the analysis and we can write @xmath160 . in this language eq
. becomes @xmath161 which is the result used in the main text . as mentioned the expression
was derived under the assumption that the gap dominates over the longitudinal size quantization energy , which is valid for few - electron quantum dot .
however , it is important to note that one could easily extend this to the more general case at higher energies by including the difference in form factors @xmath162 and @xmath163 without changing the conclusions and structure of our results qualitatively .
we treat three different qubit realizations in the main text : the spin qubit ( s ) , the kramers qubit in a magnetic field perpendicular to the carbon nanotube ( cnt ) ( kx ) , and the kramers qubit in a parallel - to - cnt magnetic field ( kz ) .
below , we express the three qubit hamiltonians @xmath164 , @xmath165 and @xmath166 as functions of system parameters and cnt deformation . to clarify the correspondence with the jaynes - cummings hamiltonian in eq .
( 2 ) of the main text , we list the formulas for the qubit frequency , the qubit - phonon coupling , as well as numerical estimates for the latter , in table [ tab : jcm - correspondence ] . at the finite value @xmath167 of a longitudinally - applied magnetic field ,
the quantum dot ( qd ) energy spectrum shows a crossing of the energies of a pair of spin states belonging to the same valley ( see fig . 2 of the main text ) .
around this point these two levels are energetically well separated from the other qd levels .
we call this two - level system the spin qubit ( s ) .
if the dynamics is restricted to these two levels , it can be described by the following effective hamiltonian : @xmath168 where we assume that the effect of valley - mixing is negligible , @xmath169 .
averaging over the @xmath8 coordinate using the charge density @xmath170 of the electron occupying the cnt qd yields @xmath171 at zero magnetic field , the ground state of the cnt qd , i.e. , the ground state of the hamiltonian @xmath172 in eq .
( 1 ) of the main text , is formed by a pair of time - reversed states ( kramers pair ) .
the twofold degeneracy is maintained even in the presence of spin - orbit interaction and valley mixing . at small enough magnetic field
these two states split up , but they remain energetically well separated from higher - lying states .
we call this two - level system the ` kramers qubit ' @xcite in a perpendicular field ( kx ) .
[ similar considerations hold for the first excited kramers pair , i.e. , the two higher - lying energy eigenstates of @xmath172 in eq .
( 1 ) of the main text . ] these energetically split states , in the absence of cnt deformation , will be denoted here as @xmath173 and @xmath174 .
starting from the hamiltonian @xmath172 in eq .
( 1 ) of the main text , averaging over @xmath8 using the electron density @xmath170 , and incorporating the effect of the two higher - lying states on @xmath173 and @xmath174 via a second - order schrieffer - wolff transformation , we find that the dynamics restricted to the kramers qubit in the presence of an external magnetic field @xmath175 and cnt deformation @xmath13 is described by the hamiltonian @xmath176.\ ] ] here , @xmath23 are the pauli matrices in the qubit basis , i.e. , @xmath177 and @xmath178 . [ cols="<,^,^,^",options="header " , ] in the absence of cnt deformation , a parallel - to - cnt magnetic field splits two low - energy kramers doublet of @xmath172 in eq .
( 1 ) of the main text .
we call this two - level system the kz qubit , and denote the two qubit states as @xmath173 and @xmath174 in this subsection . starting from the complete hamiltonian @xmath172 in eq .
( 1 ) of the main text , averaging over @xmath8 using the electron density @xmath170 , and applying a second - order schrieffer - wolff transformation to describe the effect of the higher - lying kramers pair to the kz qubit , we find that the dynamics of the latter in the presence of an external magnetic field @xmath179 and cnt deformation @xmath13 is described by the hamiltonian @xmath180.\ ] ] as before , @xmath23 are the pauli matrices in the qubit basis , i.e. , @xmath177 and @xmath178 .
all three qubit - phonon hamiltonians @xmath164 , @xmath165 and @xmath166 resemble the jaynes - cummings hamiltonian of cavity quantum electrodynamics . this becomes more apparent if we express the @xmath8-dependent displacement @xmath13 in terms of phonon annihilation @xmath18 and creation @xmath181 operators : @xmath182 here @xmath20 is a dimensionless function describing the shape of the standing - wave bending phonon mode under consideration ( normalization : @xmath183 ) and @xmath21 is the ground - state displacement of that mode . as apparent from table [ tab : jcm - correspondence ] , the qubit - phonon coupling vanishes if @xmath184 .
therefore , @xmath26 vanishes if the setup is perfectly left - right symmetric along the cnt ( @xmath8 ) axis and a bending mode with an even number of nodes is considered . to have a finite qubit - phonon coupling , either the left - right symmetry of the setup
must be broken or a flexural mode with odd number of nodes should be considered . in the main text and also here we treat the second case : we investigate the coupling of the first excited flexural phonon ( 1 node ) to the various qubits . the displacement field of the first harmonic of the resonator can be approximated by @xmath185,\ ] ] where we assume that the cnt is suspended at points @xmath186 and @xmath187 . approximating the charge density with a step function symmetrically covering the length fraction @xmath188 of the suspended part of the cnt
, we obtain @xmath189 = \frac{2 \sqrt 2 } l \frac{\sin \pi \xi } \xi\ ] ] the fraction @xmath190 is @xmath191 if @xmath188 is smaller than 1 , and therefore we make the approximation @xmath192 in the numerical estimates appearing in the main text and in table [ tab : jcm - correspondence ] .
in this section we develop semiclassical equations of motion for the coupled qubit - oscillator system , valid in the regime of large oscillator excitation .
we follow the procedure of ref . , this time allowing for different qubit ( @xmath43 ) , oscillator ( @xmath193 ) , and drive ( @xmath24 ) frequencies .
our aim will be to find a closed set of equations for the time dependence of the expectation values of the oscillator and qubit coordinates , @xmath194 , @xmath195 , and @xmath196 .
we evaluate the time derivatives of these observables using @xmath197 $ ] , with the time - dependence of the density matrix given by eq.(4 ) of the main text ( below we set @xmath198 ) , @xmath199 + ( n_b + 1 ) \gamma \left ( a\rho{a^{\dagger}}- \frac{1}{2}\{{a^{\dagger}}a,\rho\}\right ) \nonumber\\ + n_b \gamma \left ( { a^{\dagger}}\rho a - \frac{1}{2}\{a{a^{\dagger}},\rho\}\right ) , \label{eqn : applindblad_terms}\ ] ] with @xmath200 and @xmath201 . here , in the rotating frame , the reduced frequencies are given by @xmath202 .
the qubit - phonon coupling is denoted by @xmath26 , and the strength of the external driving field is denoted by @xmath25 . using the commutation rule @xmath203 and the cyclic property of the trace , @xmath204 = { \rm tr}[ba]$ ] , we find : @xmath205 we close this set of equations by neglecting correlated fluctuations between the qubit and oscillator degrees of freedom , factoring the averages as @xmath206 and @xmath207 .
using @xmath208 and @xmath209 , and as in ref . defining the complex variables @xmath210 and @xmath211 , and a real variable @xmath212 , we obtain @xmath213 in this representation , the real and imaginary parts of @xmath8 describe the oscillator coordinate and momentum , the complex variable @xmath214 describes the @xmath12 and @xmath215 bloch vector components of the qubit state , and @xmath216 describes the qubit polarization .
the dynamics described by the nonlinear system in eqs .
can be quite complex .
here we focus on steady state solutions , @xmath217 . setting @xmath218 in eq . , and introducing over - bars to indicate steady state values , we obtain @xmath219 note that @xmath220 is automatically satisfied under this condition , see eq . . within the semiclassical description , and in the absence of decoherence
acting directly on the spin , the variables @xmath214 and @xmath216 describe a vector of unit length , @xmath221 . using eq . for @xmath222 , we find @xmath223 .
taking the square root of both sides gives @xmath224 where the subscript @xmath225 indicates two branches of solutions to the square root .
setting @xmath226 in eq . , and using eqs . and , we find that the oscillator amplitude in the steady state satisfies the relation @xmath227 self - consistent solutions to eq . can easily be found numerically . in many parameter regimes ,
multiple solutions exist due to the non - linearity introduced by the qubit - oscillator coupling .
as shown in fig.3e of the main text , the branches of stable fixed points match well with the peaks in the phonon number distribution , indicating the utility of the semi - classical approach . along with the presence of multiple solutions
, we expected the typical manifestations of multistable behavior , such as hysteresis and sharp instabilities .
the sensitivity to the steady state oscillator amplitude near instability points where stable steady - state solutions disappear may be useful for sensing applications .
closely related behavior has already proved quite useful in providing a sensitive read - out mechanism for superconducting qubits.@xcite
as stated in the main text , the oscillatory motion of the cnt resonator can be detected using a charge sensing scheme in which the conductance of a mesoscopic conductor , such as a qd or quantum point contact , is modulated via capacitive coupling to the charged mechanical resonator . at a given source - drain bias on the mesoscopic conductor ,
the current depends on the displacement @xmath228 of the resonator due to their capacitive coupling .
a nonlinear dependence of the current on the resonator displacement is desired for displacement sensing , @xmath229 .
such a dependence is present in , e.g. , a qd tuned to the middle of a coulomb - blockade peak ( corresponds to @xmath230 and @xmath231 ) , or to the onset of a coulomb - blockade peak ( @xmath232 and @xmath233 ) .
here we show that under such conditions , and neglecting detector back - action on the oscillator , the steady - state time - averaged current @xmath234 through the charge - sensing mesoscopic conductor is sensitive to the steady - state average number @xmath235 of phonons in the oscillator .
therefore , the results plotted in fig .
3b , d , g , h of the main text can be interpreted as being proportional to the measured signal @xmath236 in the charge - sensing setup described above , hence that setup would allow for the experimental confirmation of the predicted features .
the steady - state time - averaged current @xmath234 through the charge - sensing conductor is given by @xmath237 the instantaneous square displacement is expressed with the steady - state density matrix @xmath71 and the position operator @xmath238 , both represented in the rotating frame , as @xmath239 ^ 2\right\}\ ] ] substitution of this expression to eq .
yields @xmath240 as stated above . | we theoretically investigate the deflection - induced coupling of an electron spin to vibrational motion due to spin - orbit coupling in suspended carbon nanotube quantum dots .
our estimates indicate that , with current capabilities , a quantum dot with an odd number of electrons can serve as a realization of the jaynes - cummings model of quantum electrodynamics in the strong - coupling regime .
a quantized flexural mode of the suspended tube plays the role of the optical mode and we identify two distinct two - level subspaces , at small and large magnetic field , which can be used as qubits in this setup .
the strong intrinsic spin - mechanical coupling allows for detection , as well as manipulation of the spin qubit , and may yield enhanced performance of nanotubes in sensing applications .
recent experiments in nanomechanics have reached the ultimate quantum limit by cooling a nanomechanical system close to its ground state @xcite . among the variety of available nanomechanical systems , nanostructures made out of atomically - thin carbon - based materials such as graphene and carbon nanotubes ( cnts ) stand out due to their low masses and high stiffnesses .
these properties give rise to high oscillation frequencies , potentially enabling near ground - state cooling using conventional cryogenics , and large zero - point motion , which improves the ease of detection @xcite .
recently , a high quality - factor suspended cnt resonator was used to demonstrate strong coupling between nanomechanical motion and single - charge tunneling through a quantum dot ( qd ) defined in the cnt @xcite . here
, we theoretically investigate the coupling of a single electron spin to the quantized motion of a discrete flexural mode of a suspended cnt ( see fig.[fig : setup ] ) , and show that the strong - coupling regime of this jaynes - cummings - type system is within reach .
this coupling provides means for electrical manipulation of the electron spin via microwave irradiation , and leads to strong nonlinearities in the cnt s mechanical response which may potentially be used for enhanced functionality in sensing applications @xcite .
in addition to their outstanding mechanical properties , carbon - based systems also possess many attractive characteristics for information processing applications .
the potential for single electron spins in qds to serve as the elementary qubits for quantum information processing @xcite is currently being investigated in a variety of systems . in many materials , such as gaas , the hyperfine interaction between electron and nuclear spins is the primary source of electron spin decoherence which limits qubit performance ( see e.g. , @xcite ) .
however , carbon - based structures can be grown using starting materials isotopically - enriched in @xmath0c , which has no net nuclear spin , thus practically eliminating the hyperfine mechanism of decoherence @xcite , leaving behind only a spin - orbit contribution @xcite . furthermore , while the phonon continuum in bulk materials provides the primary bath enabling spin relaxation , the discretized phonon spectrum of a suspended cnt can be engineered to have an extremely low density of states at the qubit ( spin ) energy splitting .
thus very long spin lifetimes are expected off - resonance @xcite . on the other hand ,
when the spin splitting is nearly resonant with one of the high - q discrete phonon `` cavity '' modes , strong spin - phonon coupling can enable qubit control , information transfer , or the preparation of entangled states .
the interaction between nanomechanical resonators and single spins was recently detected @xcite , and has been theoretically investigated @xcite for cases where the spin - resonator coupling arises from the relative motion of the spin and a source of local magnetic field gradients .
such coupling is achieved , e.g. , using a magnetic tip on a vibrating cantilever which can be positioned close to an isolated spin fixed to a nonmoving substrate .
creating strong , well - controlled , local gradients remains challenging for such setups .
in contrast , as we now describe , in cnts the spin - mechanical coupling is _ intrinsic _ , supplied by the inherent strong spin - orbit coupling @xcite which was recently discovered by kuemmeth
_ et al .
_ @xcite .
consider an electron localized in a suspended cnt quantum dot ( see fig . [
fig : setup ] ) .
below we focus on the case of a single electron , but expect the qualitative features to be valid for any odd occupancy ( see ref .
@xcite ) .
we work in the experimentally - relevant parameter regime where the spin - orbit and orbital - zeeman couplings are small compared with the nanotube bandgap and the energy of the longitudinal motion in the qd .
here , the longitudinal and sublattice orbital degrees of freedom are effectively frozen out , leaving behind a nominally four - fold degenerate low - energy subspace associated with the remaining spin and valley degrees of freedom ( see refs .
@xcite ) .
a simple model describing the spin and valley dynamics in this low - energy qd subspace , incorporating the coupling of electron spin to deflections associated with the flexural modes of the cnt @xcite , was introduced in ref .
@xcite . in principle , the deformation - potential spin - phonon coupling mechanism @xcite is also present .
the deflection coupling mechanism is expected to dominate at long phonon wavelengths , while the deformation - potential coupling should dominate at short wavelengths ( see discussion in @xcite ) .
for simplicity we consider only the deflection coupling mechanism , but note that the approach can readily be extended to include both effects .
the hamiltonian describing this system is @xcite @xmath1 where @xmath2 and @xmath3 denote the spin - orbit and intervalley couplings , @xmath4 and @xmath5 are the pauli matrices in valley and spin space ( the pseudospin is frozen out for the states localized in a qd ) , @xmath6 is the tangent vector along the cnt axis , and @xmath7 denotes the magnetic field .
note that the spin - orbit coupling has contributions which are diagonal and off - diagonal in sublattice space @xcite .
when projected onto to a single longitudinal mode of the quantum dot , the effective hamiltonian given above describes the coupling of the spin to the nanotube deflection at the location of the dot @xcite . for a nominally straight cnt
we take @xmath6 pointing along the @xmath8 direction , giving @xmath9 and @xmath10 . here
we find the low - energy spectrum shown in fig .
[ fig : qubit_spectrum ] .
the two boxed regions indicate two different two - level systems that can be envisioned as qubit implementations in this setup : we define a spin qubit @xcite ( s ) at strong longitudinal magnetic field , near the value @xmath11 of the upper level crossing , and a mixed spin - valley or kramers ( k ) qubit @xcite , which can be operated at low fields applied either in the longitudinal ( kz ) or perpendicular ( kx ) directions .
we now study how these qubits couple to the quantized mechanical motion of the cnt . for simplicity
we consider only a single polarization of flexural motion ( along the @xmath12-direction ) , assuming that the two - fold degeneracy is broken , e.g. , by an external electric field .
a generalization to two modes is straightforward .
a generic deformation of the cnt with deflection @xmath13 makes the tangent vector @xmath14 coordinate - dependent .
expanding @xmath14 for small deflections , we rewrite the coupling terms in hamiltonian ( [ eqn : h0 ] ) as @xmath15 and @xmath16 . expressing the deflection @xmath13 in terms of the creation and annihilation operators @xmath17 and @xmath18 for a quantized flexural phonon mode , @xmath19 , where @xmath20 and @xmath21 are the waveform and zero - point amplitude of the phonon mode , we find that each of the three qubit types ( s , kx , kz ) obtains a coupling to the oscillator mode which we describe as @xmath22 here the matrices @xmath23 are pauli matrices acting on the two - level qubit subspace , and we have included a term describing external driving of the oscillator with frequency @xmath24 and coupling strength @xmath25 , which can be achieved by coupling to the ac electric field of a nearby antenna @xcite .
below we describe the dependence of the qubit - oscillator coupling @xmath26 on system parameters for each qubit type ( s , kx , or kz ) . the derivation of eq .
( [ eq : jcm-1 ] ) is detailed in @xcite .
for the spin qubit ( s ) , the relevant two - fold degree of freedom is the spin of the electron itself .
therefore in eq.([eq : jcm-1 ] ) we have @xmath27 and @xmath28 , and the qubit levels are split by the zeeman energy , measured relative to the value @xmath11 where the spin - orbit - split levels cross , @xmath29
. a spin magnetic moment of @xmath30 is assumed , and @xmath31 for @xmath32 .
for the qubit - resonator coupling , we find @xmath33 , independent of @xmath34 . here
, @xmath35 is the derivative of the waveform of the phonon mode averaged against the electron density profile in the qd . for a symmetric qd , positioned at the midpoint of the cnt , the coupling matrix element proportional to @xmath35 vanishes for the fundamental and all even harmonics ( the opposite would be true for the deformation - potential coupling mechanism ) .
the cancellation is avoided for a qd positioned away from the symmetry point of the cnt , or for coupling to odd harmonics . here ,
for concreteness , we consider coupling of a symmetric qd to the first vibrational harmonic of the cnt .
using realistic parameter values @xcite , @xmath36 nm , @xmath37 pm , @xmath38ev , @xmath39ev , @xmath40 , and @xmath41 mhz , we find @xmath42 mhz , irrespective of the magnetic field strength @xmath34 along the cnt . for the kramers qubits ( kx and kz ) ,
both @xmath43 and @xmath26 depend on @xmath34 .
the qubit splitting for the kx qubit is controlled by the perpendicular field , @xmath44 , while for the kz qubit , it is controlled by the longitudinal field @xmath45 , where @xmath46 denotes the zero - field splitting between the two kramers pairs .
resonant coupling occurs when @xmath47 .
this condition sets the relevant value of @xmath48 ( @xmath49 ) in the case of the kx ( kz ) qubit ; the parameters above yield @xmath50 ( @xmath51 ) .
the qubit - cavity coupling for the kx qubit increases linearly with the applied perpendicular field , @xmath52 , while for the kz qubit it scales with the longitudinal field , @xmath53 .
using the values of @xmath48 and @xmath49 obtained above , we estimate couplings of @xmath54 for the kx qubit , and @xmath55 for the kz qubit .
thus the coupling for the kx qubit is comparable to that of the spin qubit , while the coupling of the kz qubit is much weaker .
therefore , we restrict our considerations to the spin and kx qubits below . ref .
reports the fabrication of cnt resonators with quality factors @xmath56 .
we take @xmath57 for the following estimate .
together with the oscillator frequency @xmath58 , this value of @xmath59 implies an oscillator damping rate of @xmath60 . because of the near - zero density of states of other phonon modes at @xmath43 , it is reasonable to assume a very low spontaneous qubit relaxation rate @xmath61 .
these observations suggest that the so - called `` strong coupling '' regime of qubit - oscillator interaction , defined as @xmath62 , can be reached with cnt resonators .
to quantify the system s response in the anticipated parameter regime , we study the coupled qubit - oscillator dynamics using a master equation which takes into account the finite lifetime of the phonon mode as well as the non - zero temperature of the external phonon bath . for weak
driving , @xmath63 , and @xmath64 , we move to a rotating frame and use the rotating wave approximation ( rwa ) to map the hamiltonian , eq.([eq : jcm-1 ] ) , into jaynes - cummings form @xcite @xmath65 where @xmath66 .
including the nonunitary dynamics associated with the phonon - bath coupling , the master equation for the qubit - oscillator density matrix @xmath67 reads : @xmath68 & + & ( n_b + 1 ) \gamma\ , \big
( a\rho{a^{\dagger}}- \frac{1}{2}\{{a^{\dagger}}a,\rho\}\big ) \nonumber\\ & + & n_b \gamma\ , \big ( { a^{\dagger}}\rho a - \frac{1}{2}\{a{a^{\dagger}},\rho\}\big ) , \label{eqn : lindblad_terms}\end{aligned}\ ] ] where @xmath69 is the bath - mode bose - einstein occupation factor , and @xmath70 is the boltzmann constant . because of the phonon damping , in the long - time limit the system is expected to tend towards a steady state , described by the density matrix @xmath71 .
we study these steady states , found by setting @xmath72 in eq.([eqn : lindblad_terms ] ) , using both numerical and semiclassical analytical methods . in figs .
[ fig : results]a , c we show the steady - state phonon occupation probability distribution @xmath73 as a function of the drive frequency phonon frequency detuning @xmath74 and the phonon occupation number @xmath75 , for the case where the qubit and oscillator frequencies are fixed and degenerate , @xmath47 ( see caption for parameter values ) . panels
a and c compare the cases with and without qubit - oscillator coupling . in figs .
[ fig : results]b and [ fig : results]d we show the averaged phonon occupation number @xmath76 , which is closely related to the mean squared resonator displacement in the steady state : @xmath77 . for @xmath78 ,
we observe a splitting of the oscillator resonance , which is characteristic of the coupling to the two - level system , and can serve as an experimental signature of the qubit - oscillator coupling . for drive frequencies near the split peaks ,
the phonon number distribution is bimodal ( fig .
[ fig : results]f ) showing peaks at @xmath79 and at high-@xmath75 , indicating bistable behavior ( see below ) .
for strong excitation , where the mean phonon occupation is large , we expect a semiclassical approach to capture the main features of the system s dynamics @xcite .
extending the approach described in @xcite to include distinct values of the qubit , oscillator , and drive frequencies , @xmath80 , and @xmath24 , we derive semiclassical equations of motion for the mean spin and oscillator variables ( see @xcite ) .
the steady - state values of the mean squared oscillator amplitude obtained from the resulting nonlinear system are shown in fig .
[ fig : results]e . in the vicinity of the split peak
we find two branches of stable steady - state solutions , indicative of bistable / hysteretic behavior @xcite .
the semiclassical results in fig . [
fig : results]e are in correspondence with the phonon number distribution in fig .
[ fig : results]c , and explain its bimodal character .
similar oscillator instabilities have been used as the basis for a sensitive readout scheme in superconducting qubits @xcite , and may potentially be useful for mass or magnetic field sensing applications where small changes of frequency need to be detected . to predict the oscillator response to be detected via a charge sensor ( see below ) , we solve for the stationary state of eq .
( [ eqn : lindblad_terms ] ) directly for a range of driving frequencies , qubit - oscillator detunings ( set by the magnetic field ) , and temperatures @xmath81 . in figs .
[ fig : results](g ) and [ fig : results](h ) , we show the @xmath82 and @xmath83 mk root mean squared oscillator amplitude @xmath84 as function of magnetic field @xmath34 and drive frequency , for the case of a spin ( s ) qubit .
the value @xmath85 corresponds to resonant coupling @xmath47 .
these results also apply for the kx qubit , if the magnetic field axis is adjusted appropriately . in the zero - temperature case , only half of the eigenstates @xmath86 of eq .
( [ eq : jcm ] ) can be efficiently excited by the drive at fixed @xmath87 , giving rise to the upper ( lower ) feature in fig .
[ fig : results]g for @xmath88 ( @xmath89 ) .
however , for @xmath90 , both branches of the jaynes - cummings ladder can be efficiently excited ( fig .
[ fig : results]h ) .
this is a distinct and experimentally accessible signature of the strong coupling at finite temperature .
note that the vacuum rabi splitting is also observed ( see arrows in fig .
[ fig : results]d ) , but features arising from nonlinearity in the strongly driven system dominate by more than 2 orders of magnitude .
displacement detection of nanomechanical systems is possible using charge sensing @xcite , where the conductance of a mesoscopic conductor , such as a qd or quantum point contact , is modulated via capacitve coupling to the charged mechanical resonator .
furthermore , the qubit state itself can be read out using spin - detection schemes developed for semiconductor qds @xcite , or by a dispersive readout scheme like that commonly used in superconducting qubits coupled to microwave resonators @xcite .
the dispersive regime can be rapidly accessed by , e.g. , tuning the resonator frequency using dc gate pulses which control the tension in the cnt @xcite . in summary
, we predict that strong qubit - resonator coupling can be realized in suspended cnt qds with current state - of - the - art devices .
the coupling described here may find use in sensing applications , and in spin - based quantum information processing , where the cnt oscillator enables electrical control of the electron spin , and , with capacitive couplers , may provide long - range interactions between distant electronic qubits @xcite .
combined with control of the qubit via electron - spin - resonance @xcite , the mechanism studied here could be utilized for ground - state cooling and for generating arbitrary motional quantum states of the oscillator @xcite .
we gratefully acknowledge helpful discussions with h. carmichael , v. manucharyan and p. rabl .
this work was supported by the otka grant pd 100373 , the marie curie grant cig-293834 and the qspice program of esf ( ap ) , dfg under the programs for 912 and sfb 767 ( ps and gb ) , nsf grants dmr-090647 and phy-0646094 ( mr ) , and the danish council for independent research | natural sciences ( kf ) .
_ note : _ while completing this manuscript , we became aware of a related work @xcite that describes the theory of the spin - phonon coupling in a cnt resonator qd , and its consequences in the spin blockade transport setup .
99 a. d. oconnell , m. hofheinz , m. ansmann , r. c. bialczak , m. lenander , erik lucero , m. neeley , d. sank , h. wang , m. weides , j. wenner , j. m. martinis , and a. n. cleland , nature * 464 * , 697 ( 2010 ) .
h. b. peng , c. w. chang , s. aloni , t. d. yuzvinsky , and a. zettl , phys .
rev .
lett . * 97 * , 087203 ( 2006 ) . c. chen , s. rosenblatt , k. i. bolotin , w. kalb , p. kim , i. kymissis , h. l. stormer , t. f. heinz , and james hone , nature nanotechnology * 4 * , 861 ( 2009 ) .
g. a. steele , a. k. httel , b. witkamp , m. poot , h. b. meerwaldt , l. p. kouwenhoven and h. s. j. van der zant , science * 325 * , 1103 ( 2009 ) .
h .- y .
chiu , p. hung , h. w. ch .
postma , and m. bockrath , nano lett . * 8 * , 4342 ( 2008 ) .
b. lassagne , d. garcia - sanchez , a. aguasca , and a. bachtold , nano lett . * 8 * , 3735 ( 2008 ) .
b. lassagne , d. ugnati , and m. respaud , phys .
rev .
lett .
* 107 * , 130801 ( 2011 ) .
d. loss and d. p. divincenzo , phys .
rev . a * 57 * , 120 ( 1998 ) .
j. r. petta , a. c. johnson , j. m. taylor , e. a. laird , a. yacoby , m. d. lukin , c. m. marcus , m. p. hanson , and a. c. gossard science * 309 * , 2180 ( 2005 )
. b. trauzettel , d. v. bulaev , d. loss , and g. burkard , nature phys . * 3 * , 192 ( 2007 ) .
d. v. bulaev , b. trauzettel , and d. loss , phys .
rev .
b * 77 * , 235301 ( 2008 ) .
p. r. struck and g. burkard , phys .
rev .
b * 82 * , 125401 ( 2010 ) .
a. cottet and t. kontos , phys .
rev .
lett . * 105 * , 160502 ( 2010 ) .
d. rugar , r. budakian , h. j. mamin and b. w. chui , nature * 430 * , 329 ( 2004 ) .
p. rabl , p. cappellaro , m. v. gurudev dutt , l. jiang , j. r. maze and m. d. lukin , phys .
rev .
b * 79 * , 041302(r ) ( 2009 ) .
p. rabl , s. j. kolkowitz , f. h. l. koppens , j. g. e. harris , p. zoller and m. d. lukin , nat . phys .
* 6 * , 602 ( 2010 ) .
t. ando , j. phys .
. jpn . * 69 * 1757 ( 2000 ) .
j. s. jeong and h. w. lee , phys .
rev .
b * 80 * , 075409 ( 2009 ) .
w. izumida , k. sato , and r. saito , j. phys .
. jpn . * 78 * , 074707 ( 2009 ) .
j. klinovaja , m. j. schmidt , b. braunecker and d. loss , phys .
rev .
lett . * 106 * , 156809 ( 2011 ) .
f. kuemmeth , s. ilani , d. c. ralph , p. l. mceuen , nature * 452 * , 448 ( 2008 ) .
t. s. jespersen , k. grove - rasmussen , j. paaske , k. muraki , t. fujisawa , j. nygard , and k. flensberg , nat . phys .
* 7 * , 348 ( 2011 ) .
a. plyi and g. burkard , phys .
rev .
lett . * 106 * , 086801 ( 2011 ) .
see supplemental material below .
g. d. mahan , phys .
rev .
b * 65 * , 235402 ( 2002 ) . e. mariani and f. von oppen , phys .
rev .
b * 80 * , 155411 ( 2009 ) .
m. s. rudner and e. i. rashba , phys .
rev .
b * 81 * , 125426 ( 2010 ) .
k. flensberg and c. m. marcus , phys .
rev .
b * 81 * , 195418 ( 2010 ) .
a. k. huttel , g. a. steele , b. witkamp , m. poot , l. p. kouwenhoven and h. s. j. van der zant , nano lett .
* 9 * , 2547 ( 2009 ) .
h. o. h. churchill , f. kuemmeth , j. w. harlow , a. j. bestwick , e. i. rashba , k. flensberg , c. h. stwertka , t. taychatanapat , s. k. watson , and c. m. marcus , phys .
rev .
lett . * 102 * , 166802 ( 2009 ) .
e. t. jaynes and f. w. cummings , proceedings of the ieee * 51 * , 89 ( 1963 ) .
p. alsing and h. j. carmichael , quantum opt .
* 3 * , 13 ( 1991 ) .
l. s. bishop , e. ginossar , and s. m. girvin , phys .
rev .
lett . * 105 * , 100505 ( 2010 ) .
m. d. reed , l. dicarlo , b. r. johnson , l. sun , d. i. schuster , l. frunzio , and r. j. schoelkopf , phys .
rev .
lett . * 105 * , 173601 ( 2010 ) .
r. g. knobel and a. n. cleland , nature * 424 * , 291 ( 2003 ) .
see e.g. , j. m. elzerman , r. hanson , l. h. willems van beveren , b. witkamp , l. m. k. vandersypen , and l. p. kouwenhoven , nature * 430 * , 431 ( 2004 ) .
a. wallraff , d. i. schuster , a. blais , l. frunzio , j. majer , m. h. devoret , s. m. girvin , and r. j. schoelkopf , phys .
rev .
lett . * 95 * , 060501 ( 2005 ) .
a. imamoglu , d. d. awschalom , g. burkard , d. p. divincenzo , d. loss , m. sherwin and a. small , phys .
rev .
lett .
* 83 * , 4204 ( 1999 ) .
f. h. l. koppens , c. buizert , k. j. tielrooij , i. t. vink , k. c. nowack , t. meunier , l. p. kouwenhoven , and l. m. k. vandersypen , nature * 442 * , 766 ( 2006 ) . c. ohm ,
c. stampfer , j. splettstoesser , and m. wegewijs , appl .
. lett . *
100 * , 143103 ( 2012 ) .
andrs plyi,@xmath91 p. r. struck,@xmath92 mark rudner,@xmath93 karsten flensberg,@xmath94 and guido burkard@xmath92 _ @xmath95department of physics , university of konstanz , d-78457 konstanz , germany + @xmath96department of materials physics , etvs university , h-1517 budapest pob 32 , hungary + @xmath93department of physics , harvard university , cambridge , massachusetts 02138 , usa + @xmath97niels bohr institute , university of copenhagen , universitetsparken 5 , dk-2100 copenhagen , denmark _ |
despite the success of the standard model ( sm ) , it is far from complete .
for instance , the origin of the fermion mass hierarchy and issues such as baryon asymmetry are yet to be resolved .
we would expect new physics to contribute to these areas . at the more fundamental level
, nobody knows of any deep explanation as to why there should only be three generations of quarks & leptons . for these reasons , and others , extensions of the three - generation sm
are being investigated . here
, we investigate the possibility of incorporating a fourth - generation into a gut model .
+ as is well known , all fourth - generation models must adhere to certain experimental constraints , the first of which stems from precise measurements of the decay characteristics of the @xmath1-boson performed at lep .
this has set a lower bound of @xmath2 on any non - sm particles that couple to the @xmath1-boson .
ignoring for the moment the unnatural hierarchy emerging within the neutrino sector , we assume a dirac mass @xmath3 for the heavy neutral lepton .
we label the fourth - generation explicitly with the notation @xmath4 + physics beyond the standard model is severely constrained by precision electroweak data , @xmath5 and @xmath6 , @xmath7 so that contributions to the @xmath8 decay can be ignored . ] .
assuming sm contributions , fits to lep data give the radiative correction parameter 100 gev . ]
for a heavy ( @xmath10 ) degenerate fourth - generation we obtain @xmath11 and so is ruled out at @xmath12 @xmath13
. however , maltoni _ et al . _
@xcite have shown that particles with mass @xmath14 give drastically different contributions to @xmath15 .
it is possible for a heavy neutrino @xmath16 with mass @xmath17 to cancel the contributions from the heavy @xmath18 with the sm solution ) denote pole masses and lower case letters ( @xmath19 ) denote running masses . ]
@xmath20 @xmath21 @xmath22 where the @xmath16 must be relatively stable to avoid detection ( _ i.e. _ mixing matrix elements @xmath23 ) . in eq .
( [ def ] ) , @xmath24 is the minimum mass of the b quark as allowed by experimental searches .
an extra generation can be accomodated below the @xmath25 level ( or even two generations at @xmath26 ) 180 gev ; @xmath27 130 gev gives a @xmath28 fit for the fourth - generation as good as that of the three - generation case @xcite .
perturbative unification is not consistent with such large masses . ] .
however , it was shown by gunion _ et al .
_ @xcite that the fourth - generation charged and neutral leptons must be relatively light ( @xmath29 ) in order to stay in the perturbative regime below the gut scale .
although they worked within a supersymmetric framework , similar results are expected to hold in the sm @xcite .
setting @xmath30 to achieve perturbative unification is not consistent with the above sm fits ( eq . ( [ fit ] ) ) to the precision data .
therefore , we require the cancellations to arise from another sector and so we consider a supersymmetric theory . indeed , it is shown in @xcite that light supersymmetric particles can also effect the fit to precision data , allowing for a fourth - generation below the @xmath31 level . in particular , neutralinos ( @xmath32 ) and charginos ( @xmath33 ) with masses less than @xmath34 can provide the correct sign contributions , whilst being consistent with current experimental limits if nearly degenerate @xcite and @xmath33 are defined as the lightest mass eigenstates in the neutralino and chargino sectors respectively . ] .
+ in this paper we investigate the possibility of consistently incorporating a fourth - generation into a @xmath35 r - parity conserving supergravity model .
we assume a structure akin to that of the minimal supersymmetric standard model ( mssm3 ) , adding a complete chiral fourth - generation and its associated susy partners ( the so called mssm4 ) .
specifically we require @xmath36 perturbative values for all yukawa couplings at energies up to the gut scale and @xmath37 gauge coupling unification .
these two constraints will be termed collectively as perturbative unification .
+ having satisfied precision data fits , it remains for us to provide solutions that @xmath36 unify perturbatively at the gut scale and @xmath37 evade the direct experimental searches performed at @xmath38 . in section [ sec2 ]
we investigate the specific decay channels of the fourth - generation quarks and ensure we can provide consistency with the experimental direct searches .
the solution we present requires the introduction of light susy particles ( @xmath39 ) , so that the two - body decays @xmath40 and @xmath41 are kinematically allowed
. these will always dominate over the one - loop fcnc decays @xmath42 and two - generation decays @xmath43 that traditional searches have looked for , whilst at the same time suppress the decay rate @xmath44 .
we note a light , degenerate chargino / neutralino pair is just what is needed to provide the necessary cancellations in precision data . + in section [ sec3 ]
we discuss the influence of specific fourth - generation and susy particle masses on precision data fits in more detail and point out new features that appear when trying to satisfy all constraints . in section [ sec4 ] we proceed with a renormalisation group study of the four generation mssm . under the assumption of a common mass scale for the coloured / weak sparticle spectrum ( @xmath45/@xmath46 respectively ) we will derive upper bounds on @xmath45 .
this bound stems from the fact that if we are to achieve perturbative values for the top quark yukawa coupling constant @xmath47 up to the gut scale , then the effects of the coloured sparticle spectrum must be included in the running of the strong coupling @xmath48 at an early stage .
finally , in section [ sec5 ] we investigate the lightest higgs mass in the mssm4 .
to reduce the number of free parameters we take a common mass scale for the fourth - generation quarks @xmath49 and for the leptons @xmath50 .
it is understood , however , that in reality a small non - degeneracy @xmath51 ; @xmath52 exists as will be emphasized in the following text .
we begin our discussion by considering the leptonic sector . under the assumption that the mixing between ( @xmath53 ) and the first three generations is negligible
, the decay @xmath54 will be dominant .
current experimental limits searching for @xmath54 from @xmath55 production have been performed by the opal / l3 collaborations up to the kinematic limit @xmath56 100 gev @xcite . however , to be consistent with perturbative unification we require @xmath30 .
therefore , in order to evade experimental bounds , the mass difference @xmath57 must be less than @xmath58 5 gev so as to result in a virtual @xmath59 whose decay products are too soft to be triggered . regarding the heavy neutrino , opal / l3 have set the bound @xmath60 70 - 90 gev based on the search for @xmath61 ( @xmath62 or @xmath63 ) provided the mixing matrix elements satisfy @xmath64 @xcite . however , we have already assumed that the mixing of the fourth generation leptonic sector is in fact negligible ( @xmath23 ) ; so the neutrino is stable enough to leave the detector and , in this case , only the delphi bound of @xmath60 45 gev from measurements of the @xmath1-width applies . for the purpose of the numerical study
we take @xmath65 46 gev .
+ turning to the quark sector , the requirement of perturbative unification places strict upper limits on the masses of the @xmath66 and @xmath67 and they certainly must be below the top quark whose mass is @xmath68 5.1 gev .
experimental searches for the fourth - generation quarks have mainly concentrated on the @xmath67-quark where cdf search for @xmath69 the current bound excludes @xmath70 199 gev assuming the branching ratio @xmath71 .
the search is also sensitive to other decay modes as long as @xmath72 is large enough to provide the leptonic trigger .
for instance , the decays @xmath73 and @xmath74 will be triggered , since the hadronic decay of the @xmath75 and @xmath76 is kinematically similar to those of the @xmath1 .
one possible escape might come about if we note that @xmath77 would compete with @xmath78 for @xmath79 ( dominating for larger @xmath80 ) and would provide an invisible signature .
for @xmath79 we would also expect @xmath81 @xcite .
however , cdf still exclude a @xmath67-mass in the range 104 gev @xmath82 152 gev , assuming @xmath83 and no sensitivity to the other decay modes @xcite . for the region @xmath70 104 gev ,
we require @xmath84 30% in order to evade the b search .
this comes from the fact that @xmath85 70% and that we also assume the acceptance for the two processes @xmath86 are equal .
however , this scenario is inconsistent with the latest bounds from lepii @xcite which exclude @xmath87 106 gev if @xmath84 30% . moreover , we have yet to account for the @xmath66 decays which turn out to be highly constrained if we assume @xmath88-like processes .
the channel @xmath44 is prohibited for the obvious reason that the @xmath66-quark would have been picked up in the @xmath38 search for the top quark . although we might assume that @xmath89 is dominant by suppressing @xmath90 we must notice that , since the @xmath91 production cross - section is similar to that of @xmath92 , we would effectively expect double the event rate on the @xmath67-quark search . taking this into account would further strengthen existing bounds on the @xmath67-quark mass .
+ from the ideas presented so far , we might conclude that the fourth - generation with perturbative unification is not consistent with experimental bounds on the ( @xmath93 ) quarks . we have not , however , considered the possibility of light susy particles providing decay channels for ( @xmath93 ) .
in this situation one can constrain the light ( _ i.e. _ @xmath0 ) neutralino / chargino pair , which is already required by fits to precision data , in order to provide the following two - body decays @xmath94 where @xmath95 is the lightest mass eigenstate , defined by the amount of mixing between the l / r weak eigenstates ; @xmath96 . ensuring both decay channels are kinematically accessible , combined with the fact that perturbative unification requires that @xmath97 108 gev is defined as the maximum allowed value of @xmath49 .
this is obtained by ensuring perturbative consistency in renormalization group calculations ( see section [ sec4 ] ) . ]
( see table [ range ] ) , places severe restrictions on the allowed spectrum . in the next section
we combine the experimental constraints with those obtained from precision fits to obtain the allowed fourth - generation spectrum .
.values of @xmath98 vs. @xmath99 .
[ cols="^,^,^,^",options="header " , ]
it is difficult to provide bounds from precision data without a fully consistent study taking into account exact particle masses , any light susy spectra present and mixings between different flavours .
however , maltoni _ et al . _
@xcite pointed out that a highly degenerate neutralino / chargino pair can provide the necessary contributions needed to cancel that of the fourth - generation , whilst at the same time being consistent with lep bounds .
specifically they require @xmath100 where we define the notation @xmath101 ; @xmath102 .
looking at their results , we see that the magnitude of the contribution to the fitted parameters from this sector are highly dependent on @xmath103 .
deviations from @xmath104 60 gev larger than + 30 gev / - 5 gev are ruled out at the @xmath31 level .
this scenario is inconsistent with perturbative unification , since @xmath104 60 gev with @xmath105 would require @xmath106 120 gev in order to retain the decay channels in eq .
( [ chan ] ) . ] .
one possible solution presents itself , however , if we assume that the lightest supersymmetric particle ( lsp ) is in fact the fourth - generation sneutrino @xmath107 which is stable due to r parity . assuming the leptonic mixing angle between the third- and fourth - generation is non - zero , the @xmath32 would decay invisibly via @xmath108 @xcite .
the following masses are possible : where we assume that the mixing angles @xmath109 are such that @xmath110 decouple from the z - boson at tree - level , and therefore do not contribute to its total width .
the fourth - generation bottom squark will decay via @xmath111 .
such a decay involves the factor @xmath112 , leading to a long lifetime .
current bounds looking for hadronizing sbottom quarks exclude the range 5 gev @xmath113 38 gev , if the mixing angle @xmath114 1.17 rad where the production cross - section is minimized @xcite .
we note this corresponds to the mixing angle at which @xmath95 decouples from the z - boson , as required from the z - width constraint .
+ although the sparticle masses seem to be contrived , we note that they can be obtained from reasonable assumptions about the supersymmetric sector . in order to obtain @xmath115 3 gev , we require the hierarchy @xmath116 where @xmath117 is the higgs mixing parameter . @xmath118 and @xmath119 are the electroweak gaugino masses ( for a review of supersymmetry see @xcite ) . this structure can occur naturally within specific supergravity models @xcite .
the final mass eigenstates @xmath120 ( @xmath121 ) depend on the appropriate squark / slepton soft mass terms and scalar tri - linear couplings that are present in the soft supersymmetric breaking sector , and the @xmath122 term . here , we assume it is possible to constrain these free parameters to give the required spectrum without contradicting the model in any way
. however , more detailed studies taking into account the high - energy behaviour and renormalization group improvement needs to be performed , in order to establish consistency with radiative electroweak breaking and other constraints such as the absence of charge / colour breaking _
etc_. + we note that the fits to precision data performed in @xcite constrained the fourth - generation masses to satisfy @xmath123 and @xmath124 , whereas we consider the case @xmath125 . from eqs .
( 5.95.11 , 5.13 ) in @xcite , we can see that as long as we satisfy the conditions @xmath126 , the two scenarios differ only in the universal contributions to the radiative parameters ( see maltoni , ph.d .
thesis @xcite for terminology ) .
in particular , we would require an extra ( positive ) universal contribution in order to exactly reproduce the fits .
this could arise if we were to find large su(2 ) breaking within the @xmath127 sector .
this has been shown to provide the correct type of contribution that is needed to compensate between the two scenarios @xcite .
here we investigate the effect of the fourth - generation on the evolution of couplings to the gut scale , where we require gauge coupling unification .
this places upper limits on the masses of the extra particles to ensure their yukawa couplings run perturbatively to the unification scale @xmath128 ( @xmath129 , @xmath130 ) .
starting at the low - energy scale @xmath34 , the electroweak gauge couplings @xmath131 , @xmath132 are fixed through the relations @xmath133 for @xmath134 respectively .
the strong coupling @xmath135 is taken from the particle data group ( pdg ) @xcite to be @xmath136 .
we take the best fit values for @xmath137/@xmath138 from the pdg . in principle
, one should extract the @xmath1-pole couplings assuming the full mssm4 , thereby accounting for the fourth - generation fermions and light susy spectra in a fully consistent way where both logarithmic and finite corrections will influence the extraction of the couplings . ] .
we have also performed our study in the @xmath139 scheme although it is the @xmath140 scheme that is consistent with supersymmetry .
however , differences between the @xmath140 and @xmath139 schemes are not significant at the low - energy scale as compared to other uncertainties @xcite . + in our analysis we neglect all yukawa couplings from the first three generations , except that of the @xmath141-quark whose mass we take to be @xmath142 170 gev .
as is typical with four - generation models , we require small values of @xmath143 ( the ratio of the higgs vev s ) so as to avoid @xmath144@xmath145(@xmath146 ) @xcite .
once all couplings at @xmath34 have been fixed , we integrate up in energy scale using the two - loop renormalization group equations ( rge ) .
the one - loop leading logarithmic threshold corrections from the susy sector are accounted for in the numerical procedure .
we select the point where @xmath147 as the unification scale @xmath128 with coupling @xmath148 .
any deviation in @xmath149 , which we parameterize as @xmath150 , can arise from either of two sectors .
on the one hand , we have the @xmath139 vs. @xmath140 mismatch , experimental errors in @xmath137/@xmath138 and the variations in the best fit values of @xmath137 and @xmath138 as the fourth - generation and light susy particles ( @xmath151 ) are included .
however , more importantly , assuming no intermediate scales , high - energy threshold corrections from specific gut / string models can provide contributions to @xmath152 .
following @xcite , we note that these corrections ( for particular models ) can be large .
we conservatively assume that unification is consistent if @xmath153 5 %
. + the susy threshold corrections in the mssm4 are important as they can influence whether or not a particular set of parameters ( @xmath154 ) will retain perturbative consistency to the gut scale .
this can be observed analytically if we write the one - loop leading logarithmic correction to the strong coupling from the susy sector @xmath155 here @xmath45 represents an effective threshold scale and @xmath156 represents the one - loop beta function contribution to the strong coupling in the mssm4/sm4 respectively .
this correction is implemented at the scale @xmath34 and accounts for the coloured sparticles with masses @xmath157 in the running of the strong coupling .
@xmath158/@xmath159 represents the renormalized gauge coupling just above / below the scale @xmath34 and @xmath160 is fixed to be 0.118 from experimental input . from eq .
( [ match ] ) we see that the higher the scale @xmath45 , the lower @xmath161 . writing the one - loop rge for the third - generation top yukawa coupling @xmath47 , accounting for the @xmath162 quarks and the strong coupling , we have @xmath163 where @xmath164 and @xmath165 .
we can see that the corrections from the sparticle sector ( see eq .
( [ match ] ) ) are constrained , since we require a large @xmath161 in order to compensate for the contributions from @xmath166 in eq .
( [ rgen ] ) that are driving @xmath47 to non - perturbative values .
in particular , the yukawa couplings @xmath167 are fixed at the low - energy scale through the relations @xmath168 where @xmath169 174 gev is the higgs vacuum expectation value and @xmath170 are the quark running masses at the scale @xmath171 .
for a given @xmath143 , if @xmath172 is sufficiently large so that the rg running of @xmath47 is tending towards non - perturbative values , then we obtain an upper bound @xmath173 , above which the initial @xmath161 is too small to counteract the effect of the fourth - generation couplings when solving the rge in eq .
( [ rgen ] ) . + in practice , we perform a full numerical study , accounting for the threshold corrections from @xmath162 and all sparticles by changing the @xmath174-functions and using the step - function approach in the running of the gauge couplings @xcite .
we assume separate degeneracies amongst the coloured and weak susy spectrum @xmath175 where @xmath176 and @xmath177 denote the leptons and quarks of the first three - generations respectively .
we assume 100 gev @xmath178 2 tev and we restrict the values of @xmath46 to be 500 gev , 1 tev or 2 tev .
of course , considering separate degeneracies amongst the coloured and weak sparticle spectra is an approximation .
in fact , our model demands the introduction of light sparticles to ensure consistency with experimental searches for the fourth - generation and precision data bounds , thus providing significant deviations from degeneracy . nevertheless , looking at figure [ susyr ]
, we can clearly see that the effective mass ( @xmath45 ) of the coloured spectrum is bounded from above , in the process of retaining perturbative consistency .
the effect of @xmath46 is clearly weak since @xmath131 and @xmath132 are an order of magnitude lower than the strong coupling @xmath135 . in particular , although we expect the sparticles to deviate in mass around the effective scales @xmath45 and @xmath46 , we can still conclude that as long as all the sparticles have masses less than 2 tev , then perturbative unification is consistent with @xmath179 ( 170 gev , 106 gev , 46 gev ) for @xmath180 .
in this section , the effect of the fourth - generation on the lightest higgs mass ( @xmath80 ) is investigated .
we shall employ the one - loop effective potential with contributions from the top / stop and fourth - generation fermions / sfermions .
since we are at low @xmath143 , we can ignore the contributions from the third generation bottom / sbottom masses . at tree - level ,
the higgs sector of minimal supersymmetry models is fixed by two parameters @xmath181 and @xmath182 ( the pseudoscalar higgs - boson mass ) .
we set @xmath183 1.64 from the requirement of perturbative unification ( see figure [ susyr ] ) . assuming the pseudoscalar higgs - boson mass ( @xmath184 ) is degenerate with the heavy higgs sector ( @xmath185 ) we set @xmath186 500 gev .
accounting for one - loop terms , the higgs mass @xmath80 becomes dependent on ( @xmath187 ; @xmath188 ; @xmath189 ; @xmath190 ; @xmath191 ) .
in general , the sparticle mass matrix squared in the l / r basis can be written as @xmath192 where @xmath193 , @xmath194 represents the @xmath195-term contributions ; @xmath196 and @xmath197 . since large values of @xmath117 are required by radiative electroweak breaking and also to preserve the hierarchy in eq .
( [ hier ] ) , we take @xmath198 500 gev .
the sparticle mass eigenvalues squared , denoted by @xmath199 , are obtained by diagonalizing the matrix in eq .
( [ mat ] ) .
since the higgs mass is dependent on the mixing in the sparticle sector , it is insufficient to consider a fully degenerate susy spectrum as in the approximation used in section ( [ sec4 ] ) . to account for this
, we set the soft supersymmetric breaking terms for the coloured squarks to be @xmath200 which we vary over the range 100 gev @xmath201 1 tev . for the sleptons we take @xmath202 500 gev .
we then randomly vary the two mixing parameters @xmath203 ; @xmath204 and retain the maximum value returned for the lightest higgs mass @xmath205 . as shown in previous studies of the higgs sector , this will occur for @xmath206 where @xmath207 .
large mixings will induce light sparticles ( @xmath0 ) which are required in the mssm4 to provide the decay channels to the fourth - generation quarks .
the minimum higgs mass is obtained when @xmath208 0 .
+ in the mssm4 , for @xmath209 100 gev , the channel ( @xmath210 ) , where invisible represents ( @xmath211 ) , will open and would dominate over conventional @xmath212 rates .
exclusion limits will now come from the missing energy search ( @xmath213 ) that currently sets the lower bound at 114.4 gev @xcite assuming a sm - like higgs and @xmath214 1 .
we therefore need to check , for our constrained set of parameters ( @xmath187 ; @xmath188 ; @xmath189 ; @xmath190 ; @xmath191 ; @xmath143 ) , that @xmath205 lies above this experimental lower bound .
+ we find a large region 370 gev @xmath201 1000 gev where @xmath215 114.4 gev , although some mixing in the third - generation stop and/or fourth - generation squark sector is required .
the region where @xmath216 , although excluded in the mssm3 scenario from experimental lower limits @xcite , is allowed in the mssm4 due to the extra one - loop contributions that arise from @xmath217 ; @xmath218 ; @xmath219 and @xmath220 .
the lightest higgs - boson would then decay invisibly with an upper bound on its mass of @xmath221 140 gev . by
up to -10 gev in the mssm3 @xcite . ]
we have seen that it is possible to incorporate a fourth - generation into a gut model , requiring the existence of supersymmetric particles in order to provide the necessary cancellations in precision fits .
the fourth - generation masses are then tightly constrained with typical values of @xmath222 106 gev ; @xmath65 46 gev . to retain perturbative consistency to the unification scale , assuming all sparticles have masses less than 2 tev , we constrain @xmath143 to lie in the interval 1.63 @xmath223 1.67 .
we find @xmath224 370 gev represents the scale of the soft breaking parameters in the @xmath225 sector .
] from the requirement that the lightest higgs mass must be above its present experimental lower bound , although some mixing between the l / r weak eigenstates in the third and/or fourth - generation squark sector is required . in order to provide decay channels to the fourth - generation quarks , it might be that the lsp is the sneutrino @xmath226 with a mass @xmath227 40 gev .
this has implications for dark matter constraints .
we would also expect a light @xmath95 with @xmath228 40 - 50 gev along with a degenerate neutralino / chargino pair @xmath229 55 - 65 gev
. the susy spectrum needed to satisfy all these constraints can not be obtained from msugra scenarios with universal parameters at the unification scale .
we would like to thank david sutherland for discussions and useful remarks .
99 m. maltoni , v.a .
novikov , l.b .
okun , a.n .
rozanov , m.i .
vysotsky : extra quark - lepton generations and precision measurements , phys
b476 * , ( 2000 ) , 107115 .
m. maltoni : ph.d .
thesis , hep - ph/0002143 , ch . 5 , ( 2000 ) .
novikov , l.b .
okun , a.n .
rozanov , m.i .
vysotsky : extra generations and discrepancies of electroweak precision data , hep - ph/0111028 .
gunion , d. mckay , h. pois : a minimal four - family supergravity model , phys . rev . *
d53 * , ( 1996 ) , 16161647 .
h. b. nielsen , a. v. novikov , v. a. novikov , m. i. vysotsky : higgs potential bounds on extra quark - lepton generations , phys . lett . * b374 * , ( 1996 ) , 127130 . yu .
f. pirogov , o.v .
zenin : two - loop renormalization group restrictions on the standard model and the fourth chiral family , eur .
j. * c10 * , ( 1999 ) , 629638 .
l3 coll : search for charginos with a small mass difference to the lightest supersymmetric particle at @xmath230 189 gev , phys .
b482 * , ( 2000 ) , 3142 .
opal coll .
, g. abbieudi _
et al _ : search for unstable heavy and excited leptons at lep2 , eur .
j. * c14 * , ( 2000 ) , 7384 .
, p. achard _
et al _ : search for heavy neutral and charged leptons in @xmath231 annihilation at lep , phys .
b517 * , ( 2001 ) , 7585 .
w. s. hou , r. stuart : higgs - boson production from decays of the fourth - generation @xmath67 quark , phys . rev . *
d43 * , ( 1991 ) , 36693682 .
, t. affolder _
et al _ : search for a fourth - generation quark more massive than the @xmath232 boson in @xmath233 collisions at @xmath230 1.8 tev , phys .
lett . * 84 * , ( 2000 ) , 835840 .
aleph , delphi , l3 and opal coll : searches for invisible higgs bosons : preliminary combined results using lep data collected at energies up to @xmath234 gev , hep - ex/0107032 .
delphi coll .
, p. abreu _ et al _ : a search for heavy stable and long - lived squarks and sleptons in @xmath231 collisions at energies from 130 to 183 gev , phys
b444 * , ( 1998 ) , 491502 .
martin : a supersymmetry primer , hep - ph/9709356 .
a. datta , s. maity : on the observability of invisible / nearly invisible charginos , phys . rev . *
d59 * , ( 1999 ) , 055019 .
i. v. gaidaenko , a. v. novikov , v. a. novikov , a. n. rozanov , m. i. vysotsky : enhanced electroweak radiative corrections in susy and precision data , phys . rept . * 320 * , ( 1999 ) , 119126 .
d. e. groom _ et al _ : particle data group coll .
j. * c15 * , ( 2000 ) , 1 .
w. de boer , g. burkart , r. ehret , w. oberschulte - beckmann , v. bednyakov , s. g. kovalenko : combined fit of low energy constraints to minimal supersymmetry and discovery potential at lepii , hep - ph/9507291 .
d. m. ghilencea , g. g. ross : precision prediction of gauge couplings and the profile of a string theory , nucl . phys . *
b606 * , ( 2001 ) , 101 .
s. heinemeyer , w. hollik , g. weiglein : constraints on @xmath143 in the mssm from the upper bound on the mass of the lightest higgs boson , hep - ph/9909540 .
j. r. espinosa , r. zhang : complete two - loop dominant corrections to the mass of the lightest _
cp_-even higgs boson in the minimal supersymmetric standard model , nucl
* b586 * , ( 2000 ) , 338 . | we investigate the possibility of incorporating a chiral fourth generation into a gut model .
we find that in order to do so , precision fits to electroweak observables demand the introduction of light ( @xmath0 ) supersymmetric particles .
this also enables us to provide decay channels to the fourth - generation quarks .
perturbative consistency sets an upper bound on the coloured supersymmetric spectrum .
the upper limit on the lightest higgs boson is calculated and found to be above the present experimental lower limit .
gutpa/02/03/01 + _ department of physics and astronomy _ + _ university of glasgow , glasgow g12 8qq , scotland _ |
consider a schrdinger operator @xmath3 with a rational potential @xmath4 not necessarily real .
such an operator is called _ monodromy - free _
if all the solutions of the corresponding schrdinger equation @xmath5 are meromorphic in the whole complex plane for all @xmath6 the first classification result here is due to duistermaat and grnbaum @xcite , who described all monodromy - free operators with rational potentials decaying at infinity .
the pole configurations of the corresponding potentials were studied earlier by airault , mckean and moser @xcite in relation to rational solutions of the kdv equation .
oblomkov @xcite generalised duistermaat - grnbaum s result to the quadratic growth case .
he showed that all the rational monodromy - free operators with rational potentials growing as @xmath7 at infinity are the results of darboux transformations applied to the harmonic oscillator .
the corresponding potentials have the form @xmath8 where @xmath9 is the @xmath10-th hermite polynomial , @xmath11 is a sequence of different positive integers and @xmath12 is the wronskian of functions @xmath13 we are interested in the geometry of the pole configurations of the corresponding potentials ( _ locus _ in the terminology of airault , mckean and moser ) , which are the same as the zero sets of the corresponding wronskians .
this locus has an interesting relationship with the calogero - moser problem and log - gas in a harmonic field , see @xcite . in the case when @xmath14 are consequent numbers it can be also interpreted as the pole set of some rational solutions of the fourth painlev equation and has a regular rectangle - like structure in the complex plane , as was revealed numerically by clarkson @xcite .
a natural question is what kind of pattern do we have in general .
let us label these potentials by partitions @xmath15 such that @xmath16 @xmath17 our main observation ( based on numerical experiments using mathematica ) is that although for a general partition @xmath18 the picture is quite complicated , for its doubled version @xmath19 there exists a simple qualitative relation between the shape of the young diagram and the pattern of zeroes of the corresponding wronskian . in the case of the two - term wronskian @xmath0 we have some quantitative results .
namely , for fixed @xmath10 and large @xmath20 we give an explicit formula for the curve the scaled zeroes @xmath21 in the region @xmath22 lie on asymptotically : @xmath23 is the @xmath10-th chebyshev polynomial .
the distribution of the real parts of the zeroes on this curve satisfy wigner s semicircle law .
the derivation is based on a version of the classical plancherel - rotach formula @xcite found by deift et al in @xcite .
we give also some empirical formulas for the three and four - term wronskians .
hermite polynomials @xmath24 are the classical orthogonal polynomials with gaussian weight @xmath25 ( see e.g. @xcite )
. they can be given by the formula @xmath26 and satisfy the recurrence relation @xmath27 here are the first few of them : @xmath28 @xmath29 we are using the normalisation where the highest coefficient of @xmath30 is @xmath31 but this will not be essential in what follows .
what is important for us is that @xmath32 are the eigenfunctions of the harmonic oscillator : @xmath33 let @xmath34 be a partition and consider the wronskian @xmath35 the wronskians @xmath36 have the following properties : _ 1 .
@xmath37 is a polynomial in @xmath38 of degree @xmath39
@xmath40 \3 .
@xmath41 where @xmath42 is the conjugate of @xmath18 .
recall that the conjugate to a partition @xmath18 is a new partition , whose young diagram is the transpose of the diagram of @xmath6 to prove the last ( duality ) property we note that the harmonic oscillator has also the following ( growing , hence formal ) eigenfunctions @xmath43 with the negative eigenvalues @xmath44 the claim is that the result of applications of darboux transformations at the levels @xmath45 @xmath46 is the length of the partition , and at the negative levels @xmath47 is the length of the conjugate partition , lead to the potentials different only by a constant shift .
this follows , for example , from proposition ( 1.7 ) in macdonald s book @xcite ) , or from the so - called maya representation of the partition .
recall the following well - known diagrammatic representations of a partition @xmath18 ( see e.g.@xcite ) .
the first one is the set of points @xmath48 such that @xmath49 following @xcite we will call it a _
ferrers diagram_. there are two ways to draw this .
one convention ( motivated by matrix theory ) is that the row index @xmath50 increases downwards while @xmath51 increases as one goes from left to right .
another way ( sometimes called french ) is to use a natural cartesian coordinate representation ( see fig . [ fer ] ) .
left : standard version .
right : french version , title="fig:",width=188 ] left : standard version .
right : french version , title="fig:",width=188 ] the second , most common , way , known as a _ young diagram _ , is to use boxes rather than bullets ( see fig . [ young ] ) . .
, width=377 ] since the wronskians are labelled by the partitions , we can ask a natural question : how is the geometry of the corresponding diagram of @xmath18 related to the pattern of the zeroes of @xmath37 ?
figure [ comp ] , produced with the help of mathematica , shows that in general such a relation is not easy to see .
another example is the partition @xmath52 with a triangular young diagram , for which the corresponding wronskian @xmath36 ( up to a multiple ) is simply @xmath53
so we just have one zero at @xmath54 with multiplicity @xmath55 with @xmath56 .
, width=529 ] this is why we found it very interesting that for a special class of partitions , which we call doubled , one can read off the partition from the pattern of zeroes in a straightforward way .
let @xmath57 be a partition .
define its _ doubled version _ as @xmath58 in other words , we double all parts and take them twice . for example , when @xmath59 the doubled version is @xmath60 where the power denotes how many times this part is repeated . note that the shape of the young diagram of the doubled version is similar ( with scaling factor 2 ) to the initial one .
however , for the doubled partitions there is another natural way to represent them , which combines both the usual and the french ways .
namely one can put the diagram of @xmath18 in all 4 quadrants as in fig .
one can naturally define the ferrers version , which we combine with the young version by putting bullets at the centre of each box . for @xmath61,width=377 ]
our main observation is that _ the diagram of the doubled partition @xmath62 gives a good qualitatative description of the zero set of the corresponding wronskian _
@xmath63 , see fig . [ compa1 ] and [ compa2 ] .
we believe that this works for any partition @xmath18 with distinct @xmath64 for @xmath59 and the zeroes of the corresponding wronskian @xmath63,width=529 ] , width=529 ] when some parts are equal then we may have interference between the rows of corresponding zeroes , see fig .
[ inter ] . for partition @xmath65 with two equal parts.,width=226 ] one can generalise this relation to the case of half - integer partitions @xmath66 with some of the parts being half - integers .
an example is @xmath67 for which the doubled partition is @xmath68 the corresponding ferrers diagram has some bullets on the vertical axis and gives a good qualitative picture of the zero set of the corresponding wronskian ( see fig .
[ half ] ) . and zeroes of @xmath63 ( right ) for half - integer partition @xmath69.,title="fig:",width=226 ] and zeroes of @xmath63 ( right ) for half - integer partition @xmath69.,title="fig:",width=226 ]
the following asymptotic analysis shows the limits of this comparison already in the simplest case of one - row young diagram @xmath18 .
consider now the two - term wronskian @xmath0 , corresponding to the partition @xmath70 let us fix @xmath10 and let @xmath71 to study this behaviour of zeroes in this limit we can use the following version of plancherel - rotach formula due to deift et al @xcite . in the scaled variable @xmath72
there are several regions with different asymptotic behaviour of the hermite polynomials ( see fig .
the most relevant for us is the region @xmath73 where we have the following asymptotics , width=302 ] @xmath74 with @xmath75 , @xmath76 and @xmath77 ( see @xcite ) . using this
we can show that @xmath78 ( 1+o(\frac{1}{n})),\ ] ] where @xmath79 in the upper half - plane we have two competing terms : @xmath80 term and the negative exponential component @xmath81 of @xmath82 .
for @xmath83 with small @xmath84 we can approximate @xmath85 as @xmath86 since @xmath87 equating the moduli of the two competing terms , we have @xmath88 or , @xmath89 this leads to the formula ( [ lim ] ) for the curve on which the scaled zeroes lie asymptotically as @xmath90 in the region @xmath91 comparing the arguments of the leading terms and using the relation @xmath92 we see that the real parts of the corresponding zeroes are distributed according to the famous _ wigner s semicircle law _ from the random matrix theory @xcite : the number @xmath93 of scaled zeroes @xmath94 of @xmath0 in the upper - half plane with the real parts in the interval @xmath95 satisfies @xmath96 when @xmath97 we have @xmath98 real ( scaled ) zeroes @xmath99 asymptotically given by @xmath100 : as @xmath1 @xmath101 the unscaled zeroes @xmath102 in the region @xmath103 lie asymptotically on the curve @xmath104 is the @xmath10-th chebyshev polynomial .
figure 10 shows a good agreement of this formula ( curve ) and numerical mathematica calculation of zeroes ( dots ) in the case when @xmath105 the 4 real zeroes in this case approximately are @xmath106 in agreement with the picture .
the case @xmath107 corresponds to the doubled partition @xmath108 .
figure 11 shows that shapes of the young diagram and the corresponding zero set coincide only qualitatively .
indeed , the corresponding asymptotic curve in this case is not just two straight lines but given by ( [ asym ] ) with @xmath109 : @xmath110 , width=302 ]
one can prove the following identity for 3-term wronskians of the eigenfunctions of harmonic oscillator @xmath111 we will restrict ourselves with the case @xmath112 in that case we have @xmath113 using this , previous results about 2-term wronskians and experimenting with mathematica we can suggest the following empirical formula for the limiting shape of the non - real zeroes for large @xmath20 and @xmath114 : @xmath115.,title="fig:",width=226 ] right:@xmath116.,title="fig:",width=226 ] .,width=340 ] for wronskians @xmath117 one can show that @xmath118 assuming first that @xmath119 we have @xmath120 mathematica plots suggest that the zeroes for large @xmath20 and small @xmath114 asymptotically lie on two curves , for which we have the following empirical formulas : @xmath121 for the middle curve and @xmath122 for the outside curve .
below we compare these curves and mathematica plots of zeroes for @xmath123 and values of @xmath10 ranging from 1 to 4 .
the first picture corresponds to the double partition @xmath62 with @xmath124 in all cases we expect the real parts of properly scaled zeroes to satisfy wigner s semicircle law .
we note a peculiar behaviour of zeroes near the points where @xmath125 , which needs a special investigation .
right:@xmath126.,title="fig:",width=264 ] right:@xmath126.,title="fig:",width=264 ] right:@xmath127.,title="fig:",width=264 ] right:@xmath127.,title="fig:",width=264 ] consider now the case @xmath128 and @xmath129 , corresponding to the doubled partitions @xmath62 with @xmath130 the empirical formulas for the asymptotic zero curves for large @xmath20 are @xmath131 for the middle curve and @xmath132 for the outside curve .
they seem to work fairly well when @xmath133 but not too small .
the cases with @xmath134 , @xmath135 and @xmath136 are shown below .
we should say that at the moment this part is just experimental mathematics and requires further investigation .
.,width=377 ] .,width=377 ]
the following property of the wronskians of hermite polynomials was conjectured by the third author in 1990s in relation with the corresponding locus problem solved by oblomkov @xcite .
if this property holds it would give a way to a more effective proof of his result , which is still very desirable .
for every partition @xmath18 , all the zeroes of @xmath37 are simple except possibly for @xmath137 note that the multiplicity @xmath138 of @xmath54 for @xmath36 can be easily computed and has the form @xmath139 where @xmath140 is the difference between the numbers @xmath46 and @xmath141 of odd and even elements respectively among the sequence @xmath142 in particular , for the triangular young diagram @xmath143 we have @xmath144 and @xmath145 , so the corresponding wronskian @xmath146 and all the zeroes collide at zero .
an interesting question is if the number of real zeroes of @xmath37 can be effectively described in terms of the corresponding young diagram . for doubled partitions we have the following conjecture . for doubled partitions @xmath147 with distinct parts , the wronskian @xmath148 has no real roots and has as many pure imaginary roots as there are odd numbers among @xmath149 in the special case when @xmath150 and @xmath151 we can prove this using the integral representation of the corresponding wronskian known from the random matrix theory ( see brzin - hikami @xcite ) : @xmath152 finally
, it would be very interesting to understand how special the hermite polynomials are and how much of this can be generalised to other orthogonal polynomials and to the sextic growth case @xcite .
one of us ( apv ) is grateful to the institute for mathematical research at eth zurich for the hospitality in april 2010 and to robert milson for stimulating discussions . | for a certain class of partitions , a simple qualitative relation is observed between the shape of the young diagram and the pattern of zeroes of the wronskian of the corresponding hermite polynomials . in the case of two - term wronskian @xmath0
we give an explicit formula for the asymptotic shape of the zero set as @xmath1 .
some empirical asymptotic formulas are given for the zero sets of three and four - term wronskians .
_ to boris dubrovin on his @xmath2 birthday _ |
the recent major advances in the observation of the deep universe have substantially improved our chance to reach the redshift of galaxy formation .
objects at @xmath1 begin in fact to emerge from hst observations and other deep surveys with the new - generation telescopes ( cohen _ et al . _ 2000 ) .
as far as we move to larger distances , however , optical ( and infrared ) observations probe galaxy spectral energy distribution ( sed ) at much shorter restframe wavelength , in the ultraviolet domain .
if neglected , this change in our spectral vantage point could lead to an important bias in the interpretation of high - redshift data . in this note
, we will briefly analyze some aspects of this problem dealing both with the observational side and the appropriate use of theoretical tools such as population synthesis models to match primeval galaxy evolution .
one important consequence of redshift ( besides the first and most obvious effect of moving spectral features to longer wavelength ) is that the wavelength interval @xmath2 sampled by any fixed set of photometric bands will span a narrower portion of the restframe sed when observing galaxies at increasing @xmath0 , such as @xmath3 .
this effect , by itself , can be of paramount importance as far as we try to gain information on galaxy stellar content by means of multicolor photometry .
figure [ f1 ] gives an illuminating example in this sense . for a @xmath4 galaxy , for instance ,
johnson four - color photometry in the _ bvri _ bands would span a wide wavelength range , from 4000 to 9000 , and collect a substantial contribution from the galaxy stellar population . on the contrary , when observing at @xmath5 with the same photometric system , we would be covering a scarce @xmath6 spectral window centered about 1600 in the galaxy restframe .
quite deceivingly , our `` multicolor '' photometry at @xmath4 would translate into a nearly `` monochromatic '' estimate of the galaxy flux at @xmath5 .
according to the overall color - magnitude distribution , each star in a galaxy aggregate contributes with a different weight to the integrated luminosity with varying wavelength . in buzzoni ( 1993 )
we approached this problem in a statistical way introducing the concept of `` effective '' contributors in a stellar population .
this quantity directly relates to galaxy integrated properties , such as total luminosity or surface brightness fluctuations ( as further explored , for instance , by tonry and schneider 1988 ) .
we showed , in particular , that by sampling at a given wavelength a total luminosity @xmath7 provided by @xmath8 stars , the expected poissonian fluctuation results : @xmath9 where @xmath10 is the luminosity of each composing star .
if all stars have the same luminosity , then simply @xmath11 .
more generally , however , a quantity @xmath12 can be defined , as a function of @xmath13 .
its value gives a statistical measure of the effective number of stars that provide @xmath7 at a given wavelength .
following buzzoni ( 1993 ) , a `` mean '' luminosity ( @xmath14 ) for the composing stars can also be computed as @xmath15 so that , at every wavelength , @xmath16 in fig . [ f2 ] we show the expected value of @xmath17 for the illustrative case of a 5 gyr simple stellar population ( ssp ) with salpeter imf and [ fe / h ] @xmath18 along with the integrated sed of the model .
this should closely resemble the real case of primeval elliptical galaxies ( buzzoni 1995 ) . as a stricking feature of the model , note that below 2000
the effective number of `` luminous '' contributors dramatically drops by over six ( ! )
orders of magnitude compared for example with the @xmath19 band value at 4000 . on the contrary , luminosity over the same wavelength range only dims by a factor of @xmath20 . in terms of surface brightness fluctuations
this means , for instance , that the ultraviolet image of a galaxy consisting of @xmath21 stars should fairly compare with the @xmath19 observations of a ( @xmath22 star - poorer ) galactic globular cluster .
such a severe undersampling of the galaxy stellar population in the ultraviolet range could have an important impact when studying morphology of @xmath23 ellipticals at optical wavelength ( but the same argument holds also for late - type systems ) .
in particular , high - redshift objects would more likely appear as coarser systems with increasing @xmath0 ( e.g. puerari 2001 , this conference ) .
the selective sampling of galaxy stellar population with changing wavelength has even more pervasive consequences , as far as we track evolution of starforming systems at different redshift .
more generally , we could describe a composite stellar population in terms of a suitable convolution of ssps according to the star formation rate ( sfr ) at the different epochs .
the total luminosity at age `` @xmath24 '' results therefore : @xmath25 a mean luminosity - weighted age of the composing ssps can also be derived such as @xmath26 this could be regarded as the `` representative '' age of the stars that contribute to galaxy luminosity . we know
( buzzoni 1995 ; tinsley and gunn 1976 ) that ssp luminosity evolution can be pretty well described by a simple power law such as @xmath27 with the power index @xmath28 that is a function of wavelength . as an instructive example , if we consider for the sfr a power - law time decay such as sfr @xmath29 , .
this easily derives recalling that , at age @xmath24 , @xmath30 and its time average is @xmath31 . providing that @xmath32 ,
we always have @xmath33 , from which the value of @xmath34 directly follows , by definition . ] eq .
( [ eq : ave ] ) becomes @xmath35 the integral has an analytical solution such as @xmath36 in bolometric , @xmath37 ( buzzoni 1989 ) and , for a constant sfr ( i.e. @xmath38 ) , @xmath39 . at 15 gyr ,
bright stars are therefore in average 3 gyr old , but this value changes according to the @xmath40 function . for a ssp with salpeter imf and @xmath41 , the luminosity power index smoothly increases from @xmath42 in the @xmath43 band to 0.94 in the @xmath19 band , and exceeds unity below 3500 .
a value of @xmath44 implies that @xmath45 , that is only the youngest and more massive stars are representative contributors to galaxy ultraviolet luminosity , as discussed in the previous section . the mean age of stars contributing to galaxy luminosity at different wavelength for a synthesis model of a 15 gyr magellanic irregular ( see buzzoni 2001 for details ) is reported in fig .
[ f3 ] . note from the figure that the value of @xmath46 smoothly decreases with decreasing wavelength . as a result
, high - redshift objects forcedly appear to be younger than local homologues at @xmath4 _ in spite of any intrinsic evolution . _
evolutionary population synthesis models are extensively used as a reference tool to reproduce galaxy colors and investigate galaxy spectrophotometric properties at the different cosmic epochs ( buzzoni 1989 , 1995 , 2001 ; bruzual and charlot 1993 ; worthey 1994 ; bressan _ et al . _ 1994
; fioc and rocca - volmerange 1997 ) .
besides any difference in the computational details and input physics , a crucial constraint that needs to be consistently fulfilled in the synthesis procedure concerns a suitable match of the ms and post - ms luminosity contribution in the ssp models . the so - called `` fuel consumption theorem '' of renzini and buzzoni ( 1986 ) provides a simple and very powerful tool in this sense . according to the theorem
, the post - ms bolometric luminosity in a ssp of total luminosity ( @xmath47 ) is @xmath48 in the equation , the nuclear fuel refers to the amount consumed by one star leaving the ms turn - off ( to ) point along the whole post - ms evolution .
this is a natural output of the theoretical stellar tracks and is therefore known with high accuracy .
the scaling factor @xmath49 is the `` specific evolutionary flux '' of the ssp , and directly deals with the stellar clock , that is the relationship between ms lifetime and to stellar mass ( see renzini and buzzoni 1986 ) .
in particular , we have that @xmath50 assuming that a power - law imf holds with @xmath51 ( @xmath52 for the salpeter case ) . again , the r.h . term of eq .
( [ eq : b2 ] ) is a direct output of stellar evolution theory , and @xmath53 should therefore be regarded as a `` universal '' function , that is nearly insensitive to the specific input physics of the ssp models . a comparison of the @xmath53 scaling factor , as derived from the synthesis codes of bruzual and charlot ( 1993 ) , buzzoni ( 1989 ) , and the padova group ( bressan _ et al . _ 1994 ) for a ssp with @xmath41 and salpeter imf is displayed in fig .
while both the buzzoni and padova codes consistently match the fuel consumption theorem , the bruzual and charlot models show a much higher value for the specific evolutionary flux .
the effect is a macroscopic one , and reaches about a 100% discrepancy at intermediate age .
such higher scale factor leads , in eq .
( [ eq : b ] ) , to a corresponding overestimate of post - ms luminosity at every age .
the effect is quite shifty as it will not explicitly appear when considering luminosity evolution ( cf .
[ f5 ] , left panel ) , but would strongly affect the theoretical prediction of galaxy m / l ratio and apparent color evolution .
in the latter case , enhanced post - ms contribution would lead to a larger number of red giant stars and therefore a systematically redder color for the stellar aggregate as a whole .
this is shown in fig .
[ f5 ] ( right panel ) when comparing the bruzual and charlot ( 1993 ) results with the expected evolution of elliptical galaxies according to buzzoni ( 1995 ) . | the effect of redshift on the observation of distant galaxies is briefly discussed emphasizing the possible sources of bias in the interpretation of high-@xmath0 data . a general energetic criterion to assess physical self - consistency of evolutionary population synthesis models
is also proposed , for a more appropriate use of this important tool to investigate distinctive properties of primeval galaxies . |
the lyman - alpha emission line ( @xmath0 ) at 1216 fulfills several extremely important roles in observations of the high - redshift ( @xmath4 ) universe .
recombination nebulae re - process @xmath5 of the raw ionizing power of hot stars into a single emission line which acts as a luminous spectral beacon by which to identify galaxies at the highest redshifts ( see , e.g. , hayes etal .
2010 and references therein ) .
further , at the highest redshift it provides the potential for the all - important spectroscopic confirmation of distant galaxies selected by other methods , transforming the status of candidates to objects with secure redshifts ( e.g. , stark etal . 2010 ) .
@xmath0 is a resonance line and is readily scattered by neutral hydrogen atoms .
this scattering may occur in the interstellar medium ( ism ) of the galaxies themselves , or in the circumgalactic medium ( cgm ) that immediately surrounds them .
scattering in the cgm can lead to extended @xmath0 structures such as those found in high - z @xmath0 emitters ( lae ) by @xcite and @xcite among others . inside the ism ,
the radiative transport of @xmath0 depends strongly on dust extinction @xcite .
this dependence , however , shows large scatter due to the complicated resonant scattering of @xmath0 , in which the visibility of the line is influenced by a large number of factors , including dust content ( charlot & fall 1993 ; verhamme etal .
2008 ; atek etal . 2009
; hayes etal . 2010 ) , dust geometry ( scarlata etal .
2009 ) , neutral gas content and kinematics ( kunth etal . 1998 ; mas - hesse etal . 2003 ; cannon etal .
2004 ) , and gas geometry ( neufeld 1991 ; giavalisco etal . 1996 ;
hansen & oh 2006 ; laursen et al .
2013 ; duval et al .
taken together , these myriad factors indicate that the true total escape fraction of @xmath0 photons ( @xmath6 ; defined as the ratio of observed to intrinsic @xmath0 luminosity ; hayes etal .
2005 , hayes etal .
2013 ) , is still poorly understood .
thus , photometric measurements of @xmath0 will reflect the underlying properties of galaxies only in the very broadest statistical sense .
the only way to remedy this complicated situation is to study the @xmath0 emission line in a sample of star - forming galaxies on a spatially resolved basis . from the uv perspective , obtaining @xmath0 images in the nearby universe is expensive and requires space - based platforms . to this end
, our team has acquired data from the hubble space telescope ( hst ) to produce @xmath0 images of a statistically selected sample of star - forming galaxies in the low-@xmath4 universe ( see stlin etal .
2014 ) ; details about the image processing methods can be found in hayes etal .
this program , called the `` lyman alpha reference sample '' ( `` lars '' ) , includes hst cycle 18 imaging and cycle 19 spectroscopic observations of 14 galaxies .
these hst datasets form the backbone of a comprehensive multi - wavelength observational campaign that will allow great strides forward in our understanding of @xmath0 radiative transport .
since the lars sample is selected on uv luminosity and h@xmath7 equivalent width , and not on @xmath0 characteristics , the composite sample was designed to overlap with both high-@xmath4 and low-@xmath4 star - forming galaxies .
lars will form a critical local interpretive benchmark for studies of high-@xmath4 galaxies for decades to come , and is well placed to take advantage of the remaining hst lifetime and motivate future observations with jwst . the lars program ( including sample selection )
is described in detail in @xcite and @xcite , hereafter referred to as papers i and ii respectively .
preliminary results derived from the @xmath0 images of the lars galaxies are presented in @xcite .
that work showed that in many lars galaxies , @xmath0 emission is prominent on physical scales that exceed those of both the massive stellar populations and the star formation regions that give rise to the @xmath0 photons ; this trend is very prominent in the panels of figures [ fig_opta ] , [ fig_optb ] , and [ fig_optc ] , which show optical and uv images of the lars galaxies .
this may point towards @xmath0 photons being resonantly scattered to large radii in most of the lars galaxies , although other possibilities have been proposed , such as ionization caused by hot plasma @xcite .
it is likely that multiple of the aforementioned physical properties ( dust content and geometry , neutral gas content , kinematics , geometry ) are facilitating this efficient resonant scattering . in order to decouple the effects of dust and neutral gas kinematics and densities ,
the @xmath0 images must be compared with spatially resolved information about the neutral ism .
while uv and optical absorption line spectroscopy offers one avenue to address this issue , such data are limited to those regions in the foreground of only the brightest continuum sources ( usually the emission line nebulae themselves ) . while our team is pursuing such analyses ( see rivera - thorsen etal .
, in preparation , and @xmath0 profile modeling from orlitova etal .
, in preparation ) , these avenues should be complemented by direct probes of the neutral interstellar medium in all regions of the lars galaxies . to this end , in this manuscript we begin a detailed exploration of the neutral hydrogen in the lars galaxies by presenting new single - dish hi spectroscopy and spatially resolved interferometric h i imaging .
we use these data to derive global properties and to study the properties of the h i gas on bulk scales .
previous spatially resolved h i observations of galaxies in which @xmath0 radiative transport has been studied in detail are few . while the sample of systems with spectroscopic information about @xmath0 is robust , there are comparatively few systems that have direct @xmath0 imaging ( see the discussion in hayes etal .
2005 , hayes etal .
2009 , stlin etal .
2009 , and references therein ) .
for those systems that do , there has , until the present work , been no systematic investigation of the h i on a spatially resolved basis ; only a few selected systems have detailed h i studies in the literature .
@xcite presented h i observations of two of the most well - studied and nearby @xmath0-emitting galaxies in the local universe , eso338@xmath8ig004 ( tol1924@xmath8416 ) and iras 08339 + 6517 .
in each galaxy , extended neutral gas connects the target and nearby neighbors , suggesting that interactions have played an important role in triggering the massive starbursts in eso338@xmath8ig004 and iras 08339 + 6517 .
the close proximity of the companions suggests that the interactions were recent . since both primary systems are @xmath0 emitters , these data support two possible interpretations of @xmath0 escape from starburst galaxies : either a ) the bulk ism kinematics provides the means of escape for @xmath0 photons by shifting h i atoms out of resonance , or b ) the h i is sufficiently clumpy on scales below our resolution to allow for efficient @xmath0 escape .
these interpretations are further supported by h i @xcite and @xmath0 observations @xcite of eso400-g43 . in this work , we begin to extend this type of analysis to the entire lars sample .
basic properties of the lars galaxies are summarized in table [ table_basic ] .
figures [ fig_opta ] , [ fig_optb ] , and [ fig_optc ] show optical and uv images of the 14 lars galaxies , with the left and right columns displaying the color hst images and the sdss color images respectively .
the right column includes the immediate surroundings , while the left column has smaller fields of view , and are scaled to show detail in the galaxies .
we organize this manuscript as follows . in [ s2 ]
we describe the data acquisition , reduction , and analysis .
[ s3 ] presents the single - dish h i profiles of the 14 lars galaxies , while [ s4 ]
presents spatially resolved h i images of 5 of the 14 lars systems .
we discuss each system individually in
[ s5 ] , and discuss various correlations between galaxy global properties in [ s6 ] .
we draw our conclusions in [ s7 ] . throughout this paper
we adopt the value of h@xmath9 @xmath10 [email protected] kms@xmath12mpc@xmath12 from the averaged wmap 7 year data as presented in @xcite .
we obtained data from the national radio astronomy observatory 100 m robert c. byrd green bank telescope ( gbt ) under program gbt/11a-057 ( legacy i d qo13 ; p.i .
data were acquired in four observing sessions in march and april of 2011 .
position switching observations were performed for all sources ; 32,768 channels over the 12.5 mhz of total bandwidth produce a spectral resolution of 381.469 hz ( 0.08 kms@xmath12ch@xmath12 ) .
all reductions were performed in the idl environment , using the gbtidl package designed at nrao .
we first imported the raw data for each galaxy into gbtidl using standard averaging techniques , with zenith opacity coming from atmospheric data from the nrao cleo weather system .
the data was moderately contaminated by terrestrial radio frequency interference ( rfi ) ; bad data were identified by examining individual spectra , blanked , and interpolated .
the standard position switching algorithms in gbtidl were used to produce a combined spectrum for each galaxy . to increase the signal to noise ratio ( snr ) of the observations we first averaged the reference spectrum by 16 channels .
this procedure can have a negative effect on the baseline shape , increasing the order of the polynomial ( often 3rd order ) that we used to fit and subtract the continuum ( see table [ table_gbt_props ] for the order of the fit for each galaxy ) .
the snr was quite low on most of the lars galaxies and we thus smoothed our spectrum by 256 channels resulting in a velocity resolution of @xmath13 20kms@xmath12 .
flux and width measurements were taken on this smoothed profile .
two regions were selected by hand for each galaxy : the first around the profile , and the second in an area free from rfi ( used as an rms noise region ) .
the redshifts of the lars galaxies ( see table [ table_basic ] ) place some of them in proximity to a bandpass filter in the 1.2 - 1.3 ghz region .
this bandpass filter is designed to shield against a local radar signal , but contributes higher than usual system temperature ( t@xmath14 ) values ( and therefore system noise ) to calibrated data in this frequency range .
further , the t@xmath14 values fluctuated markedly from one spectral scan to the next .
the affected galaxies were all non detections ( lars12 , lars13 , and lars14 ) and have markedly higher t@xmath14 values ( in particular lars13 ) than would otherwise be expected .
it should also be noted that these three galaxies are at the largest redshifts in our sample . systematically measuring the flux and width of irregular , low snr , h i spectra can be challenging .
to do so , we have modified previously well known measurement techniques and find that they are robust over the range of snr and profile shapes in our data .
our measurement routines take as inputs the smoothed spectrum and the region suspected of containing the galaxy . in our initial tests of archival data for ngc 5291 ( available in the distribution of gbtidl )
, we found that hand selecting regions of the profile could add additional bias , causing the width and flux measurements to vary by 2 - 5% .
instead , we opted to use @xmath15 minimization techniques to fit gaussian components ( a single component fit for galaxies with a single peak and a multi component fit for double - horned profiles ) to the smoothed spectrum .
these components were used to guide later analysis and ensured repeatability in our results even with low snr .
these gaussian fits serve only to guide analysis , however , and the flux and width measurements follow the methodology described in @xcite .
this method fits lines to the profile in the region over which it varies the most ( here between 15% and 85% of the peak value ) and measures width properties from these fits .
@xcite found that this reduced the dependence on noise on their measurements versus direct measurements .
we modified this method to respond better to low snr values and to handle single peaked data more accurately .
we consider this method to be the `` best representation '' of the width measurements for these spectra , although it is not the only method ( e.g. , the busy function ; * ? ? ?
we recorded two snrs based on the peak - to - rms ratio and on the integrated flux to integrated rms ratio .
the flux accuracy of gbt data using standard reduction techniques is quoted as 10% in the gbtidl calibration guide .
the primary error terms are the rms fluctuations of the spectra in the region around the target and the subjectivity of selecting the baseline regions and rms / profile regions .
in testing it was seen that increasing the order of the baseline generally led to increasing error .
we thus adopt a 10% flux error for `` good '' profiles and a conservative 20% for `` irregular '' profiles .
special attention was paid to the error terms presented in @xcite as modified from @xcite . the error term given
there adds an error associated with the quality of the baseline fit .
because we have assumed a separate baseline error , we remove this term to get a more or less standard error term : @xmath16 where @xmath17 is the noise in units of jy kms@xmath12 , and w is the area of integration for the profile .
the width at 50% of the maximum value ( w@xmath18 ) results reported are highly dependent on the accuracy of the method described in @xcite which , as mentioned in the above discussion , breaks down at lower snrs .
the error for the width and systemic velocity measurements is based on the equations found in @xcite .
@xmath19 this equation is doubled to give the error of the width measurements w@xmath20 and w@xmath18 , which refer to the width at 20% and 50% of the maximum , respectively .
we obtained d - configuration very large array h i spectral line imaging of five lars galaxies under program vla/13a-181 ( legacy i d ac1123 ; p.i .
these data were obtained with a standard widar correlator configuration that provides 16 mhz of total bandwidth .
1,024 channels deliver a native spectral resolution of 3.3 kms@xmath12 .
the data were acquired in march and april of 2013 , during seven observing sessions ; details are summarized in table [ table_vlaobs ] .
data reductions followed standard prescriptions in casa and in aips . the wide bandwidth and
frequency range of these observations resulted in a significant amount of radio frequency interference that was removed from the data .
bandpass and flux calibrations were derived using either 3c147 or 3c286 ( see table [ table_vlaobs ] ) .
the gains and phases were then calibrated using observations of the phase calibrators , which were typically separated by @xmath1320 minutes of on - source integration time . the wide bandwidth allowed for hundreds of channels of line - free continuum to be fit and subtracted . for lars02 and lars09 , which each had two separate observing sessions , we combined the separate observations into a single dataset for each galaxy .
the calibrated data were then inverted and cleaned using the clean algorithms in casa . to maximize snr , the datasets for lars04 and lars08
were spectrally smoothed to a final velocity resolution of 10 kms@xmath12 .
a gaussian uvtaper was applied to produce cubes with beam sizes of @xmath13 1 ; the resulting cubes were then convolved to circular beam sizes . the beam sizes and rms noises of the final datacubes are summarized in table [ table_vla_props ] .
moment maps were derived using standard blanking procedures .
first , each datacube was spatially smoothed to a circular beam size that is larger than the original .
the resulting smoothed cube was then blanked at the 2.5@xmath21 level , where @xmath21 is the rms noise in line - free channels of the cube .
this blanked cube is then inspected by hand , and only features that are spatially coincident across two or more consecutive channels are deemed to be real emission and kept in the final blanked cube .
this cube is then used as a blanking mask against the original , unconvolved cube .
moment maps are then derived using standard procedures .
we assume a 10% uncertainty on the overall calibration and absolute flux scale of the vla images .
it is important to emphasize that all of the lars galaxies are at distances of more than 100 mpc ( z@xmath22 @xmath23 0.028 ; see table [ table_basic ] ) ; even with angular resolutions that are exquisite for h i work ( e.g. , 5 @xmath10 2.4 kpc at a distance of 100 mpc ) , we are sensitive only to the bulk h i properties of the neutral gas in these galaxies . as such ,
the vla images presented here are used only to study the global morphology and large - scale h i kinematics .
a detailed study of the h i on finer physical scales will require deep observations in more extended array configurations .
we use global @xmath0 data for all 14 lars galaxies as presented in paper i and paper ii . the hst data were obtained with the wide field camera 3 ( wfc3 ) and advanced camera for surveys ( acs ) using h@xmath7 and h@xmath24 filters , and a combination of u , b , and i bands for fuv continuum imaging .
we also acquired data with the solar blind channel using a bandpass designed to isolate @xmath0 emission .
we reduced imaging data using multidrizzle in the standard hst pipelines and used the _ lyman - alpha extraction software _ ( laxs ) @xcite to continuum subtract the @xmath0 uv data .
we refer the reader to paper i and ii for details of this process .
the final global properties ( see paper i and table [ table_uv_properties ] ) are measured within apertures that correspond to twice the petrosian radius .
this reduces the effect of noise on the measurements at large radii .
presented in figure [ fig_gbt1 ] are the fully smoothed and reduced gbt h i profiles of the lars galaxies . plotted over these
are the gaussian fits in gray , solid lines in red showing the sides of the profile used in the peak w@xmath25 calculations , and the peak w@xmath25 line itself shown as a horizontal blue dotted line .
a discussion of each galaxy is included below .
the observed properties from our gbt observations of each galaxy can be found in table [ table_gbt_props ] along with properties derived from these using methods described below .
the distances of each galaxy are calculated from the standard lars luminosity distances .
the h i mass in units of m@xmath3 is calculated via @xmath26 where d is the distance in mpc and s@xmath27 is the h i flux integral in units of jy kms@xmath12 .
we will discuss the mass value for each galaxy in section [ s5 ] .
overall we find that our detections and mass estimates show a strong dependence on distance .
it is important to emphasize that the 8 primary beam of the gbt at 21 cm is sensitive to all h i - bearing objects in the observed frequency range . even for the closest lars galaxy ( lars01 ; [email protected] mpc ) ,
this 8 beam subtends a projected circular area that is nearly 280 kpc in diameter ; the potential for contamination in the gbt beam is significant .
we thus examined sdss images of the area surrounding each lars galaxy for nearby objects .
objects with spectroscopic data available from sdss that might have contributed to the gbt flux are noted below .
these objects have velocities within 300 kms@xmath12 of the target lars galaxy and are within the 8 primary beamsize of the gbt .
optical images of the fields that contain possible confusing sources ( lars06 and lars11 ) are shown in figures [ fig_lars06_field ] and [ fig_lars11_field ] ; these two sources are discussed in more detail in [ s5 ] .
we used the sdss g band to convert to luminosity l@xmath28 , which we compared with a log(m@xmath27/l@xmath29 ) value of @xmath80.5 for late type galaxies found in @xcite .
the masses of potential contamination galaxies are computed from this ratio .
we also use the 22.2 magnitude limit for the sdss to determine the possible contributions from optically dim sources of contamination . for our closest nine galaxies
this is roughly 10@xmath30 m@xmath3 of h i ; this increases to 10@xmath31 for our furthest two systems .
this information is also presented in table [ table_gbt_props ] .
the profiles in figure [ fig_gbt1 ] show a variety of h i structure .
some galaxies are simple single peak profiles ( e.g , lars04 , lars09 ) , while others show complex and asymmetric profiles ( e.g. , lars01 , lars03 ) .
still other systems have h i spectra that are almost certainly confused ; the classic double - horn profile of lars06 is likely due to the contribution of ugc10028 , a large spiral galaxy located only 1 to the southeast of lars06 ( see figure [ fig_lars06_field ] ) . various observed properties of each gbt h i profile , and the quantities derived from these , are tabulated in table [ table_gbt_props ] : column 1 identifies the galaxy number ; column 2 is the systemic velocity in kms@xmath12 ; column 3 is the linewidth at 50% in kms@xmath12 ; column 4 is the linewidth at 20% in kms@xmath12 ; column 5 is the velocity offset between the optical velocity and the h i central velocity ; column 6 is the the single - dish integrated flux in jy kms@xmath12 ; column 7 is the mass from the single - dish measurements in units of 10@xmath2 m@xmath3 ; column 8 is the snr from the integrated flux and the peak ; column 9 is the rms noise value of the gbt in units of mjy ; column 10 gives the baseline polynomial fit used in the calibration of the gbt spectrum ; column 11 is the classification of the profile ( ` s ' for single - horned , ` d ' for double - horned , ` n ' for no detection , ` i ' for highly irregular shapes , or the presence of rfi near the signal region , and ` c ' for confusion due to known nearby galaxies . )
the h i moment zero ( representing h i mass surface density ) and moment one ( representing intensity - weighted velocity ) images of the 5 lars galaxies observed in program 13a-181 are presented in figures [ fig_vla_lars02 ] , [ fig_vla_lars03 ] , [ fig_vla_lars04 ] , [ fig_vla_lars08 ] , and [ fig_vla_lars09 ] . in panel
( a ) of each figure , a digitized sky survey image is overlaid with contours of h i mass surface density ( individual contour levels are provided in the caption to each figure ) .
the beam size is shown by a blue circle , while a red square shows the location of the hst pointing for which we have uv imaging ( see discussion in paper i and ii ) . in panel ( b ) of each figure , the h i moment one image ( representing intensity weighted velocity ) is presented in a rainbow color format ; the velocity scale of each galaxy is shown by a colorbar .
based on the distances and optical angular sizes of the lars galaxies ( of order 1 or less ; see figures [ fig_opta ] , [ fig_optb ] , and [ fig_optc ] ) , one would expect the h i to be distributed on similar angular scales . at the resolution of these data ( beam sizes between 59 and 72 )
, the lars galaxies would thus appear as unresolved sources in the h i data cubes .
as we discuss below , three of the lars galaxies have h i morphologies consistent with this interpretation .
surprisingly , two of the sources have h i distributions that largely exceed the beam size ; we discuss these sources in detail below .
it is important to stress that even in cases where galaxies are unresolved by an h i beam , one can still extract bulk characteristics of the neutral ism because of the spectrally resolved nature of the data .
detailed kinematic analyses ( e.g. , rotation curve work or spatially resolved position - velocity diagrams ) will be unavailable in such cases .
however , within the spatial and spectral resolution elements of the data , one can derive bulk constraints on the motions of the h i gas .
we use the vla h i images of the 5 lars galaxies presented here to constrain two critical properties .
first , the h i data allow us to localize the h i associated with the lars galaxy .
this in turn allows us to accurately measure the total h i mass of each system .
these interferometric measurements often recover less of the h i flux , and therefore mass , than the gbt measurements .
we attribute this to the low snr of h i in these galaxies , the possible presence of low surface brightness h i features missed by the vla in extended regions or velocity space , and flux contamination from other galaxies , although we suspect that this is only relevant in the cases marked as confused .
it is interesting to note that at the distances of the lars galaxies , interferometric observations are perhaps better suited to determine the h i mass than single - dish observations ( for example , the 60 characteristic resolution of the present data are about a factor of three better than the h i beam of arecibo , the largest single - dish radio telescope in the world , and a factor of eight better than our gbt resolution ) , simply because of the potential for contamination in the single - dish beam .
second , the present data are sensitive to large - scale distributions of h i gas , some of which is expected to be tidal in origin based on the optical morphologies of the lars galaxies alone ( see , e.g. , figures [ fig_opta ] , [ fig_optb ] , and [ fig_optc ] ) . in table
[ table_vla_props ] we present vla h i quantities for our five targeted galaxies derived in the same method as described above for the gbt data . column 1 identifies the galaxy number ; column 2 is the systemic velocity in kms@xmath12 ; column 3 is the linewidth at 50% in kms@xmath12 ; column 4 is the linewidth at 20% in kms@xmath12 ; column 5 is the velocity offset between the optical velocity and the h i central velocity ; column 6 is the the interferometric integrated flux in jy kms@xmath12 ; column 7 is the mass in units of 10@xmath2 m@xmath3 ; column 8 is the rms noise value in units of mjy / beam ; column 9 is the median global column density of h i from a column density map convolved with the beam size given in 10@xmath32 @xmath33 .
we now present discussion of each lars target based on our new h i data and on the hst imaging presented in @xcite , paper i , and paper ii .
lars01 is a strong @xmath0 emitter , with a global ew of 46 (paper i ) and a moderate star formation rate ( a few m@xmath3 yr@xmath12 ) .
the single - dish h i observations ( see figure [ fig_gbt1 ] ) reveal an asymmetric double - horn profile with peak snr of 6.5 and linewidth at 50% of 160 kms@xmath12 .
we recover a total h i flux of 0.7 jy kms@xmath12 which at our distance of 118 mpc gives an h i mass of 2.3@xmath34 m@xmath3 .
a search of the sdss field revealed no other objects within a @xmath11300 kms@xmath12 velocity range . assuming that the h i profile is unconfused
, the asymmetric line shape and offset from the optical velocity is suggestive of bulk outflow of h i. consistent with this interpretation is the observation in paper ii that lars01 shows extended h@xmath7 and h@xmath24 emission .
further , preliminary results from cos spectroscopy show that within the cos aperture ( 2.5 diameter ) , large column densities of neutral hydrogen are outflowing at around @xmath8128 kms@xmath12 with respect to the h@xmath7 velocity .
note that lars01 will be studied in detail within paper i , while rivera - thorsen etal .
( in preparation ) will present the cos analysis of the whole sample .
interstellar metal lines are blueshifted with respect to the systemic velocity of the galaxy , and there is a distinct lack of static gas in the cos aperture .
this is consistent with the recent results from @xcite , whose analysis of the interstellar oi , si ii and cii lines in this galaxy also yields an outflow of the neutral gas at @xmath8130 kms@xmath12 ( not including the geocoronal velocity offset applied by these authors ) .
this large shift is also seen in our measurement of the systemic velocity of h i in lars01 .
as the closest lars galaxy , future interferometric h i observations of lars01 will be especially interesting .
this galaxy is the strongest @xmath0 emitter ( in terms of escape fraction ) in the sample with a @xmath0 ew @xmath13 82 , yet has one of the lowest ( h@xmath7 or uv - based ) star formation rates .
the gbt spectrum is double peaked ( see figure [ fig_gbt1 ] ) and has a linewidth @xmath13 140 kms@xmath12 .
the total single - dish h i flux is 0.7 jy kms@xmath12 , which gives a mass of 2.7@xmath35 m@xmath3 at our derived distance of 127 mpc .
we find no sources of potential contamination in the sdss field .
the vla images of lars02 ( see figure [ fig_vla_lars02 ] ) reveal a compact distribution of h i gas ; the h i is only slightly resolved by the 59 ( 36.3 kpc ) beam .
the h i column densities peak at @xmath132@xmath3610@xmath37 @xmath33 ( although the marginally resolved nature of the source strongly suggests that the h i distribution is more localized , with correspondingly higher mass surface densities ) .
there is evidence for coherent rotation in this source ; the isovelocity contours shown in panel ( b ) of figure [ fig_vla_lars02 ] are parallel from @xmath13@xmath815 15 kms@xmath12of the systemic velocity of 8960 kms@xmath12 .
a coarse estimate of the dynamical mass of the system can be made by assuming that this rotational velocity is occurring over the angular diameter of the beam , and that the total projected rotation velocity is symmetric about a dynamical center .
the resulting m@xmath38 = 3.8@xmath3610@xmath39 m@xmath3 ( which will increase for any non - zero value of disk inclination , but will decrease depending on the actual size of the rotating component of the disk ) is consistent with the interpretation in paper ii of lars02 being among the more dwarf - like systems in lars .
we stress that this dynamical mass estimate is meant to be representative only .
lars03 is the nuclear region of the southeastern galaxy in the spectacular arp238 interacting pair .
note that figure [ fig_opta ] shows the large - scale interacting morphology very clearly in the sdss panel , while the hst @xmath0 imaging covers only the southern of the two nuclei , and only a relatively small portion of the total interacting system .
paper ii finds that lars03 is a weak @xmath0 emitter .
interestingly , the @xmath0 luminosity increases as a function of distance from the source , suggesting that @xmath0 photons are readily being scattered to large distances even though the @xmath0 extension parameter ( @xmath40 , defined as the ratio between the @xmath0 and h@xmath7 petrosian radii ) is unity . based on the interacting morphology of the system
, we expect both an irregular global h i profile and extended h i structure .
both of these expectations are borne out by the data .
the gbt profile of lars03 ( see figure [ fig_gbt1 ] ) is double - peaked and asymmetric .
the linewidth exceeds 300 kms@xmath12 at the 50% level , and there is evidence for low surface brightness h i emission on even larger velocity scales ( w@xmath41 = 380@xmath1162 kms@xmath12 , and the profile shows weak evidence for h i emission over a range as large as @xmath13600 kms@xmath12 ) .
the h i morphology and kinematics of lars03 clearly indicate a violent interaction between the two galaxies .
the h i emission is extended many beam sizes beyond the optical galaxy . as shown in @xcite , @xmath0 emission
is found in the southeast of the two interacting galaxies ( although those authors point out that the nature of @xmath0 emission on larger spatial scales is not constrained by data presently in hand ) .
our vla images reveal that the h i surface density maximum is coincident with the northwest of the two interacting galaxies ( although this is within one beam width at the present spatial resolution ) .
this region is not included in the present hst observations .
the total h i mass in the interacting system is @xmath139@xmath3610@xmath39 m@xmath3 , and the average h i column density as revealed by our vla beam is 4.8@xmath42 @xmath33 .
an enormous tidal feature , which is contiguous in velocity space from the two interacting galaxies , extends @xmath13160 kpc to the west - southwest and contains @xmath131.6@xmath3610@xmath39 m@xmath3 of hi gas . because of the large distance separating this component from the main body of lars03 we decide to remove this mass from the total h i mass estimate of lars03 itself
this extended tidal structure is reminiscent of that in the local @xmath0-emitting starburst galaxies tol1924@xmath8416 and iras08339@xmath436517 found by @xcite , in which the large - scale neutral gas kinematics are interpreted to be critical for @xmath0 radiative transport .
although we do not probe the same physical scales in this work , we can conclude from this large tidal feature that the neutral gas is strongly disturbed . even with these interesting h i kinematics and the obvious interacting nature of the system , which contributes to its luminous infrared nature
, lars03 remains a relatively weak emitter of @xmath0 photons , that would be undetected in most high - redshift surveys . while the ever increasing @xmath0 emission to higher radii is tantalizing , more data is needed to draw firm conclusions .
lars04 has a strong single peaked h i profile with a width at 50% of 150 kms@xmath12 ( see figure [ fig_gbt1 ] ) .
we classify it as an unconfused source ; although sdssj130757.13 + 542310.6 ( v@xmath22 = 9714 kms@xmath12 ) is located 5.56 to the southwest , it is not detected in our vla imaging ( see below ) .
lars04 has an irregular morphology in the uv and is a net lya absorber , just showing a weak asymmetric emission on top of a large , damped absorption profile within the hst cos aperture ( 2.5 in diameter ) .
paper ii has shown that weak diffuse @xmath0 emission is present , but scattered to very large galactocentric radii .
we recover a total single - dish h i flux of 1.6 jy kms@xmath12 which yields an h i mass of 7.3@xmath35 m@xmath3 at the adopted distance of 140 mpc .
the vla h i imaging ( see figure [ fig_vla_lars04 ] ) reveals extended emission with a median h i column density of 4.1@xmath42 @xmath33 [ maximum angular extent roughly twice that of the 71 ( 47.5 kpc ) beam ] ; interestingly , this extended h i component is in the same directional sense as the optical morphology .
higher spatial resolution observations of lars04 with the vla are both technically feasible ( sufficient source brightness ) and will be very useful in assessing the localized morphology and kinematics . there is coherent rotation of the source apparent at this low spatial resolution . specifically , the isovelocity contours are mostly parallel over @xmath13100 kms@xmath12 ( see figure [ fig_vla_lars04 ] ) .
using the same approach as for lars02 above , this implies a dynamical mass of m@xmath38 @xmath10 1.4@xmath3610@xmath44 m@xmath3 ( again , with no inclination correction and with the assumption of the h i rotating over one beam diameter ) .
however , based on the highly irregular optical morphology ( which figure [ fig_opta ] shows to be suggestive of an interaction ) , it is likely that the localized neutral gas kinematics are more complicated than simple disk rotation ; the coarse beam size is likely smoothing out the kinematic details of this system .
lars05 is an edge - on galaxy ( see figure [ fig_opta ] ) , with net @xmath0 absorption at very small radii and emission at larger radii , and at the physical scales relevant in this work [ see discussion in paper ii ] .
the gbt h i profile ( see figure [ fig_gbt1 ] ) shows a very weak signal that is near the noise level .
while the snr is low , the peaks in the spectrum may represent the top of the horns of a typical double - horn profile that is expected for a highly inclined disk galaxy .
we derive an h i flux integral of 0.55 jy kms@xmath12 , which , at a distance of 140 mpc , corresponds to an h i mass of 2.4@xmath35 m@xmath3 .
this galaxy has no known companions or contaminants in the gbt field of view .
lars06 is an irregular galaxy that has the weakest h@xmath7 and uv - based star formation rates in the entire lars sample ; it is a @xmath0 absorber on all physical scales .
the optical morphology suggests an interaction scenario , with the southern tail being uv luminous but h@xmath7 dim compared to the main star forming knots in the northern portion of the system .
interestingly , this is the lars system with the lowest measured nebular abundance [ see discussion in paper ii ] .
the gbt h i profile of the lars06 field has the highest snr of any of the lars galaxies .
this is almost certainly due to confusion within the beam .
as figure [ fig_lars06_field ] shows , lars06 is separated from the disk galaxy ugc10028 by only @xmath131 ( 43 kpc at the adopted distance of 148 mpc ) .
although the redshift - derived velocity values differ for these two galaxies by @xmath13 200 kms@xmath12 , the h i velocity of the total system is centered directly at the velocity of ugc10028 .
this single - dish h i spectrum is a classical double horn profile indicative of the rotation of a massive and inclined h i disk ; the width of the profile is @xmath13 370 kms@xmath12 at 50% of the maximum .
the h i flux of 4.4 jy kms@xmath12 yields a total h i mass of 2.3@xmath45 m@xmath3 ; this can be compared with the estimate of the stellar mass of lars06 , 2.1@xmath3610@xmath2 m@xmath3 , from paper ii .
while gas to stellar mass ratios of this size are not unreasonable , a comparison of the optical morphologies of ugc10028 and lars06 strongly suggests that the former system is contributing significantly to the flux in this field . from the sdss g - band luminosity
, we estimate the h i mass of ugc10028 as 4.1@xmath3610@xmath2 m@xmath3 .
this represents roughly 20% of the total mass seen in the system , but we caution that this might be an over or under estimation if the galaxy does not follow the same relation as other galaxies .
there are three other galaxies in the vicinity of lars06 that could also be contributing to the observed flux integral . as discussed in the caption of figure [ fig_lars06_field ] , lars06 and ugc10028
are also in close angular proximity to 2masx j15455278 + 4415470 , 2masx j15455157 + 4415310 , and sdss j154549.41 + 441539.2 .
based on the optical velocity , 2masx j15455278 + 4415470 appears to be roughly 1000 kms@xmath12 in the background of lars06 and ugc10028 . only photometric redshifts are available for 2masx j15455157 + 4415310 and sdss j154549.41 + 441539.2 , and each appears to also be at larger distances than lars06 and ugc10028 ; emission line spectroscopy will be required to determine the absolute velocities of these systems .
wide - bandwidth vla observations of this complex field will be able to localize the neutral gas components of each system ; this is an ideal target for subsequent high spatial resolution observations .
lars06 is also the only detected galaxy that lies significantly off of our observed gas fraction - stellar mass relationship ( see discussion in [ s6 ] below ) , most likely due to this contamination .
lars07 is a near edge - on disk system that shows one of the highest @xmath0 equivalent widths of the lars sample .
the gbt spectrum shown in figure [ fig_gbt1 ] is measured as a single peak profile with a linewidth of @xmath13 92 kms@xmath12 at 50% of maximum ; a second peak is separated from this main component by @xmath13120 kms@xmath12 . due to the low snr of this putative second peak
, we measure the h i properties of the source using only the higher snr component . with this assumption , lars07 has the narrowest linewidth of any galaxy in our sample .
the h i flux is 0.47 jy kms@xmath12 , corresponding to an h i mass of 2.9@xmath35 m@xmath3 at the adopted distance of 161 mpc .
lars08 is a metal - rich , low - inclination system ( see figure [ fig_optb ] and discussion in paper ii ) .
the distribution of uv , @xmath0 , and h@xmath7 emission is strongly asymmetric in the galaxy , being much more prominent on the western side of the disk than on the eastern side .
the system is a @xmath0 emitter on all spatial scales .
the gbt spectrum of lars08 ( see figure [ fig_gbt1 ] ) shows a broad , possibly double - peaked profile with a linewidth of 310 kms@xmath12 at 50% of the peak .
the single - dish h i integrated flux is 3.4 jy kms@xmath12 , yielding a total h i mass of 2.2@xmath45 m@xmath3 at the adopted distance of 160 mpc .
the galaxy appears to be isolated , with no known neighbors in the gbt beam with similar velocities .
the median column density of h i is 2.5@xmath42 @xmath33 . as figure [ fig_vla_lars08 ] shows , the detected h i emission
is highly localized to the 72 ( 56.9 kpc ) beam .
the source appears to be undergoing bulk rotation ; while the velocity field is formally unresolved , the nearly parallel isovelocity contours span @xmath1380 kms@xmath12 ( note that the first moment map represents intensity - weighted velocity ; for faint , unresolved sources one would expect a compressed velocity scale in the first moment compared to a spectrum ) .
taken as projected rotation with no inclination correction ( which is likely substantial , given the apparent optical inclination ) , the implied dynamical mass is @xmath131.1@xmath3610@xmath44 m@xmath3 .
higher spatial resolution h i observations of lars08 would be very interesting to pursue , since this system appears to be undergoing normal rotation and lacks obvious signatures of large - scale kinematic disturbances in our h i maps , even though cos spectroscopy shows significant outflow activity ( see details in rivera - thorsen etal .
, in preparation ) .
lars09 is an extended system with two prominent arms that are rich in h@xmath7-emitting star formation regions ( note that a careful inspection of figure [ fig_optb ] shows that there is a prominent foreground star at the southern end of the disk ) .
the optical morphology is consistent with the source being either a loose spiral or an interacting pair .
the system is a global @xmath0 emitter at large scales ( r @xmath1 10 kpc ) ; the @xmath0 morphology is similar to that of the optical / uv , although more extended in all directions from the disk .
the gbt spectrum ( see figure [ fig_gbt1 ] ) is broad ( linewidth of 270 kms@xmath12 at 50% of the peak ) and relatively low snr .
the total h i single - dish flux is 1.2 jy kms@xmath12 , which gives a mass of 1.2@xmath45 m@xmath3 at a distance of 200 mpc .
the profile is not contaminated by any other known sources within the gbt beam .
the median h i column density is 1.2@xmath42 @xmath33 the vla images presented in figure [ fig_vla_lars09 ] show that the source is essentially unresolved by the 59 ( 57.2 kpc ) beam .
the h i maximum is co - spatial with the optical body ; there is weak evidence for extended gas that is only slightly larger than the beam size ( see , for example , the extended material to the east - southeast of the optical body in figure [ fig_vla_lars09 ] ) .
however , we do not interpret this as evidence of extended tidal structure due to the marginal extension compared to the beam size .
the intensity weighted velocity field of lars09 covers a large range ( @xmath1200 kms@xmath12 ) and is severely disturbed .
there are some isovelocity contours that are roughly parallel in the northwest region of the system ; however , at the position of the optical body these contours deviate significantly from regularity .
the kinematic information in these images suggests that the interaction scenario for the optical morphology may be appropriate .
the presence of diffuse @xmath0 emission in this source , combined with the irregular kinematics and morphology , make this a prime target for higher resolution h i observations .
lars10 is an irregular source that shows high @xmath0 equivalent width emission arising from a diffuse component that extends beyond the optical body of the source [ see discussion in paper i ] .
the optical morphology is consistent with an advanced merger state .
the gbt profile is broad ( @xmath13280 kms@xmath12 at 50% of maximum ) , although of low snr ; the total flux integral of 0.29 jy kms@xmath12 corresponds to a total h i mass of @xmath134.1@xmath34 m@xmath3 at the adopted distance of 250 mpc .
the weak h i flux integral will make high resolution interferometric observations of lars10 challenging .
lars11 is a dramatic edge - on galaxy that is in close angular proximity to a field spiral galaxy ( cgcg046 - 044ned01 at 14@[email protected]@xmath48 , + 062826 ) .
lars11 appears to be in the foreground of this object ( lars11 is at z@xmath22 = 0.0843 , while cgcg046 - 044ned01 has a photometric redshift derived from sdss of @xmath130.1 ) ; this is apparent by the disk of lars11 visually obstructing one of the spiral arms of cgcg046 - 044ned01 .
figure [ fig_lars11_field ] shows that lars11 is separated from two other field galaxies with similar velocities by less than the gbt beam : sdssj140401.00 + 062901.7 at 14@[email protected]@xmath48 @xmath43062902 , and sdssj140353.36 + 062504.8 at 14@[email protected]@xmath50 + 062505 .
given this complex field , we classify the gbt profile of lars11 as potentially confused .
further , rfi was present near the expected frequency of the h i spectral line from lars11 ; this required extensive flagging .
the h i profile of lars11 shown in figure [ fig_gbt1 ] is broad ( w@xmath18 = 260 kms@xmath12 ) and is distributed in multiple peaks .
this may represent a broad rotation profile from lars11 , or it may be indicative of multiple sources contributing to the h i flux within the beam . assuming that the full h i flux integral of 0.75 jy kms@xmath12 is only associated with lars11 , the implied neutral hydrogen mass is 2.3@xmath3610@xmath44 m@xmath3 at the adopted distance of 360 mpc .
vla imaging of lars11 would be very useful for localizing the h i emission from the various sources in this complex field .
due to resonant scattering with neutral hydrogen atoms , the radiative transport of @xmath0 photons may well have its strongest dependence on the characteristics of the h i that surrounds the star forming regions in which the photons are produced . with the present h i spectroscopy and imaging
, we now have a first order understanding of the neutral gas contents of many of the lars galaxies .
comparing the qualities of the global h i reservoirs with the detailed characteristics of each system derived from hst imaging allows us to probe what roles h i kinematics and density play in governing @xmath0 radiative transport .
we focus on four global properties derived from the present h i observations of the 11 lowest - redshift lars galaxies ( those for which the gbt spectra provide meaningful measurements or limits ) ; for the galaxies where it is available ( lars 02,03,04,08,09 ) we use the total fluxes and masses vla images preferentially over the gbt data . in general the vla data
recovers 30% less flux than the gbt data , which contributes to 6% shorter linewidths . as mentioned in section [ s4 ] , this is most likely due to a combination of decreased surface brightness sensitivity in the vla data and decreased contamination from other sources .
first , the h i linewidth ( specifically , the width at 50% of the maximum w@xmath25 ) is a distance - independent variable that has been shown to correlate with absolute magnitude and thus to serve as a proxy for mass in star - forming disk galaxies @xcite . for rotationally dominated galaxies , a larger h i linewidth
can often be interpreted as an indicator of a more massive galaxy . to know with certainty , a detailed study of the dynamics and complete censuses of the baryonic components of individual galaxies
is needed . in our case , the galaxies with irregular morphology may have linewidths that are significantly increased due to kinematic disturbances and merging activity .
the second global parameter we examine is the total h i mass of each lars galaxy .
this parameter is of course related to the h i linewidth , but not in a one to one sense .
the lars galaxies span a range of morphological types , including violent interactions ( e.g. , lars03 ) , disk - dominated spirals ( e.g. , lars05 and lars11 ) , and irregulars ( e.g. , lars04 , lars06 ) .
while for disk dominated systems the h i linewidth and h i mass should be closely correlated , for the other types of systems the h i line profile may not only be indicative of rotation ; h i gas from multiple components may create irregular single - dish profiles .
examples of this are clearly seen in the gbt spectra ( figure [ fig_gbt1 ] ) and in the vla images ( see , e.g. , figure [ fig_vla_lars03 ] ) .
third , we examine the offset in velocity between the centroid of the h i profile and the known systemic velocity of each lars galaxy as derived from optical spectroscopy . assuming that such offsets are caused by large - scale kinematic deviations from regularly rotating galaxy disks ( e.g.
, tidal interaction , gas outflow or infall ) , this offset parameter can be considered a rough proxy for the presence or absence of large - scale motions of neutral gas . as discussed in @xcite and above for lars03
, we find evidence for extended tidal structure in some @xmath0-emitting galaxies .
however , we also find systems that are strong @xmath0 emitters that do not show such extended neutral gas components ( e.g. , lars02 ; see [ s5.2 ] above ) . finally , we examine the gas mass fraction defined as the h i mass divided by the stellar mass .
numerous works have shown that normal , star - forming disk galaxies populate a robust trend of decreasing gas fraction with increasing stellar mass ( see recent results in huang etal . 2012 , papastergis etal . 2012 , peeples etal .
2014 , and various references therein ) . in figure [ fig_fgas ]
we show the gas fraction as a function of total stellar mass ( m@xmath51 ) derived from 2-component sed modeling to the hst data ( see paper ii and hayes etal .
2009 for details ) . except for lars06 ,
whose h i flux integral is likely strongly contaminated by the nearby field spiral ugc10028 ( see discussion in [ s5.06 ] and figure [ fig_lars06_field ] ) , the lars galaxies closely follow the derived relations for nearby galaxies as found in @xcite and @xcite .
it is important to remain cognizant of the possible confusion within the gbt beam for some of the sources without vla data . in particular ,
the h i properties of lars06 are almost certainly contaminated by ugc10028 , and those of lars11 may be contaminated by sdssj140401.00 + 062901.7 and/or sdssj140353.36 + 062504.8 .
if these sources are in fact contaminated , then the h i linewidths and masses will be overestimated ; the velocity offsets may or may not change .
we examine the relationships of these four h i - based quantities with seven global properties derived from hst data in paper ii : the @xmath0/h@xmath7 flux ratio , the @xmath0 luminosity , the h@xmath7/h@xmath24 flux ratio , the @xmath0 escape fraction , the @xmath0 equivalent width , the @xmath0 extension parameter ( @xmath0 @xmath52 ) , and the sfr per unit area within the petrosian radius ( @xmath53 = 0.2 ; see paper ii ) . as discussed in detail in @xcite , there is a complex interrelation between these ( and other ) properties in the process of @xmath0 radiative transport ; the total @xmath0 luminosity of a given galaxy does not correlate in a statistically significant way with any individual quantity .
further , there is significant variation in these values as a function of position within an individual galaxy ; some galaxies appear as net absorbers within the disk and as net emitters when the diffuse @xmath0 halo is included .
nevertheless , we seek statistical comparisons with many of the _ global _ quantities of both @xmath0 and h i. these comparisons most closely match detections of higher - redshift @xmath0 emitters and will constrain one of the many poorly understood aspects of @xmath0 propagation and detection in the distant universe . from a simplified interpretative standpoint
, these seven global quantities can be related as follows .
the h@xmath7/h@xmath24 ratio indicates extinction ; deviations above the intrinsic ratio of 2.86 @xcite indicate non - zero e(b@xmath8v ) values .
these e(b@xmath8v ) values can in turn be used to derive the intrinsic @xmath0 luminosity of a system based on its observed ( and extinction corrected ) h@xmath7 luminosity and also to estimate the @xmath0 escape fraction .
the @xmath0/h@xmath7 ratio has a ( case b ) recombination value of 8.7 .
lower values can indicate stronger suppression of @xmath0 versus h@xmath7 ( e.g. , by attenuation from dust , or by resonant scattering away from the production site ) ; super - recombination values are seen in a few localized regions of the lars galaxies , as well as in some of the large scattered @xmath0 halos surrounding some of the systems ( see paper i and paper ii ) . the @xmath0 equivalent width is related to the recent star formation history of the galaxy or region ; global values of @xmath13100 indicate constant star formation over @xmath13100 myr timescales , while values exceeding 250 occur only in the very youngest burst episodes ( schaerer etal .
2003 and raiter etal . 2010 . )
we present scatter plot comparisons of the four global h i - derived properties and seven global hst - derived properties in figures [ fig_compare_mass ] , [ fig_compare_line ] , [ fig_compare_off ] , and [ fig_compare_gasfrac ] .
we restrict the @xmath0 properties to positive values , which has the effect of setting ew and luminosity measurements equal to zero .
we mark this region on the plots with a hashed region .
we use the spearman @xmath54 correlation coefficient to quantify possible monotonic correlations between these properties ; a perfect correlation or anti - correlation between two properties will have @xmath54 values of @xmath431 or @xmath81 , respectively .
each panel of figures [ fig_compare_mass ] through [ fig_compare_off ] shows the corresponding @xmath54 value . to estimate the errors on these correlation coefficients we resample the properties 1000 times with a random realization of the associated errors and measure
, recording the correlation coefficients each time . for each correlation
we show the value from the sample of uncontaminated detections as the primary result , and include the value for the entire sample including upper limits and confused detections below .
overall the results do not show strong evidence of correlation between properties ( @xmath54 @xmath55 @xmath80.6 or @xmath54 @xmath1 @xmath430.6 ) .
we present all correlations in table [ table_correlations ] , along with the dispersion and discuss the correlations for each h i property below .
the h i line width is significantly anti - correlated with the @xmath0 extension parameter , @xmath56 ( see figure [ fig_compare_line ] ) .
this is the strongest evidence of correlation in the entire sample .
the h i line width is positively correlated with the h@xmath57/h@xmath58 ratio ( see further discussion below ) .
two factors are possibly at play here .
either larger h i quantities scatter @xmath0 photons to extended radii , or interactions increase both the linewidth and preferentially block @xmath0 photons at short radii .
two anti - correlations , and a positive correlation appear comparing the h i mass of our sample of well detected galaxies ( see figure [ fig_compare_mass ] ) .
the escape fraction ( @xmath54 = @xmath80.63 ) , and @xmath0 ew ( @xmath54 = @xmath80.60 ) show tentative evidence for anti - correlation , which do not appear when including the marginal detections .
interestingly , the h@xmath57/h@xmath58 ratio shows strong evidence of correlation ( @xmath54 = 0.75 ) .
it has been known for some time that more massive star forming galaxies ( hence those with larger h i masses and linewidths ) are dustier , and that the balmer decrement scales accordingly ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
how this correlation may be connected to the underling @xmath0 propagation is not entirely clear . we see no conclusive evidence of correlations with any of the global @xmath0 properties using either the whole sample or the positive detections .
the strongest correlation is with the h@xmath7/h@xmath24 ratio ( @xmath54 = -0.53 ) .
we find no significant correlation between the velocity offsets and the various @xmath0 properties .
this might be due to our velocity smoothing and the large extent over which we are probing the velocity information .
the strongest correlation is with with the h@xmath7/h@xmath24 ratio ( @xmath54 = 0.50 ) .
we have presented new gbt spectroscopy and vla h i spectral line imaging of the galaxies in the `` lyman alpha reference sample '' ( `` lars '' ) .
lars is a comprehensive , multi - wavelength study of 14 uv - selected star - forming galaxies that functions as a local benchmark for studies of @xmath0 emission and absorption at higher redshifts .
the centerpiece is a suite of hst images that allows us to study @xmath0 emission and related quantities on a spatially resolved basis .
the preliminary results presented in @xcite showed that most of the lars galaxies have large , diffuse halos of @xmath0 emission that exceed the sizes of the stellar populations ; this is strong evidence for the importance of resonant scattering in the neutral hydrogen gas component .
the hi structure and content are important pieces for understanding the origin of the ly - alpha halos and their different sizes .
our current hi angular resolution probes the large - scale global properties rather than those in the immediate vicinity of the ly - alpha halo regions imaged with hst .
nevertheless , these provide important insights into the galaxy structure and environment .
the lars survey products are described in paper i of this series @xcite ; the integrated properties of the lars galaxies as derived from the hst imaging are presented in paper ii of this series @xcite . in this manuscript
we present a first exploration of the h i properties of the lars galaxies . using data from the gbt 100 m telescope ,
we have presented direct measurements of the neutral gas contents of 11 of the 14 galaxies ; we place limits on the three most distant ( z @xmath1 0.1 ) lars galaxies .
these profiles show a wide variety of structure .
some systems harbor multi - component profiles that are indicative of interaction or the presence of nearby companions ; others show more simple single h i peaks .
two sources ( lars06 and lars11 ) are likely confused with nearby galaxies .
we fit each of these profiles in order to derive global line properties , including systemic velocity , linewidths , h i flux integrals , and h i masses . for a subset of five of the lars galaxies , we also present new , low - resolution h i spectral line imaging obtained with the vla .
these h i images allow us to localize the h i associated with each galaxy .
three of the systems ( lars02 , lars08 , and lars09 ) are unresolved at this angular resolution , while lars04 is resolved by a few beams .
lars03 is highly resolved by the present data ; we have discovered an enormous tidal structure that contains more than 10@xmath2 m@xmath3 of h i , and that extends more than 160 kpc from the main interacting galaxies in the lars03 system . based on the recovered flux integrals from these vla data , each of the five systems presented here would be amenable to higher spatial resolution observations .
the vla c and b configurations would offer a factor of roughly 4 and 10 improvement in resolution respectively , but at the cost of longer integration times .
in particular lars03 presents an interesting case where we have unequivocal evidence of kinematic disturbances on large scales , yet no accompanying boost relative to non - disturbed galaxies in @xmath0 escape fraction . a high resolution
follow up with the vla and future radio telescopes , especially coupled with larger maps of the @xmath0 flux , would disentangle the local versus global kinematic effects , and probe the @xmath0 escape on physical scales larger than the @xmath1310 kpc already probed in the hst imaging . using these new gbt and vla data , as well as the results derived from hst imaging and presented in paper ii , we compare various global h i and @xmath0 properties . while the sample is small , we find a few intriguing correlations in our robustly detected galaxies : the h i linewidths are strongly anti - correlated with @xmath56 ( @xmath54 = @xmath80.81 ) , while the total h i masses show anti - correlations with escape fraction and @xmath0 ew . additionally , both the h i linewidths and the h i masses are each correlated with the h@xmath57/h@xmath58 ratio . these intriguing but tentative ( anti-)correlations are in general agreement with a significant dependence of @xmath0 propagation on the total mass and thus the h i mass of the source , and is in general agreement with previous work by @xcite which found that the escape fraction decreases with increasing virial mass
. these results also support the general trend that @xmath0 properties appear to correlate with total galaxy mass , as found in paper ii .
that work showed that @xmath0 emission ( quantified by @xmath0 equivalent width , escape fraction , etc . )
is systematically larger in lower mass galaxies . given the variety of properties that scale roughly with mass ( e.g. , metallicity , dust content , star formation rate , uv colors , etc . )
, a larger statistical sample of galaxies will be needed in order to better quantify these trends .
the famous metal - poor blue compact dwarf galaxy izw18 reminds us that the trend of lower mass galaxies having larger @xmath0 luminosities or escape fractions is not always true .
while the system has an extremely low nebular oxygen abundance @xcite , an appreciable star formation rate @xcite , and a substantial uv luminosity @xcite , @xcite showed that izw18 is in fact a @xmath0 absorber .
this process was explained by numerous scattering events in the neutral , and , most importantly , static ism even in the complete absence of dust @xcite .
future local @xmath0 studies should seek to expand the number of low - mass systems with analyses similar to the ones presented here for lars .
although it is tempting to view the increasing @xmath0/h@xmath7 with decreasing mass as a result of less gas mass being available for scattering , this theory must be cast in the light of results that suggest that gas fraction increases with increasing redshift ( e.g. , * ? ? ?
* ) , and that lower mass galaxies tend to be more gas - rich @xcite and less dusty ( e.g. , * ? ? ?
in figure [ fig_fgas ] we see that the lars galaxies fit the general trend with gas mass fraction . while @xmath0 halos seem to appear in systems with a range of h i linewidths
, large @xmath0/h@xmath57 values only appear in galaxies with h i smaller linewidths .
large @xmath0 halos are also correlated with less massive galaxies , and with increasing velocity offsets .
we interpret this to mean that low h i linewidths are necessary but not sufficient for @xmath0 escape , while larger line widths contribute to the overall effects of @xmath0 destruction by dust , extinction , and scattering .
the bulk of the evidence presented by previous @xmath0 work shows the importance of h i properties , especially kinematic details , in determining the escape of photons ( e.g. mas - hesse etal .
this process is mostly governed by local effects , and it is unclear at this time what global h i properties are required for the escape and propagation of @xmath0 photons . the asymmetric gbt profiles and evidence of @xmath0 emission in the presence of large column densities of h i both point to interesting sub - resolution kinematic and density effects . a larger sample of galaxies with data like those presented here will allow us to place these correlations on a more statistically robust footing . the hst elars program has been approved for 54 orbits and contains 28 more galaxies within our nominal sensitivity range of z @xmath13 0.08 .
in addition , higher resolution follow up observations will allow us to determine kinematic details on more applicable scales for a number of our galaxies .
this program will likely be the largest statistical sample of both h i and @xmath0 detected galaxies until new advances of radio telescopes , such as the square kilometer array , push h i detections to 20 - 100 kpc scales at z @xmath59 .
the authors thank the anonymous referee for insightful comments that improved the quality of this manuscript .
j.m.c . and s.p .
thank sung kyu kim and macalester college for research support .
j.m.c . would like to thank the instituto nazionale di astrofisica and the osservatorio astronomico di padova for their hospitality during a productive sabbatical leave .
g. . is a royal swedish academy of science research fellow ( supported by a grant from the knut & alice wallenbergs foundation ) .
m.h . acknowledges the support of the swedish research council ( vetenskapsrdet ) and the swedish national space board ( sbsb ) .
this work was supported by the swedish research council and the swedish national space board .
is ended by a postdoctoral unam grant .
j.m.m - h is funded by spanish mineco grants aya2010 - 21887-c04 - 02 and aya2012 - 39362-c02 - 01 .
i.o . was supported by the sciex fellowship of the rectors conference of swiss universities .
p.l . acknowledges support from the erc - stg grant eggs-278202 .
d.k is supported by cnes ( centre national detudes spatiales ) for the hst project .
this investigation has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration , and nasa s astrophysics data system .
this investigation has made use of data from the sloan digitized sky survey ( sdss ) .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , university of cambridge , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington .
osterbrock , d. e. 1989 , research supported by the university of california , john simon guggenheim memorial foundation , university of minnesota , et al .
mill valley , ca , university science books , 1989 , 422 p. , lccccccc lars01 & mrk 259 & 13:28:44.50 & + 43:55:50 & 8394 & 0.028 & 120 & dwarf irregular + lars02 & sdssj090704.88 + 532656.6 & 09:07:04.88 & + 53:26:56 & 8934 & 0.030 & 130 & dwarf irregular + lars03 & arp 238 & 13:15:35.60 & + 62:07:28 & 9204 & 0.031 & 130 & merger + lars04 & sdssj130728.45 + 542652.3 & 13:07:28.45 & + 54:26:52 & 9743 & 0.033 & 140 & irregular + lars05 & mrk 1486 & 13:59:50.91 & + 57:26:22 & 10133 & 0.034 & 150 & edge - on dwarf spiral + lars06 & kissr 2019 & 15:45:44.52 & + 44:15:51 & 10223 & 0.034 & 150 & dwarf irregular + lars07 & iras 1313 + 2938 & 13:16:03.91 & + 29:22:54 & 11332 & 0.038 & 170 & dwarf edge - on spiral + lars08 & sdssj125013.50 + 073441.5 & 12:50:13.50 & + 07:34:41 & 11452 & 0.038 & 170 & spiral + lars09 & iras 0820 + 2816 & 08:23:54.96 & + 28:06:21 & 14150 & 0.047 & 210 & edge - on spiral + lars10 & mrk 0061 & 13:01:41.52 & + 29:22:52 & 17208 & 0.057 & 260 & merger + lars11 & sdssj140347.22 + 062812.1 & 14:03:47.22 & + 06:28:12 & 25302 & 0.084 & 380 & edge - on spiral + lars12 & sbs 0934 + 547 & 09:38:13.49 & + 54:28:25 & 30608 & 0.102 & 470 & dwarf irregular + lars13 & iras 0147 + 1254 & 01:50:28.39 & + 13:08:58 & 43979 & 0.147 & 700 & irregular + lars14 & sdssj092600.40 + 442736.1 & 09:26:00.40 & + 44:27:36 & 54172 & 0.181 & 880 & dwarf + [ table_basic ] lcccccccccc lars01 &
8339 @xmath115 & 160 @xmath1110 & 180 @xmath1116 & @xmath855 @xmath11 5 & 0.70 @xmath110.07 & 2.5 @xmath110.3 & ( 3.1 , 6.5 ) & 1.1 & 2 & d + lars02 & 8940 @xmath118 & 140 @xmath1117 & 150 @xmath1126 & 3 @xmath11 8 & 0.70 @xmath110.07 & 2.8 @xmath110.3 & ( 2.2 , 2.9 ) & 1.8 & 3 & d + lars03 & 9421 @xmath1120 & 310 @xmath1139 & 380 @xmath1162 & 230 @xmath11 20 & 1.6 @xmath110.2 & 6.9 @xmath110.7 & ( 3.1 , 4.9 ) & 1.2 & 3 & d + lars04 & 9740 @xmath1116 & 150 @xmath1131 & 260 @xmath1149 & @xmath83 @xmath11 16 & 1.6 @xmath110.2 & 7.7 @xmath110.8 & ( 5.1 , 11 . ) & 1.0 & 3 & s + lars05 & 10117 @xmath1114 & 160 @xmath1128 & 170 @xmath1144 & @xmath810 @xmath11 14 & @xmath55 0.55 & @xmath55 2.9 & ( 1.3 , 1.5 ) & 1.1 & 1 & i + lars06 & 10378 @xmath1112 & 370 @xmath113 & 390 @xmath115 & 155 @xmath11 2 & 4.4 @xmath110.4 & 23 .
@xmath112 & ( 9.5 , 18 . ) & 1.1 & 3 & d + lars07 &
11315 @xmath1116 & 100 @xmath1132 & 170 @xmath1150 & @xmath810 @xmath11 16 & 0.47 @xmath110.05 & 3.1 @xmath110.3 & ( 3.3 , 6.2 ) & 0.86 & 3 & s + lars08 & 11468 @xmath1123 & 310 @xmath1147 & 470 @xmath1173 & 20 @xmath11 23 & 3.4 @xmath110.3 & 22 . @xmath112 & ( 5.7 , 11 . ) & 0.91 & 3 & i + lars09 & 14020 @xmath1146 & 270 @xmath1192 & 490 @xmath11140 & @xmath8130 @xmath11 46 & 1.2 @xmath110.1 & 13 .
@xmath111 & ( 2.8 , 7.3 ) & 0.73 & 3 & s + lars10 & 17284 @xmath1169 & 280 @xmath11140 & 380 @xmath11220 & 70 @xmath11 69 & 0.29 @xmath110.03 & 4.5 @xmath110.5 & ( 1.1 , 2.1 ) & 0.69 & 3 & c + lars11 & 25366 @xmath1156 & 260 @xmath1111 & 290 @xmath1118 & 65 @xmath11 6 & 0.75 @xmath110.08 & 26 .
@xmath113 & ( 2.9 , 6.4 ) & 0.89 & 3 & d + lars12 & ( 30609 ) & ( 264 ) & ( 290 ) & n / a & @xmath55 2.9 & @xmath55 150 & n / a & 3.4 & 3 & n + lars13 & ( 43980 ) & ( 264 ) & ( 290 ) & n / a & @xmath55 33 & @xmath55 3800 & n / a & 38 . & 5 & n + lars14 & ( 54172 ) & ( 264 ) & ( 290 ) & n / a & @xmath55 1.7 & @xmath55 310 & n / a & 2.0 & 1 & n + [ table_gbt_props ] lccccc lars02 & 2013 apr 3 - 4 & 0542 + 498=3c147 & j0834 + 5534 & 8.28 @xmath11 0.014 & 225.9 + lars02 & 2013 apr 5 - 6 & 0542 + 498=3c147 & j0834 + 5534 & 8.35 @xmath11 0.013 & 225.5 + lars03 & 2013 apr 14 &
1331 + 305=3c286 & j1400 + 6210 & 4.42 @xmath11 0.017 & 200.4 + lars04 & 2013 mar 30 & 1331 + 305=3c286 & j1252 + 5634 & 2.29 @xmath110.025 & 120.4 + lars08 & 2013 apr 16 & 1331 + 305=3c286 & j1254 + 1141 & 0.858 @xmath11 0.0058 & 119.8 + lars09 & 2013 mar 3 - 4 & 0542 + 498=3c147 & j0741 + 3112 & 2.02 @xmath11 0.0056 & 306.2 + lars09 & 2013 mar 26 - 27 & 0542 + 498=3c147 & j0741 + 3112 & 2.09 @xmath11 0.050 & 281.1 + [ table_vlaobs ] lcccccccc lars02 & 8960 @xmath112 & 140 @xmath114 & 150 @xmath116 & 26 @xmath11 6 & 0.36 @xmath110.036 & 1.4 @xmath110.15 & 0.50 & 2.6 + lars03 & 9520 @xmath1110 & 170 @xmath1120 & 300 @xmath1132 & 316 @xmath11 32 & 2.2 @xmath110.22 & 7.8 @xmath110.94 & 1.0 & 0.96 + lars04 & 9760 @xmath112 & 180 @xmath114 & 190 @xmath116 & 15 @xmath11 6 & 1.2 @xmath110.12 & 5.8 @xmath110.58 & 1.0 & 4.8 + lars08 & 11450 @xmath112 & 240 @xmath114 & 250 @xmath117 & -1 @xmath11 7 & 1.8 @xmath110.18 & 12 .
@xmath111.2 & 1.0 & 2.6 + lars09 & 14070 @xmath119 & 310 @xmath1118 & 330 @xmath1128 & -79 @xmath11 28 & 0.60 @xmath110.060 & 6.2 @xmath110.62 & 0.50 & 1.2 + [ table_vla_props ] lcccccccc lars01 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath60 & [email protected] + lars02 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath61 & [email protected] + lars03 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath62 & [email protected] + lars04 & 0.00 & [email protected] & 0.000 & [email protected] & [email protected] & 0.00 & & [email protected] + lars05 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath63 & [email protected] + lars06 & 0.00 & [email protected] & 0.000 & [email protected] & [email protected] & 0.00 & & [email protected] + lars07 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath64 & [email protected] + lars08 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath65 & 93.3@xmath1111 + lars09 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath12.85 & [email protected] + lars10 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath66 & [email protected] + lars11 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath67 & [email protected] + lars12 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath68 & [email protected] + lars13 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath69 & [email protected] + lars14 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath70 & [email protected] + [ table_uv_properties ] ccccc @xmath0/h@xmath7 & @xmath80.52 [ 0.2 ] & @xmath80.53 [ 0.06 ] & @xmath80.083 [ 0.08 ] & 0.47 [ 0.4 ] + lum(@xmath0 ) & @xmath80.27 [ 0.1 ] & 0.05 [ 0.1 ] & 0.17 [ 0.04 ] & 0.00 [ 0.1 ] + h@xmath7/h@xmath24 & 0.67 [ 0.1 ] & 0.75 [ 0.1 ] & 0.50 [ 0.05 ] & @xmath80.45 [ 0.3 ] + f@xmath71 & @xmath80.53 [ 0.2 ] & @xmath80.63 [ 0.1 ] & @xmath80.15 [ 0.1 ] & 0.40 [ 0.4 ] + @xmath0 ew & @xmath80.33 [ 0.2 ] & @xmath80.60 [ 0.1 ] & 0.10 [ 0.1 ] & 0.45 [ 0.3 ] + @xmath56 & @xmath80.81 [ 0.1 ] & @xmath80.45 [ 0.1 ] & @xmath80.44 [ 0.2 ] & 0.57 [ 0.1 ] + sfr area@xmath12 & @xmath80.15 [ 0.1 ] & @xmath80.40 [ 0.2 ] & @xmath80.067 [ 0.04 ] & 0.43 [ 0.2 ] + [ table_correlations ] , see paper ii ) . the number in the upper left panel corresponds to the lars identification number ( e.g. , 01 is lars01 ) and each panel in a row is the same lars galaxy . in the right column are sdss color images of the lars target galaxies and the surrounding areas .
the fields of view ( which are tailored in each frame to show detail ) is larger in the right column ( 1.69 on a side ) .
b band magnitudes , given in blue , were obtained by converting sdss magnitudes using equations provided by sdss and found in @xcite .
these values are corrected for foreground extinction using values from @xcite .
, height=672 ] | we present new h i imaging and spectroscopy of the 14 uv - selected star - forming galaxies in the lyman alpha reference sample ( lars ) , aimed for a detailed study of the processes governing the production , propagation , and escape of @xmath0 photons .
new h i spectroscopy , obtained with the 100 m green bank telescope ( gbt ) , robustly detects the h i spectral line in 11 of the 14 observed lars galaxies ( although the profiles of two of the galaxies are likely confused by other sources within the gbt beam ) ; the three highest redshift galaxies are not detected at our current sensitivity limits .
the gbt profiles are used to derive fundamental h i line properties of the lars galaxies .
we also present new pilot h i spectral line imaging of 5 of the lars galaxies obtained with the karl g. jansky very large array ( vla ) .
this imaging localizes the h i gas and provides a measurement of the total h i mass in each galaxy . in one system ,
lars03 ( ugc8335 or arp238 ) , vla observations reveal an enormous tidal structure that extends over 160 kpc from the main interacting systems and that contains @xmath110@xmath2 m@xmath3 of h i. we compare various h i properties with global @xmath0 quantities derived from hst measurements .
the measurements of the @xmath0 escape fraction are coupled with the new direct measurements of h i mass and significantly disturbed h i velocities .
our robustly detected sample reveals that both total h i mass and linewidth are tentatively correlated with key @xmath0 tracers .
further , on global scales , these data support a complex coupling between @xmath0 propagation and the h i properties of the surrounding medium . |
basaltic asteroids are small bodies connected to the processes of heating and melting that may have led to the mineralogical differentiation in the interiors of the largest asteroids .
therefore , a precise knowledge of the inventory of basaltic asteroids may help to estimate how many differentiated bodies actually formed in the asteroid main belt , and this in turn may provide important constraints to the primordial conditions of the solar nebula .
in the visible wavelengths range , the reflectance spectrum of basaltic asteroids is characterized by a steep slope shortwards of 0.70 @xmath1 m and a deep absorption band longwards of 0.75 @xmath1 m .
asteroids showing this spectrum are classified as v - type in the usual taxonomies ( e.g. bus & binzel , 2002 ) .
a few years ago , most of the known v - type asteroids were members of the vesta dynamical family , located in the inner asteroid belt semi - major axis @xmath2 au. this family formed by the excavation of a large crater ( thomas et al . , 1997 ; asphaug , 1997 ) on the surface of asteroid ( 4 )
vesta , which is the only known large asteroid diameter @xmath3 km to show a basaltic crust ( mccord et al . , 1970 ) . nowadays ,
however , several v - type asteroids have been identified in the inner belt but outside the vesta dynamical family ( burbine et al . , 2001 ; florczak et al . , 2002 ; alvarez - candal et al . ,
basaltic asteroids have also been found in the middle main belt @xmath4 au ( binzel et al . , 2006 ; roig et al . , 2007 ) , as well as among the near earth asteroids ( nea ) population ( mcfadden et al . , 1985 ; cruikshank et al . , 1991 ; binzel et al . , 2004 ;
duffard et al . ,
recent works ( carruba et al . , 2005 , 2007 ; nesvorn et al . , 2007 ;
roig et al . , 2007 )
provide evidence that many of these v - type asteroids may be former members of the vesta family , that reached their present orbits due to long term dynamical evolution .
the exception is asteroid ( 1459 ) magnya , the only basaltic object so far discovered in the outer belt @xmath5 au ( lazzaro et al . , 2000 ) .
this asteroid is too far away from the vesta family and it is also too big @xmath6 - 30 km to have a real probability of being a fragment from the vesta s crust ( michtchenko et al . , 2002 ) . beyond the existence of ( 4 ) vesta and the v - type asteroids related to the vesta dynamical family ,
the paucity of intact differentiated asteroids and of their fragments observed today in the main belt is an strong constraint to the formation scenario of basaltic material .
the sample of iron meteorites collected in the earth indicates that they would come from the iron core of dozens of differentiated parent bodies .
however , there are very few olivine - rich asteroids ( classified as a - type ) which are assumed to come from the mantle of differentiated bodies , and only one asteroid , ( 1459 ) magnya , is known to sample the basaltic crust of a differentiated parent body other than ( 4 ) vesta .
finally , the other main belt asteroid families , which formed from the disruption of over fifty @xmath7 km asteroids , show little spectroscopic evidence that their parent bodies were heated enough to produce a distinct core , mantle and crust ( cellino , 2003 ) . aiming to establish if other v - type asteroids might be found together with ( 1459 ) magnya in the outer belt , thus giving support to the existence of a differentiated parent body in that part of the belt , roig & gil - hutton ( 2006 ) used the five band photometry from the 3rd release of the sloan digital sky survey moving objects catalog ( sdss - moc ; ivezi et al . , 2001 ; juri et al . ,
2002 ) to identify candidate v - type asteroids . among 263 candidates that are
not members of the vesta dynamical family , roig & gil - hutton found five possible v - type asteroids in the outer belt : ( 7472 ) kumakiri , ( 10537 ) 1991 ry16 , ( 44496 ) 1998 xm5 , ( 55613 ) 2002 ty49 , and ( 105041 ) 2000 ko41 .
however , these findings need to be confirmed by accurate spectroscopic observations .
the aim of this work is to describe the visible spectroscopic observations of two of these candidates : ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 .
our goal is to provide a more reliable taxonomic classification of these asteroids indicating that _ they would the second and third basaltic asteroids discovered up to now in the outer belt_. our observations also reveal certain peculiarities of their spectra that deserve special attention in future studies .
last but not least , our results help to validate the approach of roig & gil - hutton ( 2006 ) to predict v - type asteroids .
it is worth recalling that a similar study has been performed by roig et al .
( 2007 ) , who used visible spectroscopic observations taken at the gemini observatory to confirm the classification of two candidate v - type asteroids in the middle belt : ( 21238 ) 1995 wv7 and ( 40521 ) 1999 rl95 . in sect .
2 , we describe the observations and the reduction procedures . in sect .
3 , we present and discuss the results obtained .
finally , sect .
4 is devoted to conclusions .
low resolution spectroscopy of ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 were obtained on november 14 , 2006 , as part of a 4 nights observational run , using the calar alto faint object spectrograph ( cafos ) at the 2.2 m telescope in calar alto observatory , spain .
the prime aim of the run was to characterize v - type asteroids inside and outside the vesta family .
asteroid ( 7472 ) kumakiri was observed again on december 29 , 2006 , using the same instrument and telescope , under director s discretionary time ( ddt ) .
table [ observ ] summarizes the observational circumstances .
cafos is equipped with a 2048@xmath82048 ccd detector site-1d ( pixel size 24 @xmath1m / pixel , plate scale 0.53``/pixel ) .
we used the r400 grism allowing to obtain an observable spectral range between 0.50 and 0.92 @xmath1 m . to remove the solar component of the spectra and obtain the reflectance spectra , the solar analog stars hd 191854 , hd 20630 and hd 28099 ( hardorp , 1978 )
were also observed at similar airmasses as the asteroids . in order to estimate the quality of each night , at least two solar analogs
were observed per night and we verified that the ratios between the corresponding spectra show no significant variations .
bias frames , spectral dome flat fields and calibration lamps spectra were also taken in each night to allow reduction of the science images .
spectrum exposures for each asteroid were splitted in two exposures at two different slit positions , a and b , separated by 20 '' ( the width of the slit was 2.0 " ) .
the observations were performed with the telescope tracking at the proper motion of the asteroid .
hence by subtracting a from b and b from a , a very accurate background removal is achieved .
finally , standard methods for spectra extraction were applied .
the reflectance spectra of ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 are shown in figs .
[ fig1 ] and [ fig2 ] .
both spectra show a steep slope shortwards of 0.70 @xmath1 m and a deep absorption band longwards of 0.75 @xmath1 m . using the algorithm of bus ( 1999 )
, we determine that the spectra can be classified as v - type .
figure [ fig1 ] show that our observations are compatible with the spectra of previously known v - type asteroids ( gray lines ) taken from the smass survey ( bus & binzel , 2002 ) and the s3os2 survey ( lazzaro et al . , 2004 ) .
figure [ fig2 ] show the good agreement between the five band photometry of the sdss - moc ( black lines ) and the observed spectra .
it is worth noting that the values of maximum and minimum reflectance prevents to attribute to these spectra other taxonomic classification , like r- , o- or q - type . in view of this , ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 may be considered , together with ( 1459 ) magnya , the only v - type asteroids discovered up to now in the outer belt .
notwithstanding , the spectra of ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 show a shallow absorption feature around 0.60 - 0.70 @xmath1 m that has never been reported before in v - type asteroids .
this feature is more evident in the spectrum of ( 10537 ) 1991 ry16 .
after the its identification in the november 14 observations , and excluding possible reduction artifacts or solar analogs problems , we requested director discretionary time ( ddt ) for another observational run on december 29 .
only the spectra of ( 7472 ) kumakiri was able to be observed during this run , confirming the presence of the absorption band .
nevertheless , the band in the spectrum of ( 10537 ) 1991 ry16 has also been observed independently by moskovitz et al .
( 2007 ) . to analyze this band ,
we rectified the spectra by subtracting a linear continuum in the interval 0.55 and 0.75 @xmath1 m and then fitted several polynomials of different degrees .
this allowed to determine the center of the band at @xmath9 @xmath1 m and the fwhm of @xmath10 @xmath1 m ( e.g. fig .
[ fig3 ] ) .
the origin of this absorption band is unclear . such kind of bands
are usually believed to arise from the @xmath11 charge transfer absorptions in phyllosilicate ( hydrated ) minerals ( vilas & gaffey , 1989 ; vilas et al . , 1993 ) .
however , it is difficult to explain the presence of a hydrated mineral in the surface of a basaltic object , because the heating and melting that produce the basalt also eliminate any traces of water .
it is known that pyroxene crystals fe@xmath12 cations do not show any absorption bands in the spectral region from 0.56 to 0.72 @xmath1 m .
therefore , the origin of the observed band might be related to other impurity cations like mn@xmath12 , cr@xmath13 , and fe@xmath13 , usually located in the m1 site of terrestrial and meteorite orthopyroxenes ( shestopalov et al . , 2007 ) .
in particular , broad spin - allowed bands of trivalent chromium around 0.430 - 0.455 @xmath1 m and 0.620 - 0.650 @xmath1 m have been observed in both reflectance and transmitted spectra of cr - containing terrestrial ortho and clinopyroxenes ( see cloutis , 2002 ) , as well as in diogenite reflectance spectra ( mcfadden et al . , 1982 ) .
cr@xmath13 cations also give spin - forbidden bands near 0.480 , 0.635 , 0.655 , and 0.670 @xmath1 m but they do not give absorptions near 0.57 @xmath1 m . cloutis ( 2002 ) specifically found that cr@xmath13 gives rise to an absorption band near 0.455 @xmath1 m and a more complex absorption feature in the 0.65 @xmath1 m region .
however , changes in the grain size of the pyroxenes may have an effect on the depth of these absorption bands ( cloutis and gaffey 1991 ; sunshine and pieters 1993 ) .
therefore , the presence of specific absorption bands can be taken as an evidence for the presence of a particular cation , but the characteristics of these bands ( depth and width ) are probably not reliable enough to constrain the cation abundance ( cloutis 2002 ) .
for example , the grain size may be responsible of the different band depth observed between the spectra of ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 .
the slight differences in the band profile between the november and december spectra of ( 7472 ) kumakiri might be attributed to different rotational phases .
another interesting feature observed in our spectra is that the band center of the major absorption feature at 0.90 @xmath1 m is displaced to larger wavelenghts . in our spectra , this region is the noisiest but using different polynomial fits it was possible to estimate the center of the band nearer to 0.92 - 0.93 @xmath1 m .
this behavior may also be attributed to the presence of chromium on the surface .
actually , cloutis and gaffey ( 1991 ) suggested that the cr - rich pyroxene samples in their study have the two major absorption features ( i.e. the one centered at 0.9 and the one centered at 1.9 @xmath1 m , respectively ) displaced to larger wavelengths than expected , relative to their fe contents .
these authors also presented the predicted _ versus _ actual wavelength position of the major fe@xmath12 absorption band center in the 1 @xmath1 m region , and this center is closer to 0.92 @xmath1 m than to 0.90 @xmath1 m .
therefore , the observational evidence points to a possible cr - rich basaltic composition on the surfaces of ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 .
concerning the dynamical behavior of these two asteroids , table [ proper ] lists their proper elements and diameters , as well as those of ( 1459 ) magnya .
the three asteroids are too small to be differentiated bodies by themselves , they are quite spread in proper elements space and do not belong to any of the asteroid dynamical families identified in the outer belt .
therefore , they are likely to be fragments from more than one differentiated parent bodies . nevertheless , at variance with ( 1459 ) magnya , ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 evolve very close to the non linear secular resonance defined by the combination @xmath14 , where @xmath15 and @xmath16 represent the frequencies of the perihelion @xmath17 and node @xmath18 , respectively ( @xmath19 for asteroid , @xmath20 for jupiter , @xmath21 for uranus ; see milani & kneevi , 1992 ) . a 50 my simulation of the orbits of these two asteroids , including gravitational perturbations from the four major planets ,
indicate that they have quite stable orbits showing a slow circulation of the angle @xmath22 .
although this may be just a coincidence , a dynamical connection between ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 can not be ruled out and should be addressed by more detailed studies .
we presented visible spectroscopic observations of two asteroids , ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 , located in the outer belt .
the main goal of our work was to show that these observations are compatible with the v - type taxonomic class .
therefore , these bodies would constitute the second and third basaltic asteroids discovered up to now in that part of the main belt .
however , the presence of a shallow absorption band in the spectra around 0.65 @xmath1 m opens some questions about the actual mineralogy of these two asteroids .
this band is likely to be related to the presence of cr@xmath13 cations , and provides evidence for a possible a cr - rich basaltic surface . the spectroscopic similarities among the two asteroids , together with some shared dynamical properties , point to the idea of a common origin from the breakup of a differentiated parent body in the outer belt
. further studies , including near infrared ( nir ) spectroscopic observations , are mandatory to better address these issues .
we thank calar alto observatory for allocation of director s discretionary time to this programme .
fruitful discussions with d. nesvron are also highly appreciated .
based on observations collected at the centro astronmico hispano alemn ( caha ) at calar alto , operated jointly by the max- planck institut fr astronomie and the instituto de astrofsica de andaluca ( csic ) .
rd acknowledges financial support from the mec ( contract juan de la cierva ) .
alvarez - candal , a. , duffard , r. , lazzaro , d. , & michtchenko , t.a .
2006 , a&a , 459 , 969 asphaug , e. 1997 , meteor . & planet .
, 32 , 965 binzel , r. , rivkin , a. , stuart , s. , et al . 2004 , icarus , 170 , 259 binzel , r.p . , masi , g. , & foglia , s. 2006 , bull .
soc . , 38 , 627 burbine , t.h . ,
buchanan , p.c , binzel , r.p .
, et al . 2001 , meteor .
, 36 , 761 bus , s.j . 1999 , phd thesis , mit bus , s.j . , & binzel , r.p .
2002 , icarus , 158 , 146 carruba , v. , michtchenko , t.a . , roig , f. , et al .
2005 , a&a , 441 , 819 carruba , v. , roig , f. , michtchenko , t.a . ,
2007 , a&a , 465 , 315 cellino , a. , bus , s. j. , doressoundiram , a. & lazzaro , d. in asteroids iii ( eds bottke , w. f. et al . ) 632643 ( univ .
arizona press , tucson , 2003 ) cloutis , e. a. ; gaffey , m. j. journal of geophysical research , vol .
96 , 1991 , p. 22809 - 22826 .
cloutis , e.a .
journal of geophysical research ( planets ) , volume 107 , issue e6 , pp . 6 - 1 .
cruikshank , d.p . ,
tholen , d.j . ,
bell , j.f .
, et al . 1991 , icarus , 89 , 1 delbo , m. , gai , m. , lattanzi , m.g . , et al .
2006 , icarus , 181 , 618 duffard , r. , de len , j. , licandro , j. , et al .
2006 , a&a , 456 , 775 florczak , m. , lazzaro , d. , & duffard , r. 2002 , icarus , 159 , 178 hardorp , j. 1978 , a&a , 63 , 383 ivezi , . ,
tabachnik , s. , rafikov , r. , et al .
2001 , aj , 122 , 2749 juri , m. , ivezi , . ,
lupton , r.h .
, et al . 2002 , aj , 124 , 1776 lazzaro , d. , michtchenko , t.a . ,
carvano , j.m . ,
2000 , science , 288 , 2033 lazzaro , d. , angeli , c.a . , carvano , j.m . , et al . 2004 , icarus , 172 , 179 mccord , t.b . , adams , j.b . , & johnson , t.v . 1970 , science , 168 , 1445 mcfadden , l.a . ,
gaffey , m.j . ,
takeda , h. , jackowski , t.l . ,
reed , k.l . ,
25 , 188 - 206 .
mcfadden , l. , gaffey , m.j .
, & mccord , t. 1985 , science , 229 , 160 michtchenko , t.a . , lazzaro , d. , ferraz - mello , s. , et al .
2002 , icarus , 158 , 343 milani , a. , & kneevi , z. 1992 , icarus , 98 , 211 moskovitz , n.a . ,
willman , m. , lawrence , s.j .
, et al . 2007 , lpi conf . , 38 , 1663 nesvorn , d. , roig , f. , gladman , b. , et al .
, icarus ( 2007 ) , doi:10.1016/j.icarus.2007.08.034 roig , f. , & gil - hutton , r. 2006 , icarus , 183 , 411 roig , f. , nesvorn , d. , gil - hutton , r. , & lazzaro , d. , icarus ( 2007 ) , doi:10.1016/j.icarus.2007.10.004 shestopalov , d. i. ; mcfadden , l. a. ; golubeva , l. f. , icarus , volume 187 , issue 2 , p. 469 - 481 .
sunshine , j. m. ; pieters , c. m. journal of geophysical research , vol .
e5 , p. 9075 - 9087 .
thomas , p.c . ,
binzel , r.p .
, gaffey , m.j .
1997 , science , 277 , 1492 vilas , f. , & gaffey , m.j .
1989 , science , 246 , 790 vilas , f. , larson , s.m . ,
hatch , e.c .
1993 , icarus , 105 , 67 lccccccc asteroid & ut & @xmath23 [ au ] & @xmath24 [ au ] & @xmath25 [ deg ] & @xmath26 [ mag ] & airmass & @xmath27 + november 14 & & & & & & & + ( 7472 ) kumakiri & 03:25:00 & 2.920 & 2.123 & 13.5 & 16.6 & 1.037 & 3400 sec + ( 10537 ) 1991 ry16 & 00:05:08 & 3.040 & 2.053 & 1.8 & 17.1 & 1.091 & 4000 sec + december 29 & & & & & & & + ( 7472 ) kumakiri & 21:48:52 & 2.873 & 1.901 & 3.8 & 15.9 & 1.094 & 3600 sec + ccccc asteroid & @xmath28 [ au ] & @xmath29 & @xmath30 & @xmath31 [ km ] + + ( 1459 ) magnya & 3.14986 & 0.2183 & 0.2651 & @xmath32 + ( 7472 ) kumakiri & 3.01033 & 0.1372 & 0.1562 & 8.5 + ( 10537 ) 1991 ry16 & 2.84958 & 0.1023 & 0.1101 & 7.3 + m and shifted by 0.5 units in reflectance for clarity . to remove the solar contribution ,
we have used the solar analog hd 191854 in the november 14 observations and the solar analog hd 28099 in the december 29 observation.,width=491 ] m interval .
the spectrum has been rectified by subtracting a linear continuum in this interval . from a polynomial fit ( thick line ) , the center of the absorption band
is detected at 0.63 @xmath1 m with a fwhm of 0.1 @xmath1 m . ] | the identification of basaltic asteroids in the asteroid main belt and the description of their surface mineralogy is necessary to understand the diversity in the collection of basaltic meteorites .
basaltic asteroids can be identified from their visible reflectance spectra and are classified as v - type in the usual taxonomies . in this work ,
we report visible spectroscopic observations of two candidate v - type asteroids , ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 , located in the outer main belt ( @xmath0 ua ) .
these candidate have been previously identified by roig and gil - hutton ( 2006 , icarus 183 , 411 ) using the sloan digital sky survey colors .
the spectroscopic observations have been obtained at the calar alto observatory , spain , during observational runs in november and december 2006 .
the spectra of these two asteroids show the steep slope shortwards of 0.70 @xmath1 m and the deep absorption feature longwards of 0.75 @xmath1 m that are characteristic of v - type asteroids .
however , the presence of a shallow but conspicuous absorption band around 0.65 @xmath1 m opens some questions about the actual mineralogy of these two asteroids .
such band has never been observed before in basaltic asteroids with the intensity we detected it .
we discuss the possibility for this shallow absorption feature to be caused by the presence of chromium on the asteroid surface .
our results indicate that , together with ( 1459 ) magnya , asteroids ( 7472 ) kumakiri and ( 10537 ) 1991 ry16 may be the only traces of basaltic material found up to now in the outer main belt . |
protoplanetary disk around pre - main sequence stars are potential sites for planet formation . in the standard model ( e.g. * ? ? ?
* ) , after settling into equilibrium , the gas and solid components in the disk are separated and dust grains begin sinking to the midplane to form a thin dust layer which eventually fragments into planetesimals , ingredients of planets .
the radial profile of the surface density structure at this stage is likely one of the most important parameters to control what kinds of planets form in the disk @xcite .
since the gas and dust are well mixed in the beginning , the grain growth and sedimentation are key processes to separate dust from gas .
theoretical studies have suggested that dust grains are removed from the flaring surface and will sink to the midplane with growing by coalescence along the disk evolution @xcite .
one way to find such evolved disks is to carefully examine the spectral energy distribution ( sed ) of young stars as well as detection of ( sub)millimeter continuum emission , because the shape of the sed of young stars is sensitive to its temperature structure mainly determined by the vertical distribution of dust grains @xcite .
@xcite studied the dust growth and settling in passive disks to investigate the evolution of disk structure and seds .
their numerical results indicated that the flux densities at mid - infrared ( mir ) from a disk get lower as a result of dust grain growth and settling from the surface layer . in the last two decades , extensive observations from optical to millimeter wavelengths
have been conducted to study disk properties ultimately determining the nature of forming planets .
pre - main sequence stars , particularly in the quiescent phase such as weak line t tauri stars ( wttss ) or stars are good candidates for searching such evolved dusty disks in the process of sedimentation .
a pioneer survey in ( sub)millimeter continuum emission of the circumstellar dust around young stars was conducted by @xcite and followed by many other groups @xcite .
@xcite made an imaging survey of single t tauri stars in taurus and derived the physical properties of the disks combined with the seds .
they found the radial expansion of the disks attributed to outward transport of angular momentum with decreasing h@xmath6 line luminosity , a measure of disk evolution .
@xcite made a large survey of disk emission of more than 100 young stars in taurus and found that there is a distinct difference in detection rates between classical t tauri star ( ctts ; 90% ) and wtts ( 15@xmath75% ) .
no massive evolved disk around wtts , however , was found . in the present paper ,
we report the discovery of an unusual cold massive disk around the t tauri star , , at the southern tip of lupus 3 at a distance of 150 pc .
the source has a quiescent accretion disk presumably in the transient stage from ctts to wtts because its equivalent width of h@xmath6 ranged 714 @xmath8 @xcite .
is known as a single star whose spectral type , mass , and age are k6 , 0.78 @xmath5 , and 3 myr , respectively @xcite .
our observations have been performed with the atacama submillimeter telescope experiment ( aste ) , the submillimeter telescope operated by nobeyama radio observatory ( nro ) , a branch of national astronomical observatory japan ( naoj ) , in collaboration with university of chile , and japanese institutes including university of tokyo , nagoya university , osaka prefecture university , ibaraki university , and hokkaido university .
the telescope is located at pampa la bola , chile , an altitude of 4860 m , and is remotely operated from the aste operation facilities in chile and japan through the network observation system n - cosmos3 @xcite .
we employed two receivers , aztec 1.1 mm camera @xcite and receiver @xcite , for millimeter continuum observations and molecular line observations at submillimeter wavelengths , respectively .
aztec is a 144-element bolometer array camera installed on the aste between april 2007 and december 2008 @xcite . in this period ,
the array was configured to operate at a wavelength of 1.1 mm .
the angular resolution and field of view of aztec on aste were 28and 78 , respectively .
we have made wide - field ( 40@xmath940 ) millimeter continuum observations of the molecular cloud including the target , , in july 2007 .
the individual on - the - fly maps were obtained in the raster - scan mode with a scanning velocity of 100@xmath10 s@xmath11 .
the telescope was scanned along the az - el coordinates with a spacing of 195 , a quarter of the field of view .
all the data were taken at night or early morning and the average 270 ghz radiometric opacity at zenith was @xmath12 during the observation period .
telescope pointing was checked by mapping the quasar every one and half , or two hours .
the pointing corrections are applied during the data analysis .
the flux conversion factor from optimal loading to source flux density of the aztec array was calibrated by observing uranus , neptune , or 3c279 every night .
the aztec data set was reduced by using the aztec data reduction pipeline @xcite for the identification of point sources .
following @xcite , we employed the principle component analysis ( pca ) cleaning method to remove the atmospheric emission .
the pixel size of the final map was set to be 3@xmath10 , 10 % of aztec beam size .
the noise level and effective resolution of the final map are 5.0 mjy beam@xmath11 and 35@xmath10 , respectively .
@xmath1co and @xmath2co j=32 line observations were performed in march 2008 with a 2 sideband - separating ( 2sb ) receiver , cats345 , on aste .
the half power beam widths ( hpbws ) at 345.79 and 330.59 ghz were @xmath13 and @xmath14 , respectively , which corresponds to 3300 and 3450 au at the distance to the target .
the main beam efficiency was about 60 % around these frequencies .
the system noise temperatures were typically 400 and 350 k for @xmath1co and @xmath2co , respectively .
for the backend , we used a 1024 channel digital auto - correlator , which has a band width of 128 mhz and a resolution of 125 khz , corresponding to 111 and 0.11 km s@xmath11 , respectively , at 345 ghz .
the telescope pointing calibration was performed every 1.5 - 2 hours by observing , and the resulting pointing accuracy was typically @xmath15 . for the absolute flux calibration , the integrated intensities of and
were measured two or three times per day and compared with those by @xcite .
data reduction and analysis were made with the newstar software package developed in nro , which is a frontend of the astronomical image processing system ( aips ) developed in the national radio astronomy observatory ( nrao ) .
we employed the five - point observation method in which spectra are obtained not only at a stellar position but also at four adjacent points @xmath13 apart from the stellar position in the r.a . or dec .
this method enables us to deduce the excess emission toward the central star by subtracting the average spectrum of the surrounding four spectra from the central spectrum , even if the star is located in extended components , such as an ambient cloud .
the 1.1 mm continuum map around is shown in figure [ fig : cont]a .
strong compact continuum emission was detected towards the stellar position . from the 2-d gaussian
fitting , the peak position of the emission is obtained to be 16@xmath1608@xmath17361 , @xmath18392316 in j2000 , which agrees well with the 2mass position of within 1@xmath195 . the total flux density at 1.1 mm was measured to be [email protected] mjy from the deconvolved cleaned map ( see below ) , which is one of the strongest 1.1 mm emission among t tauri stars in the lupus molecular cloud .
the negative contours around the emission are due to a point spread function ( i.e. , a beam ; see figure [ fig : cont]c ) with a negative hole generated from the pca cleaning method .
figure [ fig : cont]b shows a deconvolved map by subtracting the measured point spread function ( psf ) from the emission via the _ clean _ algorithm .
the psf of the aztec map for is shown in figure [ fig : cont]c . from the deconvolved map
, we found that the emission is composed largely of a spatially unresolved component .
the derived clean components are located around the star over two pixels ( @xmath20 in size ) with its separation comparable to the pointing uncertainty estimated from the pointing measurements ( figure [ fig : cont]b ) .
therefore , we consider that the strong millimeter emission comes from a compact ( @xmath21 au ) circumstellar dusty disk associated with .
this is supported by the fact that , unlike the case of a stellar type flare seen in wtts @xcite , did not show any time variation of the flux density at millimeter wavelengths during the observing period .
figure [ fig : sed ] shows the sed of .
the flux density at 1.1 mm is unusually strong as the 1.1 mm to infrared flux density ratio is near unity and larger than those for t tauri stars in taurus @xcite .
adapting the extinction relation of @xcite , we measured the spectral index for the dereddened fluxes in the range from 2 to 10 @xmath3 , @xmath22 , to be @xmath23 , where @xmath24 $ ] .
this value suggests that should be categorized as a object .
the shape of the measured sed is unique in the sense that there are not distinct broadband spectral features , such as a large dip , from near - infrared ( nir ) to millimeter wavelengths with a power - law like monotonic decrease .
the index of the decrease is measured to be @xmath25 in the range from 2 to 1100 @xmath3 .
however , there is a small turnover around 10 @xmath3 , i.e. , the index at the longer wavelength , from 20 to 1100 @xmath3 , follows nearly the entire trend to be @xmath26 while that from 2 to 10 @xmath3 is @xmath23 .
furthermore , there is no dip feature in the nir to mir wavelength range caused by clearing material in the inner parts of the disk , such as @xcite .
the five - point spectra of the @xmath1co and @xmath2co j=32 emissions are shown in figure [ fig:5point ] .
the significant @xmath1co and @xmath2co emissions were detected over the five - points around the star . there is a clear intensity gradient along the east west direction in the @xmath2co spectra as well as in @xmath1co . in the @xmath1co spectra
, we could not find any symmetrical wing feature of an outflow with a velocity width of @xmath2710 km s@xmath11 which is often seen around a low - mass young stellar object .
the @xmath1co and @xmath2co spectra are peaked around the lsr velocity of 4.9 km s@xmath11 , which is slightly shifted from the systemic velocity of the main cloud ( 4.1 km s@xmath11 ; * ? ? ?
the peak intensity in @xmath28 scale and the velocity width in fwhm are 7.0 k and 0.66 km s@xmath11 , respectively , on average over the five - points for @xmath1co , and 0.7 k and 0.85 km s@xmath11 for @xmath2co . from the intensity ratio of the @xmath1co to @xmath2co spectra
, we estimate the optical depth of both lines toward the star .
if we assume the same excitation temperature , the mean optical depths of @xmath1co and @xmath2co are obtained to be 16.7 and 0.19 , respectively .
since the @xmath2co emission is optically thin in contrast to @xmath1co , the associated gas with the star can be expected to be seen in @xmath2co through the parent cloud as excess emission towards the star . by assuming the local thermodynamic equilibrium with the excitation temperature , @xmath29
, the h@xmath30 mass of the gas in the beam solid angle toward the source , @xmath31 , can be estimated from the intensity of the optically thin @xmath2co j=32 line as follows : @xmath32 } \biggr)^2\\ \nonumber & \times & \frac{\exp(15.9/t_\mathrm{ex})}{1-\exp(-15.9/t_\mathrm{ex } ) } \frac{t_\mathrm{ex}}{j(t_\mathrm{ex})-0.05 } \\ & \times & \frac{\tau_\mathrm{^{13}co}}{1-\exp(-\tau_\mathrm{^{13}co } ) } \int \frac{t_\mathrm{a}^\ast(\mathrm{^{13}co})}{\eta_\mathrm{mb } } dv \ \ \ \ \ [ m_\sun]\mathrm { , } \label{eq : mass}\end{aligned}\ ] ] where @xmath33 $ ] , @xmath34 , @xmath35 is the fractional abundance of @xmath2co with respect to h@xmath30 , @xmath36 is the distance to the source , @xmath37 is the optical depth , and @xmath38 is the main beam efficiency of aste @xcite .
we adopt an excitation temperature 10 k which is the typical value in the ambient molecular cloud @xcite , and @xmath39 for @xmath35 by assuming that the abundance ratio of @xmath40 and the @xmath41 conversion factor are 60 and @xmath42 , respectively .
our mass as well as column density are estimated to be @xmath43 @xmath5 and @xmath44 @xmath45 by using equation ( [ eq : mass ] ) .
the column density is much smaller than the averaged one in the lupus 3 cloud @xcite , suggesting that there is no large and massive envelope around . to search for the compact excess emission towards the star , we made residual spectra of the @xmath1co and @xmath2co emission as shown in figure [ fig : residual ] .
each residual spectrum was obtained by subtracting the average of the surrounding four spectra from the central one .
the excess emission above the 2@xmath46 noise level over two velocity channels can be discerned in the @xmath2co residual spectrum around the velocity of 6 km s@xmath11 .
the peak intensity and the velocity width in fwhm are measured to be 0.12 k and 1.43 km s@xmath11 , respectively .
since the velocity of the excess emission agrees well with the radial velocity of the star , [email protected] km s@xmath11 @xcite , this may be the possible detection of the gas disk around .
the @xmath1co residual emission was detected at @xmath47 4.1 and 6.4 km s@xmath11 whose peak intensities are 0.30 and 0.43 k in @xmath28 , respectively . although the @xmath1co emission is shown to be optically thick , the double - peaked profile of the residual spectrum towards the star might suggest the presence of a rotating gas disk around .
we tried to fit the rotating disk model @xcite to the both spectra to examine quantitatively .
we could not , however , find any parameter set to reproduce the both residual spectra consistently .
for example the @xmath1co intensity can be reproduced for the larger rotating disk model .
such a disk model expects the stronger @xmath2co intensity than the obtained and can not explain the both intensities compatibly .
in addition , both velocity widths are not well fitted by the model simultaneously .
consequently , the residual spectrum in @xmath1co is not likely to originate from the circumstellar gas disk .
the measured sed of has the unique feature : there is no significant turnover and a monotonic decrease from nir to millimeter .
this feature is well described by a flattened disk for the dust component . according to @xcite , the flattened disk has a lower temperature than that of a flared disk with the well - mixed dust because the central star can not illuminate the flattened disk efficiently . in this case , the sed is predicted to have an asymptotic power - law slope with an index of @xmath48 from 30 @xmath3 to 1 mm .
the mir to millimeter continuum radiation mainly comes from the optically thick interior of the disk . on the other hand ,
the sed of a flared disk rises in the mir to far - infrared ( fir ) regime owing to the strong emission from a hot surface layer and gradually falls down toward the millimeter regime , mainly tracing the cooler interior . to deduce disk parameters
, we performed the least square fitting to the sed of using a simple power - law disk model ( same as model 1 case in * ? ? ? * ) .
our disk model is assumed to have a power - law form of the surface density and temperature variations over the entire area of the disk , @xmath49 and @xmath50 , respectively : @xmath51 } } \biggr)^{-p}\ ] ] and @xmath52}}\biggr)^{-q } \mathrm{.}\ ] ] here , @xmath53 is the radial distance in units of au , @xmath54 is the surface density at 100 au , and @xmath55 is the temperature at 1 au .
we applied the model to the data without the extinction correction because it was corrected in the model by calculating the mass opacity at each frequency from @xmath56 with the extinction curve ( figure 1 of * ? ? ?
the lower limit of the temperature is fixed at 10 k which is the typical temperature of the lupus 3 cloud @xcite .
we adopt the effective temperature of the star of 4255 k , the stellar mass of 0.78 @xmath5 and the visual extinction toward the star of 1.75 mag @xcite .
the outer radius of the disk , @xmath57 , and the disk inclination angle , @xmath58 , are fixed to be 320 au and 45 degree , respectively .
the outer radius is derived from the pixel size of 3@xmath10 as an upper limit because the emission was not spatially resolved .
stellar radius , @xmath59 , inner radius of the disk , @xmath60 , @xmath55 , @xmath61 and @xmath54 are the free parameters of the disk model .
the model seds are computed and fitted to the observational data for the following four cases : @xmath62 and @xmath63 ( minimum mass solar nebula by * ? ? ?
* ) and @xmath64 and 1 , where @xmath65 is the power - law index of the dust mass opacity coefficient , @xmath66
@xmath67 g@xmath11 @xcite . in the fitting
, we treated all the data points with the equal weights of the largest uncertainty because the sed fitting tends to be biased toward the data points with small uncertainties at shorter wavelengths .
these model seds seem to agree with the observed sed as shown in figure [ fig : sed ] and its reduced chi - squares are measured to be @xmath68 .
the derived parameters by the fitting are listed in table [ tab : fitting ] and [ tab : fitting_result ] . in figure [
fig : sed ] , we plotted the fitting results only for @xmath69 case because the two best - fit curves with @xmath62 and 1.5 are almost identical for each @xmath65 value .
two remarkable disk features are derived in the fitting .
first , the disk is very cold with a temperature of 99 k at 1 au .
the fitted stellar luminosity of 0.85 @xmath70 suggests that @xmath55 in the condition of radiative equilibrium is expected to be 269 k when the disk is in hydrostatic equilibrium and gas and dust are well mixed .
previous studies of the sed fitting have found that most t tauri stars in taurus have a higher temperature of 100400 k at 1 au @xcite .
our derived low temperature suggests a flattened dust disk as predicted by theoretical studies ( e.g. * ? ? ?
the second remarkable finding is that the disk is massive .
our estimated disk mass ranges from 0.03 to 0.12 @xmath5 and the ratio of the disk to stellar mass is 313% .
these values are much higher than the typical values for t tauri stars in the taurus - auriga region : median disk masses are @xmath71 and @xmath72 @xmath5 for ctts and wtts , respectively , and the median disk - to - star mass ratio is only 0.5 % @xcite .
it is very rare to find a massive disk in a quiescent accretion phase like a wtts one .
for example , @xcite have found that only 15 % of wttss are detectable at the submillimeter wavelengths .
the above arguments are not affected much by either the disk inclination angle or the disk outer radius .
if we change the disk inclination angle from 20 to 70 degrees with @xmath73 au and @xmath69 , the temperature at 1 au varies from 90 to 128 k , which is still low , and the index @xmath61 remains around 0.68 .
the variation of the disk mass is less than 20 % , remaining the massive disk .
furthermore , when we adopt a smaller @xmath57 than 320 au , a higher surface density is required to reproduce the observed sed , resulting the massive disk . for example , in the @xmath57=110 au case , @xmath54 becomes more than 5 times larger than that for the @xmath57=320 au case and the disk mass increases up to 0.42 @xmath5 , which remains massive . for a smaller opaque disk
, we can not accurately estimate the disk mass , though our interpretation of the massive disk is still valid .
therefore , we definitely conclude that the disk around is colder than the usual flared disk and that the disk mass is much larger than the typical value of t tauri stars in taurus . in our sed fitting , we adopted a low threshold of 10 k in the temperature profile because the source is likely embedded in its parent cloud as shown by our @xmath1co observations .
this threshold causes an increment of flux densities around at 300 @xmath3 in the model sed because the temperature of 10 k corresponds to 290 @xmath3 by wien s law ( see figure [ fig : sed ] ) . to investigate the effect of the temperature threshold
, we also analyzed the sed by using a disk model with a threshold of 2.7 k , the temperature for the cosmic background radiation . in this case , the observed sed can be well fitted only when the disk inclination angle is less than 30 degree .
the fitted sed for @xmath69 and @xmath74 case is shown by a gray line in figure [ fig : sed ] ; the best - fit curve can also explain the observed sed , and the disk mass becomes four times larger than that in the 10 k threshold case .
note that such a massive disk relative to the stellar mass is likely to be gravitationally unstable . on the other hand ,
@xmath55 and @xmath61 decrease only by 2 % and 11 % , respectively .
consequently , the threshold in the temperature profile of the disk model does not change our interpretation of a cold massive disk around .
the cold and massive disk we found around strongly suggests the presence of a flattened disk where a large number of small dust grains are settling . since seems to be in the quiescent phase of disk accretion
, the majority of dust particles are expected to just settle into the midplane with dust growth through coalescence .
theoretically , the thin dust layer in the midplane fragments into numerous planetesimals by its gravitational instability , and as a result , a hole appears at the inner part of the disk to produce a dip feature at @xmath75 @xmath3 in sed .
however , we could not find such a feature at the ir wavelengths in the sed of , indicating that there is no inner hole created by the formation of planetesimals .
indeed , the strong continuum emission toward that we detected suggests that a large number of small particles rather than larger planetesimals with sizes of @xmath75 km is the dominant component in the circumstellar disk .
consequently , the disk around is most likely in the quiescent stage after the dust sedimentation in the midplane , but just prior to the planetesimal formation .
is a rare object in terms of the shape of its sed . to find a similar object with a massive cold disk to
, we checked the available photometric data at the mir and millimeter wavelengths @xcite and measured the flux ratio of mir flux to millimeter flux .
we found that only , a class ii ctts in taurus , has a high intensity ratio of the 1 mm to 70 @xmath3 as well as to 25 @xmath3 ( @xmath76 ) , suggesting the strong millimeter flux despite its weak mir flux and the sed with a monotonic slope from mir to millimeter .
the temperature at 1 au and the disk mass of are estimated to be 87 k and 0.13 @xmath5 , respectively @xcite .
we can roughly estimate the timescale for the duration of a cold and massive disk like those around and as follows .
given about 30 samples in taurus whose flux densities at 1 mm and mir were previously determined , the timescale of the cold massive disk is estimated to be at least one order of magnitude shorter than the typical age of a t tauri star ( @xmath77 yr ) , if we assume that new stars randomly form with a constant rate .
such a short timescale is consistent with the gravitational instability of the thin dust layer after the dust sedimentation at the midplane .
it is suggested that the dust layer produced by settling to the equatorial plane becomes gravitationally unstable and rapidly fragments into numerous planetesimal ( @xmath7810@xmath79 yrs ; * ? ? ?
it is very interesting that the physical properties of such a quiescent disk provide the initial conditions of the successive formation of planets ( e.g. * ? ? ?
in particular , it is crucial to reveal the radial variation of the disk surface density with high - resolution and high - sensitive observations in future .
for the dust coagulation , it is theoretically expected that the index of the dust opacity coefficient , @xmath65 , is lower than unity as a result of dust growth @xcite .
we treated @xmath65 as fixed parameter in the sed fitting because the photometric data at the millimeter and sub - millimeter wavelengths are limited .
more ( sub-)millimeter data are required to determine the @xmath65 value accurately .
additionally , high - resolution molecular line observations will reveal the disk kinematics and the initial conditions of the planet forming disk more quantitatively .
we have detected the strong 1.1 mm continuum emission toward the t tauri star , , in the lupus molecular cloud with aztec on aste .
in addition , we have obtained the @xmath1co and @xmath2co j=32 spectra at the five positions around the source .
the main results of our observations are summarized in the following . *
significant continuum emission was detected just toward the stellar position .
the emission is not spatially resolved with the effective aste beam size of 35@xmath10 , indicating that a dust disk is associated with the star .
the total flux density is 272 mjy at 1.1 mm which is the strongest among t tauri stars in the lupus molecular cloud . *
the @xmath1co and @xmath2co emissions were detected over the five - points around the star . the five - points spectra show no signature of a large - scale molecular outflow or a massive envelope .
we made residual spectra by subtracting the average spectrum of the surrounding four spectra from the central spectra .
the @xmath1co excess emission was found and the @xmath2co one was marginally discerned towards the star .
the @xmath2co excess may come from the gas disk while the @xmath1co excess emission detected can not be explained by the gas disk . * the observed sed shows two prominent features .
first , there is no significant turnover with a constant slope from nir to millimeter which is well reproduced by a flattened disk .
second , there is no signature of a dip at around 10 @xmath3 suggesting no inner hole in the disk .
* we fit a power - law disk model to the sed when @xmath80 is fixed to be 1.0 and 1.5 , and @xmath65 is 0.0 and 1.0 .
the best fit parameters of the disk are @xmath81(1.254.97 ) g @xmath45 , @xmath8299 k , @xmath83 , and @xmath84(0.030.12 ) @xmath5 .
the disk temperature at 1 au is lower than that of a flared disk in hydrostatic equilibrium .
the disk mass is much higher than the typical values for t tauri stars . *
the cold and massive disk around can be well interpreted as a partially settled disk of a large number of small dust grains . together with no signature of an inner hole in the sed
, it is most likely that the disk is in the quiescent phase after the dust sedimentation in the midplane , but just prior to the planetesimal formation .
the may be a rare sample to investigate the initial conditions for the formation of a planetary system .
we appreciate the referee for the constructive comments that have helped to improve this manuscript .
we acknowledge the aste and aztec staff for the operation and maintenance of the observing instruments .
we are grateful to , , and for testing the data reduction software .
we thank , , and for the pointing model correction of aztec on aste .
observations with aste were in part carried out remotely from japan by using ntt s gemnet2 and its partner r&e ( research and education ) networks , which are based on accessnova collaboration of university of chile , ntt laboratories , and naoj .
this work is supported in part by a grant - in - aid for scientific research ( a ) from the ministry of education , culture , sports , science and technology of japan ( no .
19204020 ) and by jsps ( no .
18204017 ) .
lcccc @xmath65 & & + @xmath80 & 1.0 & 1.5 & 1.0 & 1.5 + @xmath59 [ au ] & + @xmath89 [ @xmath70 ] & + @xmath60 [ au ] & + @xmath54 [ g @xmath45 ] & 1.25 & 1.47 & 4.31 & 4.97 + @xmath55 [ k ] & 99.0 & 99.0 & 99.3 & 99.4 + @xmath61 & + @xmath92 [ @xmath5 ] & 0.03 & 0.04 & 0.10 & 0.12 + [ tab : fitting_result ] | we present the discovery of a cold massive dust disk around the t tauri star in the lupus molecular cloud from the 1.1 millimeter continuum observations with aztec on aste . a compact ( @xmath0320 au ) continuum emission coincides with the stellar position having a flux density of 272 mjy which is largest among t tauri stars in lupus .
we also present the detection of molecular gas associated with the star in the five - point observations in @xmath1co j=32 and @xmath2co j=32 . since our @xmath1co and @xmath2co observations did not show any signature of a large - scale outflow or a massive envelope , the compact dust emission is likely to come from a disk around the star .
the observed sed of shows no distinct turnover from near infrared to millimeter wavelengths , which can be well described by a flattened disk for the dust component , and no clear dip feature around 10 @xmath3 suggestive of absence of an inner hole in the disk .
we fit a simple power - law disk model to the observed sed .
the estimated disk mass ranges from 0.03 to @xmath40.12 @xmath5 , which is one or two orders of magnitude larger than the median disk mass of t tauri stars in taurus .
the resultant temperature is lower than that of a flared disk with the well - mixed dust in hydrostatic equilibrium , and is probably attributed to the flattened disk geometry for the dust which the central star can not illuminate efficiently . from these results together with the fact that there is no signature of an inner hole in the sed , we suggest that the dust grains in the disk around sank into the midplane with grain growth by coalescence and is in the evolutional stage just prior to or at the formation of planetesimals . |
a number of interesting features of the iron - based superconductors have attracted much attention . among them
is the material dependence of the superconductivity .
not only the transition temperature @xmath3 , but also the form of the superconducting gap seems to be material dependent .
for instance , although a number of experiments suggest that the gap is fully open on the fermi surface on some of the arsenides , there are several experiments suggesting the presence of nodes in the superconducting gap lafepo@xcite , bafe@xmath0(as@xmath4p@xmath5)@xmath0 @xcite , or in lifep@xcite . also in the electron - doped 122 systems , ba(fe@xmath4m@xmath5)@xmath0as@xmath0(m = co , ni )
, the presence of gap nodes has been suggested.@xcite for lafepo , theoretical studies suggest the presence of nodes on the electron fermi surface around the wave vector @xmath6 in the unfolded brillouin zone .
@xcite from this viewpoint , the heavily hole doped end material kfe@xmath0as@xmath0 is of special interest because in this material , the electron pockets present in most of the iron - based superconductors are known to be absent due to the low position of the fermi energy.@xcite nonetheless , the material becomes superconducting , and interestingly , several experiments suggest the presence of nodes in the gap .
@xcite since there are no appreciable electron pockets in kfe@xmath0as@xmath0 , the nodes are likely to be on the hole fermi surface .
theoretically , a possibility of @xmath2-wave pairing has been proposed recently.@xcite another interesting observation in kfe@xmath0as@xmath0 is the spin fluctuations found in neutron - scattering experiments . in ref . , incommensurate spin fluctuation has been observed , and we have theoretically provided its explanation in terms of the multiorbital nature of the fermi surface . given this background , the aim of the present study is to understand the relation between the spin fluctuations and the superconducting gap of kfe@xmath0as@xmath0 .
we first construct a ten - orbital model for kfe@xmath0as@xmath0 and also obtain a five - orbital model by approximately unfolding the brillouin zone .
we adopt three - dimensional models in order to take into account the possible variation of the superconducting gap along the @xmath7 direction .
we apply random phase approximation to these models and compare the calculation results at high temperature , which gives reasonable agreement , although the brillouin zone unfolding is not exact . then , we go down to low temperature using the five - orbital model .
we first summarize our understanding on the origin of the incommensurate spin fluctuations@xcite and also analyze its strength when the material composition is varied from bafe@xmath0as@xmath0 to kfe@xmath0as@xmath0 .
we finally discuss the superconducting gap when cooper pairing is mediated by spin fluctuations .
an analysis based on the linearized eliashberg equation shows that @xmath1-wave and @xmath2-wave pairings are found to be in close competition , and the former is found to have essentially the @xmath8-wave form in the sense that the gap function around the wave vector @xmath9 ( where the hole fermi surfaces exist ) and around @xmath6 ( where the electron fermi surface would have existed in a less doped material ) has the opposite sign .
this sign change is driven by the incommensurate spin fluctuation , which is due to the interband interaction between states around ( 0,0 ) and @xmath6 . since the electron fermi pockets are absent , the latter states are somewhat away from the fermi level , so that this interband sign change of the gap function should not be detected experimentally . as for the possible explanation of the nodes of the gap observed experimentally ,
we raise several possibilities concerning the hole fermi surface .
as@xmath0 presented in the unfolded brillouin zone of the st lattice structure .
the dots in the left are the original first - principles band calculation , while the solid lines are the tight - binding model band dispersion obtained by using maximally localized wannier orbitals .
the horizontal cuts of the fermi surface at @xmath10 and @xmath11 are presented in the bottom panels of ( a ) and ( b ) .
translating the left ( right ) panel of the five - orbital fermi surface ( b ) by @xmath12 and superposing it to the right ( left ) corresponds to the ten - orbital fermi surface shown in the right ( left ) .
( c ) the bct and the st lattice structure .
the former ( latter ) unit cell with two ( one ) iron(s ) is adopted for the ten ( five)-orbital model .
the figure has been produced using vesta@xcite .
( d ) the brillouin zone correspondence between the ten - orbital ( black line ) and five - orbital ( blue line ) models .
the z point of the second brillouin zone of the bct structure coincides with the m point of the unfolded brillouin zone.,width=272 ] as@xmath0 refolded into the original brillouin zone ( solid red ) , and the band structure of the ten - orbital model ( dashed black ) .
( bottom panels ) horizontal cuts of the fermi surface for the ten - orbital ( upper ) and five - orbital ( lower ) models .
the thickness represents the @xmath13 , @xmath14 and @xmath15 orbital characters .
, width=302 ] first we perform a first - principles band calculation for kfe@xmath0as@xmath0 adopting the experimentally determined lattice structure@xcite and using the quantum espresso package.@xcite then we exploit maximally localized wannier orbitals@xcite to obtain the ten - orbital model , whose band structure and the fermi surface are shown in fig .
[ fig1 ] ( a ) .
there are ten @xmath16 bands because there are two iron atoms per unit cell in the body - centered tetragonal ( bct ) lattice structure . around the x point ,
the electron fermi surfaces , which are present in iron pnictides with fewer holes , are absent .
nonetheless , the band that gives rise to the electron fermi surface for smaller number of holes sits very close to the fermi level .
three cylindrical hole fermi surfaces are present around the @xmath17 point , which we name @xmath18 , @xmath19 , and @xmath20 from the inner one to the outer .
also around the z point there is a tiny hole fermi pocket . for 122 systems ,
the brillouin zone unfolding procedure@xcite that adopts the reduced unit cell ( with one iron ) can not be done strictly,@xcite namely , the hopping integrals that go out from and come into fe1 and fe2 in fig .
[ fig1](c ) are somewhat different .
nonetheless , if we adopt a reduced unit cell by neglecting the difference , we end up with an approximate five - orbital model.@xcite we show the band structure and the fermi surface of this model in fig .
[ fig1 ] ( b ) in the `` unfolded '' brillouin zone of the simple tetragonal ( st ) lattice structure .
the correspondence between the unfolded st and the folded bct brillouin zone is shown in fig . [ fig1](d ) .
as seen from this correspondence , the band structure of the five - orbital model can be refolded into the original bct brillouin zone by translating the bands in the second bct brillouin zone by a wave vector ( @xmath21,@xmath21,@xmath21 ) .
the comparison of the band structure between the original ten - orbital model and the five - orbital model in the refolded brillouin zone is shown in fig .
[ theory ] .
they coincide with each other in most of the portions , but there is some discrepancy along @xmath17-z . in the middle panels of fig .
[ theory ] , we show the vertical cuts of the fermi surface , where the thickness represents the strength of the @xmath13 , @xmath14 , or @xmath22 orbital character .
all of the hole fermi surfaces are very cylindrical , so that the two - dimensionality is strong . however , if we look at the orbital character , @xmath20 and @xmath19 fermi surfaces have strong @xmath14 and @xmath15 character , respectively , regardless of @xmath7 , while the @xmath18 fermi surface is mainly composed of @xmath23 orbital around the z point , while the @xmath14 character becomes stronger around the @xmath17 point . in this sense , @xmath18 can be considered as less two - dimensional compared to the other two - hole fermi surfaces . in the bottom panels of fig .
[ theory ] , we show similar vertical cuts of the fermi surface for the five - orbital model , refolded into the original brillouin zone .
the result in most portions is close to that obtained for the ten - orbital model , but the @xmath18 hole fermi surface , which is a cylinder in the latter , splits into a three - dimensional pocket around the z point and a strongly warped cylinder .
this split occurs between portions of the fermi surface having strong @xmath13 and @xmath14 character .
although this may seem like a large difference , the effect of this splitting is not so large as far as the quantities we focus on are concerned , as we shall see later .
for the many - body part of the hamiltonian , we consider the intraorbital @xmath24 , the interorbital @xmath25 , the hund s coupling @xmath26 , and the pair hopping interaction @xmath27 .
we consider orbital - dependent interactions,@xcite and the interaction part of the hamiltonian is given as @xmath28 where @xmath29 denotes the sites and @xmath30 , @xmath31 stand for the orbitals .
we apply random - phase approximation ( rpa ) to this model and obtain the spin and charge susceptibility matrices .
namely , using the bare green s function @xmath32 @xmath33 in the orbital representation , the irreducible susceptibility matrix is given as @xmath34 the spin and the charge ( orbital ) susceptibility matrices are obtained as @xmath35 @xmath36 where @xmath37 and @xmath38 are interaction vertex matrices .
we refer to the largest eigenvalue of the matrices @xmath39 and @xmath40 for @xmath41 as the stoner factor and the spin susceptibility , respectively .
the magnetic instability is signaled by the stoner factor exceeding unity . in the rpa calculation ( where the self - energy correction is neglected ) , adopting the interaction values evaluated by first - principles calculation@xcite easily results in a magnetic instability even at high temperature , so we multiply all the electron - electron interaction by a constant factor @xmath42@xcite . to analyze superconductivity , the green s function and the effective singlet pairing interaction , @xmath43
are plugged into the linearized eliashberg equation , @xmath44 the eigenvalue @xmath45 increases upon lowering the temperature and reaches unity at @xmath3 . instead of going down to @xmath3 , which will be a tedious calculation
, we perform the calculation at a fixed temperature and obtain the eigenvalue and the eigenvector for @xmath1-wave and @xmath2-wave pairing symmetries .
the comparison of the eigenvalue at a fixed ( low ) temperature for different pairing symmetries tells us which pairing state is actually realized in the system .
the eigenfunction of the eliashberg equation is referred to as the superconducting gap function .
to ( @xmath21,0 ) . @xmath46ev with @xmath47 @xmath48-point meshes.,width=283 ] we first compare the five- and ten - orbital models at a high temperature of @xmath46ev .
we take @xmath47 @xmath48-point meshes and 128 matsubara frequencies .
the interaction reducing factor is @xmath49 .
the stoner factor calculated for this parameter set is @xmath50 and 0.95 for the ten- and five - orbital models , respectively , in reasonable agreement . in fig . [
1025chi ] , we show the spin susceptibility along the wave vector ( 0,0 ) to @xmath51 in the unfolded brillouin zone .
the slight difference in the stoner factor results in a difference in the peak value of the spin susceptibility , but the @xmath52 position at which the spin susceptibility is maximized is the same between the two models .
it can also be seen from this figure that the three - dimensionality of the spin fluctuations is very small . in fig.[1025gap ] , we compare the superconducting gap for the two models in the folded brillouin zone . for the five - orbital model ,
the results are obtained first in the unfolded brillouin zone , and then those in the second brillouin zone of the original bct lattice structure is moved to the first brillouin zone . here again they have similar structures for most of the @xmath7 planes given here .
the eigenvalue of the eliashberg equation is 0.66 and 0.76 for @xmath1-wave , and 0.51 and 0.60 for @xmath2-wave for the ten- and five - orbital models , respectively .
since the fermi surface is different between the two models for some @xmath7 portions , the gap function differs between the two at those portions of the fermi surface .
nonetheless , the present comparison shows that the five - orbital model can be considered as reliable as far as the form of the superconducting gap on most of the @xmath7 planes are concerned .
hereafter , we concentrate on the five - orbital model and go down to lower temperatures , where larger - number @xmath48-point meshes and matsubara frequencies are required . gap function between ( a ) the ten - orbital and ( b ) the five - orbital models at @xmath46ev with @xmath47 @xmath48-point meshes .
contour plots are shown for five horizontal cuts around @xmath17-z and x - p - x in the original brillouin zone of the bct lattice .
the five - orbital results , obtained originally in the unfolded brillouin zone , are refolded into this brillouin zone .
solid black lines are the fermi surface and the red dashed lines are the nodal lines of the gap . the brillouin zone is shown along with the regions where the gap is presented.,width=207 ]
ev with ( a ) @xmath53 ( kfe@xmath0as@xmath0 ) , ( b ) @xmath54 ( ba@xmath55k@xmath56fe@xmath0as@xmath0 ) , and ( c ) @xmath57 ( ba(fe@xmath58co@xmath59)@xmath60as@xmath0 ) .
( d ) the stoner factor against the band filling for the models of kfe@xmath0as@xmath0 and bafe@xmath0as@xmath0 .
the dots indicate the positions of the actual band filling of the corresponding material . at @xmath54
, we adopt the model of ba@xmath55k@xmath56fe@xmath0as@xmath0.,width=302 ] .
@xmath15 ( a),(c ) and @xmath61 ( b),(d ) orbitals for the band fillings @xmath53 ( a),(b ) and @xmath57 ( c),(d ) .
the arrows indicate the wave vectors connecting the portions of the fermi surface having similar orbital character .
these arrows correspond to the peak position in the spin susceptibility.,width=302 ] we now discuss the spin fluctuations using the five - orbital model . here
we take @xmath62ev , adopting @xmath63 @xmath48-point meshes and 512 matsubara frequencies .
one of the merits to adopting the five - orbital model is that it captures the essence of the spin fluctuations observed in neutron - scattering experiments . in fig.[chi ] , we show the contour plot of the spin susceptibility of the five - orbital model in the unfolded brillouin zone at @xmath10 .
in addition to the results for kfe@xmath0as@xmath0 ( band filling @xmath53 ) , we also show for comparison the results for ba@xmath55k@xmath56fe@xmath0as@xmath0 ( @xmath54 ) and ba(fe@xmath58co@xmath59)@xmath0as@xmath0 @xmath64 . for ba@xmath55k@xmath56fe@xmath0as@xmath0
, we use the five - orbital model that is obtained by making a linear combination of the hopping integrals of bafe@xmath0as@xmath0 and kfe@xmath0as@xmath0 using the lattice structure of ba@xmath55k@xmath56fe@xmath0as@xmath0 given in ref .
, as was done in ref . . for @xmath57
, we simply use the five - orbital model of bafe@xmath0as@xmath0 adopting the experimental lattice structure.@xcite the interaction reducing rate is taken to be @xmath65 , so that the stoner factor does not exceed unity in all of the three cases considered here . in fig .
[ chi](a ) , the peak position is located at an incommensurate wave vector [ ( 1 - 2@xmath66)@xmath21,0 ] , @xmath67 , which is in very good agreement with a recent neutron - scattering experiment for kfe@xmath0as@xmath0@xcite . for ba@xmath55k@xmath56fe@xmath0as@xmath0 ( @xmath54 ) , the peak moves to the commensurate position @xmath51 . in the case of electron doping , @xmath57 , the peak moves toward @xmath68 in the unfolded brillouin zone , which is consistent with the experimental results of ba(fe , co)@xmath0as@xmath0 in the high - energy regime . the above tendency has also been obtained in other theoretical studies that use realistic band structures.@xcite in the above , the adopted band structure is different among figs .
[ chi](a),5(b ) and 5(c ) . to focus on the evolution of the spin fluctuations
, we have also fixed the band structure to that of kfe@xmath0as@xmath0 or bafe@xmath0as@xmath0 , and varied the band filling hypothetically from 5.5 to 6.2 .
we find that for both of the band structures , the spin fluctuation becomes commensurate around @xmath69 .
for both of the models , it is found that the peak position continuously moves from the commensurate to the incommensurate position as holes are doped,@xcite indicating that the spin fluctuation has the same origin in the entire ( ba , k)fe@xmath0as@xmath0 system , namely , the interband scattering between the band around @xmath9 and @xmath6 . the stoner factor obtained for the above models
is shown in fig .
[ chi](d ) as functions of the band filling .
this shows that the strength of the spin fluctuations is stronger for the model of bafe@xmath0as@xmath0 than in kfe@xmath0as@xmath0 when compared at the same band filling .
therefore , when the composition of the material is changed continuously from ba to k , we would expect the spin fluctuation to be maximum around the undoped regime ( the ba - rich regime ) , and decreases when the holes are doped by ba@xmath70k replacement .
however , in the heavily hole doped regime ( the k - rich regime ) , the strength of the spin fluctuations should barely decrease , and may become even stronger with hole doping approaching the end material kfe@xmath0as@xmath0 .
as for the electron - hole asymmetry of the incommensurability of the spin fluctuations , we have analyzed its origin in ref . as follows .
the spin fluctuations are mainly governed by the interaction between portions of the fermi surface having similar orbital character , so in fig .
[ orbcha](a)-(d ) , we show the strength of the orbital character on the fermi surfaces of kfe@xmath0as@xmath0 @xmath71 and bafe@xmath0as@xmath0 with @xmath57 in the unfolded brillouin zone . here
we plot the states within a finite energy range @xmath72 with @xmath73ev , which can contribute to the spin fluctuations .
although fermi surfaces near the wave vector @xmath6 are barely present in kfe@xmath0as@xmath0 , there are states in the vicinity of the fermi level originating from a nearly flat band lying close to the fermi level .
the position of this flat band with respect to the fermi level can be clearly seen in the magnifications of figs . [ fig1](a ) and 1(b ) . as mentioned in ref .
, the spin fluctuations develop at wave vectors which bridge the portions of the fermi surface having similar orbital character .
the wave vectors connecting @xmath15 orbital portions and those connecting @xmath61 ( or @xmath74 , not shown ) portions nearly coincide in each of the two cases , so that this wave vector should correspond to the peak position of the spin fluctuations . in the electron - doped case ,
the connecting wave vector deviates from @xmath51 toward @xmath68 , while in the hole - doped case it deviates toward @xmath9 .
thus , the difference of the spin fluctuation incommensurability between electron- and hole - doped cases originates from the multiorbital nature of the system .
finally , we discuss the superconducting gap of kfe@xmath0as@xmath0 .
we obtain the superconducting gap at @xmath62ev for @xmath1-wave and @xmath75-wave pairings . here
we take @xmath76 @xmath48-point meshes and 1024 matsubara frequencies .
we show in fig .
[ fig7 ] the @xmath1-wave ( a ) and @xmath2-wave ( b ) gap functions of the five - orbital model refolded into the original folded brillouin zone . here
we take the interaction reducing rate @xmath77 .
the rate @xmath42 is increased compared to the values adopted for the spin susceptibility because here we focus only on kfe@xmath0as@xmath0 , so that the stoner factor does not exceed unity even for this @xmath42 .
we take larger values of the interaction so as to enhance the eigenvalue of the eliashberg equation , which enables us to obtain a more reliable form of the superconducting gap .
the eigenvalue of the eliashberg equation is @xmath78 and @xmath79 , close to unity .
these values indicate that the pairing symmetry competition is close .
in fact , the possibility of @xmath2-wave pairing was proposed in a recent functional renormalization group study@xcite and an effective hamiltonian approach.@xcite even at this low temperature , the gap is essentially similar to the one obtained at higher temperature in sec . [ 1025sec ] .
the @xmath1-wave gap is basically @xmath8-wave in the sense that the gap function changes sign between the states near @xmath80 and @xmath81 .
the origin of this sign change is strongly related to the spin fluctuation discussed in the present section .
namely , although the fermi surface is barely present around @xmath81 , there are states in the vicinity of the fermi level originating from the bands that produces the electron fermi surface for smaller hole content .
the pair scattering from those states to the hole fermi surface and vice versa mediated by the incommensurate spin fluctuations is the origin of the sign change of the gap function . on the other hand
, the @xmath2-wave pairing originates essentially due to the intraband repulsive interaction .
in fact , we find that the spin fluctuations originating from the intraband interaction ( within the hole fermi surfaces ) develops especially for the band filling below @xmath53 . -wave ( a ) and @xmath75-wave ( b ) at @xmath62ev .
the solid black lines represent the fermi surfaces , while dashed red lines are the nodes of the gap .
( c ) a schematic figure of the horizontal nodes in the @xmath18 fermi surface in the case of @xmath1-wave pairing .
the arrow indicates the intraband scattering that drives the sign change.,width=170 ] although the origin of the stabilization of the @xmath1-wave gap is the incommensurate spin fluctuation and the gap function ( eigenfunction of the eliashberg equation , to be strict ) changes its sign between @xmath82 and @xmath83 , theoretically , the fermi surface around the latter wave vector is barely present for kfe@xmath0as@xmath0 , so , experimentally , this sign change should not be detectable .
however , there are several possibilities of ( nearly ) vanishing gap function , which may be relevant to the experimental observations that suggest the presence of nodes in the superconducting gap .
first , the @xmath1-wave superconducting gap on the @xmath15 fermi surface is rather small .
one can see in fig .
[ fig7](a ) that the nodal line runs somewhat close to the @xmath15 fermi surface .
this may experimentally be detected as nearly vanishing gap , although there is no actual sign change in the gap function here .
second , the @xmath1-wave gap on the @xmath18 fermi surface has different signs between the @xmath7 plane that includes the @xmath17 point and that with the z point , as seen in fig .
[ fig7](a ) .
this means that the gap has horizontal nodes on the @xmath18 fermi surface , which is schematically shown in fig .
[ fig7](c ) .
although the five - orbital model is not reliable concerning this portion of the fermi surface , this sign change is also found in the ten - orbital model ( at higher temperature ) as seen in fig .
[ 1025gap ] in sec.[1025sec ] .
the sign change in this @xmath18 hole fermi surface is a common feature found in the rpa studies of the 122 systems,@xcite and indeed some recent experiments@xcite support this possibility in phosphorous doped 122 materials .
third , there is a possibility of a @xmath2-wave gap , which is in close competition with the @xmath1-wave .
this possibility , however , may not be consistent with an experiment on the magnetic vortex lattice , which shows that the gap is not strongly anisotropic within the planes.@xcite
in the present study , we have constructed an approximate five - orbital model of kfe@xmath0as@xmath0 and compared it with the original ten - orbital model .
we find the former to be reliable as far as the peak position of the spin fluctuations and form of the gap function are concerned .
using the five - orbital model , we have investigated the spin fluctuations and the spin - fluctuation - mediated superconducting gap of kfe@xmath0as@xmath0 .
we find that the spin fluctuation has an incommensurate peak structure , whose origin is the multiorbital nature of the fermi surfaces .
the peak position of the spin fluctuation is in excellent agreement with the experiment.@xcite as for the superconductivity , @xmath1-wave and @xmath2-wave pairings are found to be in close competition .
the @xmath1-wave gap has essentially the @xmath8-wave form in the sense that the gap function changes its sign between the wave vectors around @xmath9 and @xmath6 . although there are barely fermi surfaces around @xmath6 , the states in the vicinity of the fermi level are nonetheless effective to give the incommensurate spin fluctuations and thus the sign change in the superconducting gap function .
although the sign change exists in the gap function , this should not be detectable in experiments because the fermi surface is barely present around @xmath84 .
as for the explanation for the nodes in the superconducting gap observed experimentally , besides a possible @xmath2-wave pairing , we raise two possibilities for the @xmath1-wave pairing : either horizontal nodes in the @xmath85 fermi surface or the small gap on the entire @xmath15 fermi surface .
we are grateful to c.h .
lee , h. fukazawa , h. ikeda , t. shibauchi y. matsuda , h. eisaki , k. yamada , t. shimojima , and k. okazaki for valuable discussions .
numerical calculations were performed at the facilities of the information technology center , university of tokyo , and also at the supercomputer center , issp , university of tokyo .
this study has been partially supported by grant - in - aid for scientific research from mext of japan and from the japan society for the promotion of science .
99 j. d. fletcher , a. serafin , l. malone , j. g. analytis , j - h chu , a.s .
erickson , i. r. fisher , and a. carrington , phys .
lett . * 102 * , 147001 ( 2009 ) . c. w. hicks , t. m. lippman , m. e. huber , j. g. analytis , j .- h .
chu , a. s. erickson , i. r. fisher , and k. a. moler , phys .
* 103 * , 127003 ( 2009 ) .
m. yamashita , n. nakata , y. senshu , s. tonegawa , k. ikada , k. hashimoto , h. sugawara , t. shibauchi , and y. matsuda , phys .
b * 80 * , 220509(r ) ( 2009 ) .
k. hashimoto , m. yamashita , s. kasahara , y. senshu , n. nakata , s. tonegawa , k. ikada , a. serafin , a. carrington , t. terashima , h. ikeda , t. shibauchi and y. matsuda , phys .
b * 81 * , 220501(r ) ( 2010 ) .
y. nakai , t. iye , s. kitagawa , k. ishida , s. kasahara , t. shibauchi , y. matsuda and t. terashima : phys .
b. * 81 * ( 2010 ) 020503(r ) .
m. yamashita , y. senshu , t. shibauchi , s. kasahara , k. hashimoto , d. watanabe , h. ikeda , t. terashima , i. vekhter , a.b .
vorontsov , and y. matsuda , phys .
b * 84 * 060507(r ) ( 2011 )
. k. hashimoto , s. kasahara , r. katsumata , y. mizukami , m. yamashita , h. ikeda , t. terashima , a. carrington , y. matsuda , and t. shibauchi , arxiv:1107.4505 .
i. i. mazin , t. p. devereaux , j. g. analytis , jiun - haw chu , i. r. fisher , b. muschler and r. hackl , phys .
b * 82 * , 180502 ( 2010 ) . c. martin , h. kim , r. t. gordon , n. ni , v. g.kogan , s. l. budko , p. c. canfield , m. a. tanatar , r. prozorov , phys . rev .
b * 81 * , 060505 ( 2010 ) .
k. kuroki , h. usui , s. onari , r. arita and h. aoki , phys .
b * 79 * , 224511 ( 2009 ) .
f. wang , h. zhai , and d .- h .
lee , phys .
b * 81 * , 184512 ( 2010 ) .
h. ikeda , r. arita , and j. kune , phys .
b * 81 * , 054502 ( 2010 ) . t. kariyado and m. ogata , j. phys .
jpn . * 79 * , 033703 ( 2010 ) .
r. thomale , c. platt , w. hanke , b.a .
bernevig , phys .
lett . * 106 * , 187003 ( 2011 ) .
t. sato , k. nakayama , y. sekiba , p. richard , y .- m .
xu , s. souma , t. takahashi , g. f. chen , j. l. luo , n. l. wang and h. ding , phys .
. lett . * 103 * , 047002 ( 2009 ) .
t. terashima , m. kimata , n. kurita , h. satsukawa , a. harada , k. hazama , m. imai , a. sato , k. kihou , c .- h .
lee , h. kito , h. eisaki , a. iyo , t. saito , h. fukazawa , y. kohori , h. harima , and s. uji , j. phys .
jpn . * 79 * , 053702 ( 2010 ) .
h. fukazawa , y. yamada , k. kondo , t. saito , y. kohori , k. kuga , y. matsumoto , s. nakatsuji , h.kito , p.m. shirage , k. kihou , n. takeshita , c. h. lee , a. iyo and h. eisaki , j. phys .
. jpn . * 78 * , 083712 ( 2009 ) .
j. k. dong , s. y. zhou , t. y. guan , h. zhang , y. f. dai , x. qiu , x. f. wang , y. he , x. h. chen and s. y. li , phys . rev .
* 104 * , 087005 ( 2010 ) .
s.w . zhang , l. ma , y.d .
hou , j. zhang , t .- l .
xia , g.f.chen , j.p .
luke and w. yu , phys .
b * 81 * , 012503 ( 2010 ) .
k. hashimoto , a. serafin , s. tonegawa , r. katsumata , r. okazaki , t. saito , h. fukazawa , y. kohori , k. kihou , c. h. lee , a. iyo , h. eisaki , h. ikeda , y. matsuda , a. carrington and t. shibauchi , phys .
b * 82 * , 014526 ( 2010 ) .
r. thomale , c. platt , w. hanke , j. hu , and b. a. bernevig , phys .
lett . * 107 * , 117001 ( 2011 ) .
s. maiti , m. korshunov , t. maier , p.j .
hirschfeld , and a.v .
chubukov , phys .
lett . * 107 * , 147002 ( 2011 ) , arxiv:1109.0498 . c. h. lee , k. kihou , h. kawano - furukawa , t. saito , a. iyo , h. eisaki , h. fukazawa , y. kohori , k. suzuki , h. usui , k. kuroki and k. yamada , phys
lett . * 106*,067003 ( 2011 ) .
s. rozsa , h. u. schuster , z. naturforsch .
b * 36 * , 1668 ( 1981 ) .
s. baroni , a. dal corso , s. de gironcoli , p. giannozzi , c. cavazzoni , g. ballabio , s. scandolo , g. chiarotti , p. focher , a. pasquarello , k. laasonen , a. trave , r. car , n. marzari , a. kokalj : http://www.pwscf.org/. here we adopt the exchange correlation functional introduced by j. p. perdew , k. burke , and y. wang ( phys .
b * 54 * , 16533 ( 1996 ) ) , and the wave functions are expanded by plane waves up to a cutoff energy of 40 ry .
8@xmath86 @xmath48-point meshes are used .
n. marzari and d. vanderbilt , phys . rev .
b * 56 * 12847 ( 1997 ) ; i. souza , n. marzari and d. vanderbilt , phys .
b * 65 * , 035109 ( 2001 ) .
the wannier functions are generated by the code developed by a. a. mostofi , j. r. yates , n. marzari , i. souza and d. vanderbilt , ( http://www.wannier.org/ ) for the energy window @xmath87 ev @xmath88 3.6ev , where @xmath89 is the eigenenergy of the bloch states and @xmath90 the fermi energy .
k. momma and f. izumi , j. appl .
crystallogr . *
41 * , 653 ( 2008 ) .
k. kuroki , s. onari , r. arita , h. usui , y. tanaka , h. kontani , and h. aoki , phys .
* 101 * , 087004 ( 2008 ) . a five orbital model for bafe@xmath0as@xmath0
was constructed in ref .
, which has a different orbital character in the @xmath18 fermi surface compared to our model . t. miyake , k. nakamura , r. arita , and m. imada , j. phys .
. jpn . * 79 * , 044705 ( 2010 ) . m. rotter , m. tegel , d. johrendt , phys . rev
* 101 * , 107006 ( 2008 ) .
k. suzuki , h. usui , k. kuroki , j. phys .
. jpn . * 80 * , 013710 ( 2010 ) m. rotter , m. tegel , d. johrendt , i. schellenberg , w. hermes and r. pttgen , phys .
b * 78 * , 020503 ( 2008 ) .
s. graser , a. f. kemper , t. a. maier , h .-
cheng , p. j. hirschfeld , and d. j. scalapino , phys .
b * 81 * , 214503 ( 2010 ) . j. t. park , d. s. inosov , a .
yaresko , s. graser , d. l. sun , ph .
bourges , y. sidis , y. li , j .- h .
kim , d. haug , a. ivanov , k. hradil , a. schneidewind , p. link , e. faulhaber , i. glavatskyy , c. t. lin , b. keimer , and v. hinkov , phys .
b. * 82 * , 134503 ( 2010 ) .
j. knolle , i. eremin , and r. moessner phys .
b * 83 * , 224503 ( 2011 ) a. n. yaresko , g .- q . liu , v. n. antonov , and o. k. andersen , phys .
b * 79 * , 144421 ( 2009 ) .
k. suzuki , h. usui , and k. kuroki , to be published in the proceedings of lt26 t. dulguun , h. mukuda , h. kinouchi , m. yashima , y. kitaoka , t. kobayashi , s. miyasaka , and s. tajima , arxiv:1108.4480 .
y. zhang , z. r. ye , q. q. ge , f. chen , j. jiang , m. xu , b. p. xie , and d. l. feng , arxiv:1109.0229 . h. kawano - furukawa , c. j. bowell , j. s. white , r.w .
heslop , a.s .
cameron , e.m .
forgan , k. kihou , c. h. lee , a. iyo , h. eisaki , t. saito , h. fukazawa , y. kohori , r. cubitt , c. d. dewhurst , j. l. gavilano and m. zolliker , phys . rev .
b * 84 * , 024507 ( 2011 ) . | we study the relation between the spin fluctuation and superconductivity in a heavily hole doped end material kfe@xmath0as@xmath0 .
we construct a five orbital model by approximately unfolding the brillouin zone of the three - dimensional ten - orbital model obtained from first - principles calculation . by applying the random - phase - approximation
, we obtain the spin susceptibility and solve the linearized eliashberg equation . the incommensurate spin fluctuation observed
experimentally is understood as originating from interband interactions , where the multiorbital nature of the band structure results in an electron - hole asymmetry of the incommensurability in the whole iron - based superconductor family . as for superconductivity , @xmath1-wave and @xmath2-wave pairings
are found to be in close competition , where the sign change in the gap function in the former is driven by the incommensurate spin fluctuations .
we raise several possible explanations for the nodes in the superconducting gap of kfe@xmath0as@xmath0 observed experimentally . |
in 1998 watts and strogatz introduced in a seminal paper small world networks and provided a simple model of such graphs @xcite .
the key property of small world graphs is the simultaneous presence of small diameter ( at most logarithmic in the vertex size ) and high clustering coefficient .
extensive investigations over the last decade of real networks like the worldwideweb , citation networks , friendship networks and many others have shown that these graphs are really of small world type .
but there is a second remarkable property shared by most real networks namely that they have a power law like fat tail distribution ( see @xcite for an excellent recent survey about the whole subject ) .
there are many models which produce graphs with small diameter and power law distribution .
if additionally local search rules are incorporated one easily obtains high clustering as well .
but up to now it was common believe that for the small diameter property a certain amount of essentially independent random edge formations like in classical random graphs where necessary . to obtain a scale free distribution for the degree the basic conviction was that the random edge formation should be biased according to the `` the richer you are the richer you get- principle '' @xcite where richness is just measured by the degree of a vertex . in the present paper
we will present a model which shows that even by entirely local edge formation rules without preferential choices graphs are obtainable which admit all the three above mentioned properties .
the results are entirely based on numerical simulations since an analytic treatment of the model seems to be rather difficult .
nevertheless we hope to present in a forthcoming paper a more theoretical study of the main properties .
looking at the edge formation process in social networks there are essential two ways to get a new contact . first by the random event of meeting somebody for instance in a train or airplane .
humans refer to this kind of relation creation often as fate .
second there are contacts which where created by the local rule `` let me introduce you to one of my friends '' , certainly a very common process in real life . in the language of graphs this translates to the formation of an edge between two vertices
say @xmath0 and @xmath1 which had distance 2 with respect to the underlying graph metric and we refer to this kind of edge formation as the `` my friends are your friends '' principle ( ffp ) .
graphs which where partially build by using the ff - principle are already described and analyzed in @xcite . in the following we want to describe a model based on the iteration of this principle - the iff - model . a typical example of an essential ff - graph (
although a directed one and with increasing number of vertices ) is the citation network .
in detail , we want to discuss the following model of a time dependent random graph space @xmath2 with @xmath3 vertices and an explicit integer parameter @xmath4 tuning the total number of edges .
we start with a circular chain of @xmath3 vertices with the usual nearest neighbor edges , i.e. vertex @xmath5 is connected to vertex @xmath3 and @xmath6 , vertex @xmath6 is connected to vertex @xmath5 and @xmath7 and so on .
we call this configuration a circle and refer to the corresponding edges as fixed base edges .
the random graph space @xmath2 will be defined through the following algorithm : in each of the first @xmath4 time - steps , a vertex say @xmath0 is chosen randomly . from the set of neighbors @xmath8 of this vertex ( at the given time @xmath9 ) one element
say @xmath1 is chosen randomly and then randomly from @xmath10 the neighbor set of @xmath1 at time @xmath9 with the exclusion of @xmath0 another element say @xmath11 .
if there was no edge between @xmath11 and @xmath0 a new one is created between the two vertices otherwise nothing happens . in the first case
we say that @xmath0 has created an ff - edge to @xmath12 respectively @xmath11 was chosen by @xmath0 . after repeating @xmath4 times
the above procedure one continues for @xmath13 the same way with the additional rule that whenever a vertex say @xmath0 has created a new ff - edge randomly an ff - edge containing @xmath0 is deleted .
note that the total number of edges stays therefore constant for @xmath14 the first surprising observation is the collapse of the diameter from @xmath15 to @xmath16 for sufficiently large @xmath9 , @xmath4 and typical elements of @xmath2 .
this happens despite the fact that only local edge formations where used .
[ fig_1 ] shows the diameter versus @xmath9 for different @xmath3 and @xmath4 .
our simulations indicate that in the case @xmath17 one gets an asymptotic regime in which the diameter is very small and shows no strong dependence on @xmath3 .
[ tab_1 ] shows for three choices of @xmath18 the mean diameter @xmath19 in the asymptotic regime .
the diameter @xmath20 is approximately proportional to @xmath21 which is one of the characteristics of small world graphs . in the case
@xmath22 an asymptotic regime with still small diameter but large fluctuations starts approximately at @xmath23 .
note that this value is much bigger than the collapse time in the cases @xmath24 .
it seems that @xmath18 @xmath25 is just the borderline where a collapse of the diameter appears .
this corresponds to the case where the mean number of non - base edges equals @xmath3 . for smaller values of @xmath4 we found the diameter to stay of order @xmath26 with an almost linear function @xmath27 ( see fig .
[ fig_2 ] ) .
this indicates an interesting phase transition in the total number of edges reminiscent to the famous phase transition in the size of the largest component in classical random graph spaces like the @xmath28 the space of all graphs with @xmath3 vertices and @xmath29 edges equipped with the uniform probability measure . ;
@xmath30 versus time , averaged over 10 realizations ] ; @xmath30 versus time , averaged over 10 realizations ] ; @xmath30 versus time , averaged over 10 realizations ] * * * * [ tab_1 ] @xmath31 calculated by averaging the values @xmath32 for @xmath33 ( @xmath34 ) or @xmath35 ( @xmath36 ) and 10 realizations ; effective non - base edges @xmath37 ; the mean diameter @xmath38 of @xmath39 is calculated by averaging over 100 realizations , left axis : mean diameter @xmath30 calculated at @xmath40 ( boxes with lines ) and @xmath41 ( triangles with lines ) , right axis : effective non - base edges @xmath42 ( crosses ) versus @xmath18 calculated at @xmath40 ; averaged over 10 realizations ] * * fig .
[ fig_3 ] shows one realization of the time evolution of a typical graph from @xmath43 with @xmath44 and @xmath45 .
it illustrates how the diameter of the graph becomes very small only by using the iff principle . before the asymptotic state
is reached an interesting symmetry breaking can be observed : the appearance of a few components with long range edges ( taken the circle as a reference system ) and hence small diameter .
the components itself are only connected via the circle skeleton .
it takes a relative long time till the components finally merge and the circle is filled uniformly with edges .
the same phenomenon can be seen for larger values of @xmath3 and different @xmath18 .
, @xmath44 ] * * of @xmath47 versus @xmath48 ; averaged over 100 realizations ] * * it is interesting to compare our model with a version of the random graph space @xmath49 where an underlying circle is added to keep the graph connected . to be precise
let @xmath50 be the random graph space where @xmath29 edges are randomly added to a circle graph with @xmath3 vertices .
[ fig_4 ] shows the diameters @xmath38 of the average of a few samples of @xmath50 as a function of @xmath48 . here
the diameter clearly decays like @xmath51 and reaches the @xmath21 regime already for very small values of @xmath48 ( @xmath52 ) .
note that this value is much smaller than the threshold value @xmath53 @xmath54 for the emergence of a giant component in @xmath55 .
let us remark that the @xmath50 model is very close in spirit to the original small world model of watts and strogatz .
the main difference is with respect to the clustering coefficient since the underlying skeleton the circle has zero clustering coefficient .
replacing the standard circle with next neighbor connections by one where also next - next neighbor connections are edges would give an essentially equivalent model to the watts - strogatz one .
another unexpected property of our model is the fat tail of the degree distribution .
[ fig_5]a / b show the degree distribution for @xmath56 and @xmath57 at various times .
, * a * @xmath44 , @xmath58 ( circles ) , @xmath59 ( boxes ) , @xmath60 ( triangles ) , * b * @xmath61 , @xmath62 ( circles ) , @xmath59 ( boxes ) , @xmath60 ( triangles ) ; the big dotted line shows the degree distribution of @xmath63 , all values averaged over 10 realizations ] , * a * @xmath44 , @xmath58 ( circles ) , @xmath59 ( boxes ) , @xmath60 ( triangles ) , * b * @xmath61 , @xmath62 ( circles ) , @xmath59 ( boxes ) , @xmath60 ( triangles ) ; the big dotted line shows the degree distribution of @xmath63 , all values averaged over 10 realizations ] * * let us compare the degree distribution in the asymptotic regime with those of the random graph @xmath64 , where @xmath42 is the expectation of the number of edges in the iff random graph space @xmath2 ( see fig .
[ fig_2 ] for a plot of @xmath65 ) .
the resulting degree distributions are shown in fig .
[ fig_5]a and fig .
[ fig_5]b by big dotted lines . at the beginning
( that is for @xmath66 ) the iff - model has essential the degree distribution of @xmath67 but in the asymptotic regime the distributions differ drastically since the iff - model has gotten a fat tail . at the moment the limited numerical data still do nt allow to check whether it is really a power law .
we want to close this section with a remark about the degree preferences if a vertex @xmath0 was chosen in the formation of a new edge .
although never explicitly included in the model there is a strong numerical evidence ( see fig .
[ fig_6 ] ) that the probability of a vertex to be chosen is proportional to its degree .
that is one of the basic assumptions in the albert barabsi model .
a heuristic explanation for this property in our model is the following argument based on two independence assumptions .
let @xmath68 and @xmath69 be the expected values for the first and second neighborhood sizes .
assume that the conditional expectation of @xmath70 for @xmath0 having degree @xmath71 equals @xmath72 and assume further that for @xmath73 the expected value of @xmath74 equals @xmath69 . clearly with these conditions
one can easily compute the probability of a vertex @xmath0 with degree @xmath71 to be chosen within the process of an edge formation to be equal to @xmath75 .
, @xmath44 and @xmath58 ( circles ) , @xmath59 ( boxes ) , @xmath60 ( triangles ) , the big dotted line shows the constant probability in @xmath76 averaged over 10 realizations ; basic edges system : circle ] * * a basic quantity to measure the local clustering around a vertex @xmath0 is the number of triangles @xmath77 containing @xmath0 as a vertex . note that for classical random graph spaces the expectation of this number is is of order @xmath78 .
we are interested in the averaged number of triangles in the asymptotic regime of the iff - model .
[ fig_7 ] shows the mean number of triangles @xmath79 for all vertices with a fixed degree for @xmath80 and @xmath60 .
the mean number of triangles per vertex ( independent of its degree ) is shown by the big symbols on the figure frame .
the plot for the two time - values coincides practically .
the number of triangles increases nearly linear with the degree @xmath71 .
the reason for the fluctuations for high @xmath71 is the low number of vertices with such high degrees . fig . [ fig_7 ] shows also the mean clustering coefficient @xmath81 which is directly connected to the mean number of triangles .
namely the clustering coefficient @xmath82 of a vertex @xmath0 with degree @xmath71 is given by the normalized triangle coefficient @xmath83 . and mean clustering coefficient @xmath81 per vertex versus degree @xmath71 , @xmath84 , @xmath44 , the big symbol shows the mean values of all vertices independent of their degree , all values averaged over 10 realizations , the basic edges system forms a circle : number of triangles ( left axes ) : @xmath59 ( dashed line / plus sign ) , @xmath85 ( dotted line / cross ) ; clustering coefficient ( right axes ) : @xmath59 ( boxes ) , @xmath60 ( circles ) ] * * * * [ tab_2 ] @xmath86 calculated with @xmath84 and after @xmath9 steps , averaged over 10 realizations there is another interesting quantity , defined via the second shortest distance between two vertices , which characterizes additional clustering properties . for pairs of vertices with a common edge
let @xmath87 be the distance between @xmath0 and @xmath1 after removal of the connecting edge .
define for a given graph @xmath88 the quantity @xmath89 as the fraction of pairs of vertices with distance @xmath5 whose @xmath90-distance is larger @xmath91 ( for a given random graph space @xmath92 let @xmath93 the expectation of @xmath94 ) . in case when the diameter of a random graph space is small due to the presence of independently generated edges the @xmath95 value is usually very large since no short second - shortest paths between two vertices connected by an independently generated edge exist .
this remains also true for the watts - strogatz small world graph . in our model
we get by the very construction process a small value of @xmath96 ( 3 ) similar to the situation met in real networks ( see tab .
up to now , the system of base edges forms a circle .
it is natural to ask if one gets qualitatively the same results with a two - dimensional system of base edges like a torus lattice .
this means that at the beginning the vertices are connected in form of a torus or a lattice with periodic boundaries .
the algorithm to generate ( and define ) @xmath97 remains the same .
for the diameter one gets again an asymptotic regime with collapse up to logarithmic size but the asymptotics starts earlier as in the `` circle''-case . even for @xmath22
we get an asymptotic regime starting at @xmath98 .
[ tab_3 ] shows the mean diameter @xmath20 in the case of a torus base for various values of @xmath4 and @xmath3 . * * [ tab_3 ] @xmath99 calculated with the values of the mean diameters in the range @xmath100 averaged over 10 realizations ; the value @xmath38 is calculated by averaging over 100 realizations , @xmath44 ; * a , b * the edges at different times are plotted , the dots characterize the vertices , the color of an edge characterizes its length @xmath101 ; * c * the dots characterize the vertices , the color characterize the length @xmath101 of the longest edge of the vertex ] * * there is another difference to the one - dimensional case which can be seen in the realization shown in fig .
[ fig_8 ] : the graph evolution is more uniformly with a two - dimensional base edge system than with a one - dimensional ( compare to fig .
[ fig_3 ] ) .
we obtain qualitatively the same results as in the one - dimensional case with respect to the degree distribution and the clustering properties for @xmath102 and @xmath44 , @xmath61 . in the asymptotic regime one
sees again a clear fat tail distribution .
the distribution of the number of triangles and the clustering coefficient for @xmath103 and @xmath44 gives qualitatively the same results as in the `` circle''-case , i.e. the distributions are nearly equal for @xmath59 and @xmath60 and we have a linear dependence of the number of triangles on the degree @xmath104 finally we want to discuss the stability of the graphs in the asymptotic regime under removal of the circle edges or torus edges , i.e. at @xmath105 we remove the base edges and examine in which way the graph breaks into connected components and how these components evolve by continuing using our algorithm .
[ fig_9 ] shows the number of connected components with different size @xmath106 ( number of vertices ) and different diameter @xmath30 for @xmath107 directly after deleting the base edges at @xmath105 for the `` circle '' and the `` torus '' case .
there is still one big component .
the time evolution of the biggest component can be seen in fig .
[ fig_10 ] which shows its size @xmath106 and diameter @xmath30 versus time .
an interesting result is that the diameter of the biggest component is nearly constant for @xmath108 , nearly independent of @xmath18 and also nearly independent of using a circle or a torus as base edge system .
let us note that without the skeleton there can be no merging of different components by the very nature of the iff principle .
therefore , as can easily be seen , the only stable asymptotic configurations are components which are totally connected ( i.e. every vertex has an edge to all other vertices in the component ) since no further changes in such components can happen . but it is very likely that the time scale till this phenomenon can be seen is huge compared to the time scales we where studying . to compare with the classical random graph situation observe that the removal of the circle in @xmath109 gives just the model @xmath110 the component distribution for this random graph space is well known . especially for @xmath111
there is always a giant component of size @xmath112 but in contrast to our model , the diameter @xmath30 of the biggest component depends via the constant on @xmath113 . ,
@xmath44 , one realization ] * * ; * b * time dependence of its size @xmath106 ; circle , @xmath84 , @xmath44 ( plus signs ) ; circle , @xmath84 , @xmath61 ( crosses ) ; torus , @xmath103 , @xmath44 ( circles ) ; torus , @xmath103 , @xmath61 ( boxes ) ; the symbols on the left vertical axes indicate the values of the parameters directly before deleting the basic edges ; all values averaged over 10 realizations ] ; * b * time dependence of its size @xmath106 ; circle , @xmath84 , @xmath44 ( plus signs ) ; circle , @xmath84 , @xmath61 ( crosses ) ; torus , @xmath103 , @xmath44 ( circles ) ; torus , @xmath103 , @xmath61 ( boxes ) ; the symbols on the left vertical axes indicate the values of the parameters directly before deleting the basic edges ; all values averaged over 10 realizations ] * *
we have presented a model where the iteration of the my friends are your friends ( ff )
principle has produced random graphs with small world properties ( logarithmic diameter , high clustering ) and a fat tail distribution for the degree .
the model used a fixed regular graph of large diameter as a skeleton , reminiscent to a pre - given geographical structure with pre - given neighborhood relations for vertex pairs .
but in contrast to the small world graphs by watts and strogatz no random , global edge formation processes are involved to obtain small diameter .
the high clustering is an immediate consequence of the ff -principle .
the fat tail distribution is surprising since no preferential choice mechanisms are contained in the graph generation algorithm .
so far our investigations are entirely numerical and clearly a more theoretical explanation of the observed phenomena is the desired next step .
first , in our model the total number of edges is kept fixed after the build up phase . instead of that one could use probabilistic rules which keep the number of edges only fixed in mean .
this would match better real situations where the ff - principle is of relevance .
second , like in the albert & barabsi network @xcite , a growing number of vertices could be considered .
growth should happen here in form of offsprings of already existing vertices to be able to apply the ff - principle to the
newcomers .
in this situation it could well be the case that even without the iteration of the ff - principle every vertex when entering the network forms just once a number of edges according to the ff - rule
small world graphs plus fat tail for the degree distribution are obtainable .
this model variant is actually the typical situation for the growth and formation of the network of citations or collaboration . | we study graphs obtained by successive creation and destruction of edges into small neighborhoods of the vertices . starting with a circle graph of large diameter
we obtain small world graphs with logarithmic diameter , high clustering coefficients and a fat tail distribution for the degree .
only local edge formation processes are involved and no preferential attachment was used .
furthermore we found an interesting phase transition with respect to the initial conditions . |
in 2009 , in inclusive @xmath17 interactions , two new charmed - strange states @xmath4 and @xmath5 were observed by the babar collaboration in both @xmath18 and @xmath19 channels@xcite .
the available experimental results on these two states are as follows : @xmath20 with @xmath21 and @xmath22 . in @xmath19 channel
, the babar collaboration also found the evidence for the @xmath6 whose mass and width are @xmath23 mev and @xmath24 mev , respectively .
there is no signal of @xmath6 in @xmath18 channel@xcite .
more recently , in inclusive @xmath17 collisions , four new charmed states @xmath0 , @xmath1 , @xmath2 , and @xmath3 were found by the babar collaboration@xcite . the @xmath0 and @xmath2 were observed in @xmath25 channel , the @xmath3 was observed in @xmath26 channel , and the @xmath1 was observed in both @xmath25 and @xmath26 channels .
the isospin partners of the @xmath27 and @xmath28 were also observed in the @xmath29 channel .
the resulting masses and widths of these four states with neutral - charge are @xmath30 and the following ratios of branching fractions were also obtained : @xmath31 due to the poor information on the higher excitations of @xmath32 and @xmath33 mesons , the find of these open - charm states is clearly of importance to complete the @xmath32 and @xmath33 spectra . to understand their observed properties ,
various efforts have been carried out under the assumption that all the observed open - charm states are dominated by the simple @xmath34 quark content@xcite .
it is natural and necessary to exhaust the possible conventional @xmath34 descriptions before resorting to more exotic interpretations@xcite .
further theoretical efforts are still required in order to satisfactorily explain the data concerning these open - charm states . in this work , we shall investigate the masses as well as strong and radiative decays of these newly observed states in the nonrelativistic constituent quark model and try to clarify their possible quarkonium assignments by comparing our predictions with the experiment .
the organization of this paper is as follows . in sec .
ii , we calculate the open - charm mesons masses in a nonrelativistic constituent quark model and give the possible assignments for these open - charm states based on their observed masses and decay modes . in sec .
iii , we investigate , with the @xmath35 decay model , the strong decays of these states for different possible assignments . the radiative transitions of these states are given in sec .
the summary and conclusion are given in sec .
to estimate the masses of @xmath36 and @xmath37 states , we employ a simple nonrelativistic constituent quark model which was proposed by lakhina and swanson and turns out to be able to describe the heavy - light meson and the charmonium masses with reasonable accuracy.@xcite . in this model ,
the hamiltonian is @xmath38 where @xmath39 is the zeroth - order hamiltonian , @xmath40 is the spin - dependent hamiltonian , and @xmath41 is a constant .
the @xmath39 is @xmath42 where @xmath43=@xmath44 is the @xmath34 separation , @xmath45 ; @xmath46 and @xmath47 ( @xmath48 and @xmath49 ) are the mass and spin of the constituent quark @xmath50 ( antiquark @xmath51 ) , respectively .
the @xmath40 is @xmath52\cdot{{\bf l}}v_4.\end{aligned}\ ] ] here @xmath53 is the relative orbital angular momentum between @xmath50 and @xmath54 , and @xmath55,\nonumber\\ & & v_2=-\frac{4}{3}\frac{\alpha_s}{r}-\frac{1}{9\pi}\frac{\alpha^2_s}{r}\left [ -18\ln(\sqrt{m_qm_{\bar{q}}}~r)+54\ln(\mu r)+36\gamma_e+29\right ] , \nonumber\\ & & v_3=\frac{4\alpha_s}{r^3}+\frac{1}{3\pi}\frac{\alpha^2_s}{r^3}\left [ -36\ln(\sqrt{m_qm_{\bar{q}}}~r)+54\ln(\mu r)+18\gamma_e+31\right ] , \nonumber\\ & & v_4=\frac{1}{\pi}\frac{\alpha^2_s}{r^3}\ln\left(\frac{m_{\bar{q}}}{m_q}\right),\nonumber\\ & & { \bf s}_{\pm}={\bf s}_q\pm{\bf s}_{\bar{q}},\end{aligned}\ ] ] where @xmath56 and the scale @xmath57 has been set to 1.3 gev .
the model parameters have been chosen to reproduce the low lying @xmath32 and @xmath33 masses and are @xmath58 , @xmath59 gev@xmath60 , @xmath61 gev , @xmath62 gev , and @xmath63 gev .
the constituent quark masses are taken to be @xmath64 gev , @xmath65 gev , and @xmath66 gev .
these quark masses are also used in both strong and radiative decays computations .
the heavy - light mesons are not the charge conjugation eigenstates and hence mixing can occur between the two states with @xmath67 .
this mixing can be parameterized as@xcite @xmath68 where @xmath69 is the mixing angle and @xmath50 denotes @xmath70 or @xmath71 quark .
the @xmath72 refers to the higher mass state . with the help of the mathematica program@xcite , solving the schr@xmath73dinger equation with hamiltonian @xmath39 and evaluating the @xmath40 in leading - order perturbation theory ,
one can obtain the open charm mesons masses as shown in tables [ tab : mesonmass]-[tab : mesonmass1 ] .
, @xmath74 , @xmath75 , @xmath76 , @xmath77 , @xmath78 , @xmath79 , and @xmath80 . ] for comparison , the corresponding masses predicted by some other approaches such as the blankenbecler - sugar equation@xcite and the relativistic quark model@xcite are also listed .
.[tab : mesonmass]the charmed meson masses in gev . [ cols="<,^,^,^,^,^,^,^,^",options="header " , ]
the discovery of @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , and @xmath6 provides a good opportunity to test our present understanding of charmed mesons and is also of importance to further establish the @xmath32 and @xmath33 spectra .
we are trying to shed some light on the natures of these open - charm states by investigating their masses and decays in the nonrelativistic constituent quark model .
we first calculated the charmed meson spectrum in the nonrelativistic constituent quark model .
the overall agreement between our predicted masses and those from other approaches such as the blankenbecler - sugar equation and the relativistic quark models turns out to be satisfactory . according to the observed decay modes and by comparing the measured masses with our predictions
, we presented the possible assignments of these newly observed open - charm states .
mass spectra alone are insufficient to determine the quantum numbers of these open - charm states .
studies on the decay dynamics of these states are needed .
we therefore further evaluated the strong and @xmath15 and @xmath16 radiative decays of these open - charm states for possible assignments .
comparing our predictions with the experiment , we conclude that ( 1 ) if the @xmath0 is indeed the @xmath7 state , its width could be overestimated experimentally ; ( 2 ) the @xmath1 and @xmath4 can be identified as the @xmath8-@xmath9 mixtures ; ( 3 ) if the @xmath2 and @xmath3 are the same state , they could be interpreted as the @xmath10 ; otherwise , they could be assigned as the @xmath10 and @xmath11 , respectively ; ( 4 ) both the @xmath12 and @xmath4 s partner assignments for the @xmath5 are likely ; and ( 5 ) both the @xmath13 and @xmath14 interpretations for the @xmath6 seem possible .
further experimental studies on these states are needed .
we acknowledge prof .
xiang liu for very helpful suggestions and discussions .
this work is supported in part by hancet under contract no .
2006hancet-02 , and by the program for youthful teachers in university of henan province .
p. colangelo , f. defazio , and s. nicotri , phys .
b * 642 * , 48 ( 2006 ) ; + p. colangelo , f. defazio , s. nicotri , and m. rizzi , phys .
d * 77 * , 014012 ( 2008 ) .
p. colangelo , f. defazio , phys .
d * 81 * , 094001 ( 2010 ) .
x. h. zhong and q. zhao , phys .
d * 78 * , 014029 ( 2008 ) . c. amsler et al.(partical data group ) , phys .
b * 667 * , 1 ( 2008 ) .
k. abe et al.(belle collaboration ) , arxiv : hep - ex/0608031 ; + j. brodzicka et al.belle collaboration ) , phys .
rev.lett.*100*,092001 ( 2008 ) .
a. le yaouanc , l. oliver , o. pene , and j - c .
raynal , hadron transitons in the quark model ( gordon and breach science publishers , new york , 1988 ) .
w. roberts and b. silvestr - brac , few - body syst . * 11 * , 171 ( 1992 ) . h. g. blundel , arxiv : hep - ph/9608473 .
d. m. li and b. ma , phys .
d * 77 * , 074004 ( 2008 ) ; + d. m. li and b. ma , phys .
d * 77 * , 094021 ( 2008 ) ; + d. m. li and s. zhou , phys .
d * 78 * , 054013 ( 2008 ) ; + d. m. li and e. wang , eur .
j. c * 63 * , 297 ( 2009 ) .
w. kwong and j. l. rosner , phys .
d * 38 * , 279 ( 1988 ) ; + s. godfrey , phys .
d * 70 * , 054017 ( 2004 ) ; + s. godfrey , phys .
d * 72 * , 054029 ( 2005 ) ; + g. j. ding , j. j. zhu , and m. l. yan , phys .
d * 77 * , 014033 ( 2008 ) . | comparing the measured properties of the newly observed open - charm states @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , and @xmath6 with our predicted spectroscopy and strong decays in a constituent quark model , we find that : ( 1 ) the @xmath7 assignment to @xmath0 remains open for its too broad width determined by experiment ; ( 2 ) the @xmath1 and @xmath4 can be identified as the @xmath8-@xmath9 mixtures ; ( 3 ) if the @xmath3 and @xmath2 are indeed the same resonance , they would be the @xmath10 ; otherwise , they could be assigned as the @xmath10 and @xmath11 , respectively ; ( 4 ) the @xmath5 could be either the @xmath4 s partner or the @xmath12 ; and ( 5 ) both the @xmath13 and @xmath14 interpretations for the @xmath6 seem likely .
the @xmath15 and @xmath16 radiative decays of these sates are also studied .
further experimental efforts are needed to test the present quarkonium assignments for these new open - charm states .
-1 cm * pacs numbers : * 12.39.-x , 13.20.fc , 13.25.ft , 14.40.lb |
consider a sensor network in which a sensor measures a certain physical quantity @xmath0 over time @xmath1 the aim of the sensor is communicating a symbol - by - symbol processed version @xmath2 of the measured sequence @xmath3 to a receiver .
as an example , each element @xmath4 can be obtained by quantizing or denoising @xmath0 , for @xmath1 to this end , based on the observation of @xmath5 and @xmath6 , the sensor communicates a message @xmath7 of @xmath8 bits to the receiver ( @xmath9 is the message rate in bits per source symbol ) .
the receiver is endowed with sensing capabilities , and hence it can measure the physical quantity @xmath6 as well .
however , as the receiver is located further away from the physical source , such measure may come with some delay , say @xmath10 for some @xmath11 . assuming that at time @xmath12 the decoder must put out an estimate @xmath13 of the @xmath14th source symbol @xmath4 by design constraints
, it follows that the estimate @xmath13 can be made to be a function of the message @xmath7 and of the delayed side information @xmath15 ( see @xcite for an illustration ) . following related literature ( e.g. , @xcite )
, we will refer to @xmath16 as the delay for simplicity .
delay @xmath16 may or may not be known at the sensor .
the situation described above can be illustrated schematically as in fig .
[ fig0 ] for the case in which the delay @xmath16 is known at the encoder . in fig .
[ fig0 ] , the encoder ( enc ) represents the sensor and the decoder ( dec ) the receiver .
the decoder at time @xmath14 ( more precisely , @xmath12 ) has access to delayed _ side information _
@xmath17 with delay @xmath18 fig .
[ fig2 ] accounts for a setting where the side information at the decoder , unbeknownst to the encoder , _ may _ be delayed by @xmath16 or not delayed , where the first case is modelled by decoder 1 and the second by decoder 2 .
note that , in the latter case , the receiver has available the sequence @xmath19 at time @xmath14 . for generality , in the setting in fig .
[ fig2 ] , we further assume that the encoder is allowed to send additional information in the form of a message @xmath20 of @xmath21 bits when the side information is not delayed
. this can be justified in the sensor example mentioned above , as a non - delayed side information may entails that the receiver is closer to the transmitter and is thus able to decode an additional message of rate @xmath22 ( bits / source symbol ) . to start ,
let us first assume that sequences @xmath5 and @xmath6 are _ memoryless sources _ so that the entries ( @xmath23 ) are arbitrarily correlated for a given index @xmath14 but independent identically distributed ( i.i.d . ) for different @xmath24 to streamline the discussion , the following lemma summarizes the optimal trade - off between rate @xmath9 and distortion @xmath25 , as measured by a distortion metric @xmath26 , for the point - to - point setting of fig .
[ fig0 ] with memoryless sources .
similar conclusions apply for the more general set - up of fig .
[ fig2 ] .
@xcite for memoryless source , and zero delay , i.e. , @xmath27 , the rate - distortion function for the point - to - point system in fig .
[ fig0 ] is given by the conditional rate - distortion function @xmath28\leq d}}}i(x;z|y).\ ] ] this result remains unchanged even if the decoder has access to non - causal side information , i.e. , if the reconstruction @xmath13 can be based on the entire sequence @xmath6 , rather than only @xmath29 . instead , for strictly positive delay @xmath30
, the rate - distortion function is the same as if there was no side information , namely @xmath31\ensuremath{\leq}d}}}i(x;z)$ ] .
similar conclusions can be easily shown to apply also for the more general model of fig .
[ fig2 ] , as it will be discussed in the paper ( see sec . [
sec : lossy - source - coding ] ) . specifically , if @xmath30 and the sources are memoryless , the rate - distortion function for the system of fig .
2 with @xmath32 reduces to the one obtained by kaspi in @xcite for a model in which decoder 1 has no side information , and , for general @xmath33 , the rate - distortion region coincides with the one obtained in @xcite for a model with no side information at decoder 1 .
we have seen in lemma 1 that , for memoryless sources , no advantages can be accrued by leveraging a ( strictly ) delayed side information , i.e. , with @xmath34 however , this conclusion does not generally hold if the sources have memory . in this context , a number of works have focused on the scenario of fig .
[ fig0 ] where @xmath35 for @xmath36 this entails that the decoder observes sequence @xmath5 itself , but with a delay of @xmath16 symbols .
this setting is typically referred to as _ source coding with feedforward _ , and was introduced in @xcite .
reference @xcite derived the rate - distortion function for this problem ( i.e. , fig .
[ fig0 ] with @xmath35 ) for ergodic and stationary sources in terms of multi - letter mutual informations .
the result was also extended to arbitrary sources using information - spectrum methods .
achievability was obtained via the use of a codebook of codetrees .
the function was explicitly evaluated for some special cases in @xcite ( see also @xcite ) , and @xcite proposed an algorithm for its numerical calculation . the more general case of fig .
[ fig0 ] with @xmath37 was studied in @xcite assuming stationary and ergodic sources @xmath5 and @xmath6 .
the rate - distortion function was expressed in terms of multi - letter mutual informations .
no specific examples were provided for which the function is explicitly computable .
we finally remark that for more complex networks than the ones studied here , strictly delayed side information may be useful even in the presence of memoryless sources .
this was illustrated in @xcite for a multiple description problem with feedforward .
the goal of this work is to characterize the rate - distortion trade - offs for the setting in fig .
[ fig0 ] and the more general set - up in fig .
[ fig2 ] for a specific class of sources @xmath5 and @xmath6 .
specifically , we assume that @xmath6 is a markov chain , and @xmath5 is such that @xmath4 is obtained by passing @xmath0 through a channel @xmath38 for @xmath39 as illustrated in fig .
[ figmarkov ] .
the process is thus a hidden markov model .
this model complies with the type of sensor network scenarios described above , where @xmath6 is the physical quantity of interest , modelled as a markov chain , and @xmath5 is a symbol - by - symbol processed version of @xmath40 the main contributions and the paper organization are as follows .
after the description of the system model in sec .
[ sec : system - model ] , for the source statistics described above , * we derive a single - letter characterization of the minimal rate ( bits / source symbol ) required for asymptotically lossless compression in the point - to - point model of fig .
[ fig0 ] for any delay @xmath11 ( sec .
[ sec : lossless - source - coding ] ) .
achievability is based on a novel scheme that consists of simple multiplexing / demultiplexing operations along with standard entropy coding techniques ; * we derive a single - letter characterization of the minimal rate ( bits / source symbol ) required for lossy compression for the point - to - point model of fig .
[ fig0 ] and , more generally , for the model of fig .
[ fig2 ] in which the side information may be delayed , for delays @xmath27 and @xmath41 ( sec . [ sec : lossy - source - coding ] ) ; * we solve a number of specific examples , namely binary - alphabet sources with hamming distortion and gaussian sources with minimum mean square error distortion , and present related numerical results ( sec .
[ sec : examples ] ) .
we present the system model for the scenario of fig .
[ fig2 ] . as detailed below , the scenarios of fig .
[ fig0 ] is obtained as a special case .
the system is characterized by a delay @xmath11 ; finite alphabets @xmath42 , @xmath43 , @xmath44 , @xmath45 conditional probabilities @xmath46 , with @xmath47 and @xmath48 with @xmath49 and @xmath50 ( i.e. , we have @xmath51 and @xmath52 for all @xmath53 ) ; and distortion metrics @xmath54 : @xmath55 $ ] , such that @xmath56 for all @xmath57 for @xmath58 .
as explained below , the subscript `` 1 '' in @xmath46 indicates that @xmath46 denotes one - step transition probabilities . the random process @xmath59 , @xmath60 , is a stationary and ergodic markov chain with transition probability @xmath61=w_{1}(a|b).$ ]
we define the probability @xmath62\triangleq\pi(a)$ ] and also the @xmath63-step transition probability @xmath64\triangleq w_{k}(a|b),$ ] which are both independent of @xmath14 by the stationarity of @xmath0 .
these quantities can be calculated using standard markov chain theory from the transition matrix associated with @xmath46 ( see , e.g. , @xcite ) .
we also set , for notational convenience , @xmath65 .
sequence @xmath3 is thus distributed as @xmath66 for any integer @xmath67 the random process @xmath68
, @xmath60 is such that vector @xmath69 , for any integer @xmath70 , is jointly distributed with @xmath6 so that @xmath71 in other words , process @xmath68 , @xmath60 corresponds to a hidden markov model with underlying markov process given by @xmath40 we now define encoder and decoders for the setting of fig .
specifically , an @xmath72 code is defined by : ( _ i _ ) an encoder function @xmath73\times\lbrack1,2^{n\delta r}],\label{eq : encoder}\ ] ] which maps sequences @xmath5 and @xmath6 into messages @xmath74 $ ] and @xmath75;$ ] ( _ ii _ ) a sequence of decoding functions for decoder 1 @xmath76\times\mathcal{y}^{i - d}\rightarrow\mathcal{z}_{1},\label{eq : decoder}\ ] ] for @xmath77 $ ] , which , at each time @xmath78 map message @xmath79 or rate @xmath9 [ bits / source symbol ] , and the delayed side information @xmath17 into the estimate @xmath80 ; ( _ iii _ ) a sequence of decoding function for decoder 2 @xmath81\times[1,2^{n\delta r}]\times\mathcal{y}^{i}\rightarrow\mathcal{z}_{2}\label{eq : decoder2}\ ] ] for @xmath77 $ ] , which , at each time @xmath78 map messages @xmath79 or rate @xmath82 and @xmath83 of rate or rate @xmath84 and the non - delayed side information @xmath29 into the estimate @xmath85 . in ( [ eq : encoder])-([eq : decoder2 ] ) , for @xmath86 integer with @xmath87 , we have defined @xmath88 $ ] as the interval @xmath89 $ ] with @xmath88=\phi$ ] if @xmath90 . and @xmath91
are implicitly considered to be rounded up to the nearest larger integer . ] encoding / decoding functions ( [ eq : encoder])-([eq : decoder2 ] ) must satisfy the distortion constraints @xmath92\leq d_{j},\text { for } j=1,2.\label{dist constraints}\ ] ] note that these constraints are fairly general in that they allow to impose not only requirements on the lossy reconstruction of @xmath4 or @xmath0 ( obtained by setting @xmath93 independent of @xmath94 or @xmath95 respectively ) , but also on some function of both @xmath4 and @xmath0 ( by setting @xmath93 to be dependent on such function of ( @xmath96 ) ) . given a delay @xmath11 , for a distortion pair ( @xmath97 )
, we say that rate pair ( @xmath98 ) is achievable if , for every @xmath99 and sufficiently large @xmath100 , there exists a @xmath101 code .
we refer to the closure of the set of all achievable rates for a given distortion pair ( @xmath97 ) and delay @xmath16 as the _ rate - distortion region _ @xmath102 . from the general description above for the setting of fig .
[ fig2 ] , the special case of fig .
[ fig0 ] is produced by neglecting the presence of decoder 2 , or equivalently by choosing @xmath103 . in this case , the rate - distortion region @xmath102 is fully characterized by a function @xmath104 as @xmath105@xmath106 .
function @xmath104 hence characterizes the infimum of rates @xmath9 for which the pair @xmath107 is achievable , and is referred to as the _ rate - distortion function _ for the setting of fig .
[ fig0 ] . for the special case of the model in fig .
[ fig2 ] in which @xmath32 , we define the rate - distortion function @xmath108 in a similar way . _ notation _ : for @xmath86 integer with @xmath109 , we define @xmath110 ; if instead @xmath109 we set @xmath111 .
we will also write @xmath112 for @xmath113 for simplicity of notation . given a sequence
@xmath114 $ ] and a set @xmath115,$ ] we define sequence @xmath116 as @xmath117 $ ] where @xmath118 .
random variables are denoted with capital letters and corresponding values with lowercase letters .
given random variables , or more generally vectors , @xmath119 and @xmath120 we will use the notation @xmath121 or @xmath122 for @xmath123 $ ] , and @xmath124 or @xmath125 for @xmath126 $ ] , where the latter notations are used when the meaning is clear from the context .
given set @xmath42 , we define @xmath127 as the @xmath100-fold cartesian product of @xmath42 .
we denote any function of @xmath99 that tends to zero as @xmath128 as @xmath129 . when referring to @xmath130typical sequences , we refer to the notion of strong typicality as treated in @xcite .
in this section , we study the point - to - point model in fig .
[ fig0 ] .
we start by characterizing the rate - distortion function @xmath104 for any delay @xmath11 under the hamming distortion metric for @xmath131 .
the hamming distortion metric is defined as @xmath132 , where @xmath133 if @xmath134 is true and @xmath135 otherwise .
this implies that the distortion constraint ( [ dist constraints ] ) for @xmath136 becomes @xmath137=\frac{1}{n}\sum\limits _ { i=1}^{n}\pr[x_{i}\neq z_{1i}]=0.\label{dist constraints-1}\ ] ] in other words , from the definition of achievability given above , we impose that the sequence @xmath5 be recovered with vanishingly small average symbol error probability as @xmath138 .
we refer to this scenario as asymptotically lossless , or lossless for short .
we have the following characterization of @xmath139 . [
prop : for - any - delay 1]for any delay @xmath11 , the rate - distortion function for the set - up in fig .
[ fig0 ] under hamming distortion at @xmath131 is given by @xmath140 where the conditional entropy is calculated with respect to the distribution @xmath141 } } \prod\limits _ { i=2}^{d+1}w_{1}(y_{i}|y_{i-1})q(x_{i}|y_{i})\text { for } d\geq1.\label{eq : distributions}\end{aligned}\ ] ] the proof of converse of the proposition above is based on an appropriate use of the fano inequality and is reported in appendix [ sub : proof - of - converse ] .
to prove the direct part of the proposition , we propose a simple achievable scheme , which , to the best of the authors knowledge , has not appeared before , in sec . [
sub : proof - of - achievability ] .
expression ( [ lossless ] ) consists of a conditional entropy of @xmath142 random variables , namely @xmath143,@xmath144 , ... , @xmath145 .
these variables are distributed as the corresponding entries in the random vectors @xmath5 and @xmath6 , as per ( [ eq : distributions - a])-([eq : distributions ] ) ( cf .
( [ eq : hidden markov ] ) ) .
we have therefore used the same notation for the involved random variables as in sec . [
sec : system - model ] .
proposition 1 provides a `` single - letter '' characterization of @xmath139 for the setting of fig .
[ fig0 ] , since it only involves a finite number of variables .
this contrasts with the general characterization for stationary ergodic processes of @xmath146 given in @xcite , which is a `` multi - letter '' expression , whose computation can generally only attempted numerically using approaches such as the ones proposed in @xcite .
note that a multi - letter expression is also given in @xcite to characterize @xmath146 for i.i.d .
sources with _ negative _ delays @xmath147 .
finally , it should be emphasized that the simple characterization ( [ lossless ] ) for the scenario of interest here hinges on the assumed statistics of the sources ( @xmath148 ) . by setting @xmath27 in ( [ lossless ] )
we obtain @xmath149 .
this result generalizes ( * ? ? ?
* ; * ? ? ?
* remark 3 , p. 5227 ) from i.i.d .
sources ( @xmath148 ) to the hidden markov model ( [ eq : hidden markov ] ) considered here . note that , for @xmath41 , we instead obtain @xmath150 as another notable special case , if side information is absent , or equivalently @xmath151 , in accordance to well - known results , we obtain that @xmath152 equals the entropy rate ( see , e.g. , @xcite ) @xmath153 in fact , we have @xmath154 by ( * ? ? ?
* theorem 4.5.1 ) .
[ rem : is - delayed - side]is _ delayed _ side information useful ( when known also at the encoder ) ? that this is generally the case follows from the inequality @xmath155 since @xmath152 is the required rate without side information . this result is proved by the chain of inequalities @xmath156 where the first inequality follows by the data processing inequality and the second by conditioning reduces entropy
. however , inequality ( [ eq : entropy rate in ] ) may not be strict , and thus side information may not be useful .
a first example is the case where @xmath4 is an i.i.d .
process , which is obtained by making @xmath38 independent of @xmath94 . as another example ,
consider the setting of source coding with feedforward @xcite , i.e. , @xmath35 . in this case
, our assumption ( [ eq : hidden markov ] ) entails that @xmath5 is a markov chain , and we have @xmath157 for @xmath158 . therefore ,
delayed feedforward ( with @xmath159 is not useful for the lossless compression of markov chains , as already shown in @xcite .
this conclusion need not hold for lossy compression ( i.e. , for @xmath160 ) @xcite ( see also sec . [
sub : binary - hidden - markov ] ) . if @xmath148 are general jointly stationary and ergodic processes ( and not necessarily stationary ergodic hidden markov models )
, one can adapt in a straightforward way the proofs of appendix [ sub : proof - of - converse ] and sec .
[ sub : proof - of - achievability ] , and conclude that the rate distortion function can be written as @xmath161 where @xmath162 is the causally conditioned entropy @xmath163 ( see , e.g. , @xcite ) . comparing ( [ eq : directed info ] ) with the rate @xmath164 necessary in the absence of any side information , we conclude that the reduction in the compression rate obtained by leveraging delayed side information at the decoder , when side information is known at the encoder , is given for stationary and ergodic processes by @xmath165 in ( [ eq : directed mut info ] ) , we have used the definition of _ directed _ mutual information @xmath166 ( see , e.g. , @xcite ) .
note that the rate gain ( [ eq : directed mut info ] ) complements the results given in @xcite on the interpretation of the directed mutual information ( see also next remark ) .
[ rem : consider - a - variable - length]consider a _ variable - length _ ( strictly ) lossless source code that operates symbol by symbol such that , for every symbol @xmath167 $ ] , it outputs a string of bits @xmath168 which is a function of @xmath169 and @xmath17 .
encoding is constrained so that the code @xmath170 for each ( @xmath171 ) is prefix - free .
the decoder , based on delayed side information , can then uniquely decode each codeword @xmath170 as soon as it is received .
following the considerations in ( * ? ? ?
* ; * ? ? ?
iv ) , it is easy to verify that rate @xmath139 ( and , more generally , ( [ eq : directed info ] ) ) is also the infimum of the average rate in bits / source symbol required by such code .
moreover , it is possible to construct universal context - based compression strategies by adapting the approach in @xcite .
we refer to sec .
[ sec : examples ] for some examples that further illustrate some implications of proposition 1 .
, for @xmath172 ( symbols corresponding to out - of - range indices are set to zero).,width=345 ] ( achievability ) here we propose a coding scheme that achieves rate ( [ lossless ] ) .
the basic idea is a non - trivial extension of the approach discussed in ( * ? ? ?
* ; * ? ? ?
* remark 3 , p. 5227 ) and is described as follows . a block diagram is shown in fig .
[ figmux ] for encoder ( fig .
[ figmux]-(a ) ) and decoder ( fig .
[ figmux]-(b ) ) .
we first describe the _ encoder , _ which is illustrated in fig .
[ figmux]-(a ) . to encode sequences
@xmath173 we first partition the interval @xmath174 $ ] into @xmath175 subintervals , which we denote as @xmath176 $ ] , for all @xmath177 and @xmath178 .
every such subinterval @xmath179 is defined as @xmath180\text { and } y_{i - d}=\tilde{y},\text { } x_{i - d+1}^{i-1}=\tilde{x}^{d-1}\}.\label{eq : i(xy)}\ ] ] in words , the subinterval @xmath179 contains all symbol indices @xmath14 such that the corresponding delayed side information available at the decoder is @xmath181 and the previous @xmath182 samples in @xmath183 are @xmath184 .
we refer to the value of the tuple ( @xmath185 ) as the _ context _ of sample @xmath186 .
, this definition of context is consistent with the conventional one given in @xcite when specialized to markov processes .
see also remark [ rem : consider - a - variable - length ] . ] for the out - of - range indices @xmath187 $ ] , one can assume arbitrary values for @xmath188 and @xmath189 , which are also shared with the decoder once and for all . note that @xmath190 $ ] .
[ figillustration ] illustrates the definitions at hand for @xmath172 . as a result of the partition described above , the encoder `` demultiplexes '' sequence @xmath183 into @xmath175 sequences @xmath191 , one for each possible context ( @xmath192 .
this demultiplexing operation , which is controlled by the previous values of source and side information , is performed in fig .
[ figmux]-(a ) by the block labelled as `` demux '' , and an example of its operation is shown in fig .
[ figillustration ] . by the ergodicity of process
@xmath4 and @xmath0 , for every @xmath99 and all sufficiently large @xmath100 , the length of any sequence @xmath191 is guaranteed to be less than @xmath193 symbols with probability arbitrarily close to one .
this because the length @xmath194 of the sequence @xmath191 equals the number of occurrences of the context ( @xmath195 ) and by birkhoff s ergodic theorem ( see ( * ? ? ?
16.8 ) ) . in particular
, for any @xmath99 we can find an @xmath100 such that @xmath196\leq\frac{\epsilon}{2|\mathcal{x}|^{d-1}|\mathcal{y}|},\label{eq : error_1}\ ] ] where we have defined the `` error '' event @xmath197 each sequence @xmath191 is encoded by a separate encoder , labelled as `` enc '' in fig . [
figmux]-(a ) . in case
the cardinality @xmath194 does not exceed @xmath193 ( i.e. , the `` error '' event @xmath198 does not occur ) , the encoder compresses sequence @xmath191 using an entropy encoder , as explained below . if the cardinality condition is instead not satisfied ( i.e. , @xmath198 is realized ) , then an arbitrary bit sequence of length @xmath199 , to be specified below , is selected by the encoder `` enc '' .
the entropy encoder can be implemented in different ways , e.g. , using typicality or huffman coding ( see , e.g. , @xcite ) . here
we consider a typicality - based encoder .
note that the entries @xmath4 of each sequence @xmath200 are i.i.d . with distribution @xmath201 , since conditioning on the context @xmath202 makes the random variables @xmath4 independent . as it is standard practice , the entropy encoder assigns a distinct label to all @xmath203-typical sequences @xmath204 with respect to such distribution , and an arbitrary label to non - typical sequences . from the asymptotic equipartion property ( aep )
, we can choose @xmath100 sufficiently large so that ( see , e.g. , @xcite ) @xmath205\leq\frac{\epsilon}{2|\mathcal{x}|^{d-1}|\mathcal{y}|},\label{eq : error_2}\ ] ] where we have defined the `` error '' event @xmath206 moreover , by the aep , a rate in bits per source symbol of @xmath207 is sufficient for the entropy encoder to label all @xmath203-typical sequences . from the discussion above
, it follows that the proposed scheme encodes each sequence @xmath191 with @xmath208 bits . by concatenating the descriptions of all the @xmath175 sequences
@xmath209 , we thus obtain that the overall rate @xmath9 of message @xmath7 for the scheme at hand is @xmath210 .
the concatenation of the labels output by each entropy encoder is represented in fig .
[ figmux]-(a ) by the block `` mux '' .
we emphasize that encoder and decoder agree a priori on the order in which the descriptions of the different subsequences are concatenated .
for instance , with reference to the example in fig .
[ figillustration ] ( with @xmath172 ) , message @xmath7 can contain first the description of the sequence corresponding to @xmath211 , then @xmath212 , etc .
we now describe the _ decoder , _ which is illustrated in fig . [ figmux]-(b ) . by undoing the multiplexing operation just described
, the decoder , from the message @xmath7 , can recover the individual sequences @xmath191 through a simple demultiplexing operation for all contexts @xmath213 .
this operation is represented by block `` demux '' in fig . [ figmux]-(b ) . to be precise , this demultiplexing is possible , unless the encoding `` error '' event @xmath214 takes place .
in fact , occurrence of the `` error '' event @xmath215 implies that some of the sequences @xmath191 was not correctly encoded and hence can not be recovered at the decoder .
the effect of such errors will be accounted for below .
assume now that no error has taken place in the encoding . while the individual sequences @xmath191 can be recovered through the discussed demultiplexing operation
, this does not imply that the decoder is also able to recover the original sequence @xmath183 .
in fact , that decoder does not know a priori the partition @xmath216 : @xmath177 and @xmath217 of the interval @xmath174 $ ] and thus can not reorder the elements of sequences @xmath191 to produce @xmath183 .
recall , moreover , that such re - ordering operation should be done in a causal fashion following the decoding rule ( [ eq : decoder ] ) .
we now argue that the re - ordering mentioned above is in fact possible using a decoding rule that complies with ( [ eq : decoder ] ) via a multiplexing block controlled by the previous estimates of the source samples ( block `` mux '' in fig .
[ figmux]-(b ) ) .
in fact , note that at time @xmath14 , the decoder knows @xmath218 and the previously decoded @xmath219 and can thus identify the subinterval @xmath179 to which the current symbol @xmath4 belongs .
this symbol can be then immediately read as the next yet - to - be - read symbol from the corresponding sequence @xmath191 .
note that for the first @xmath16 symbols , the decoder uses the values for @xmath186 and @xmath220 at the out - of - range indices @xmath14 that were agreed upon with the encoder ( see above ) . in conclusion , we remark that the scheme described above , by choosing @xmath203 small enough and @xmath100 large enough , is able to satisfy the constraint ( [ dist constraints-1 ] ) to any desired accuracy .
we also note that the controlled multiplexing / demultiplexing operation used in the proof is reminiscent of the scheme proposed in @xcite for transmission on fading channels with side information at the transmitter and receiver .
we finally need to study the effect of errors . given the choices made above , we have that the probability of an encoding error is @xmath221 & \leq\sum_{\tilde{x}^{d-1}\in\mathcal{x}^{d-1},\mbox { } \tilde{y}\in\mathcal{y}}\pr[\mathcal{e}_{1}(\tilde{y},\tilde{x}^{d-1})]+\pr[\mathcal{e}_{2}(\tilde{y},\tilde{x}^{d-1})]\leq\epsilon,\end{aligned}\ ] ] where the first inequality follows from the union bound and the second from ( [ eq : error_1 ] ) and ( [ eq : error_2 ] ) .
this implies that the distortion in ( [ dist constraints-1 ] ) is upper bounded by @xmath203 as desired .
in fact , from the definition of encoder and decoder given above , we can conclude that @xmath222=\pr[\mathcal{e}]\leq\epsilon$ ] , where we recall that @xmath223 is the sequence reconstructed at the decoder .
moreover , the following inequality holds in general @xmath224\geq\frac{1}{n}\sum\limits _
{ i=1}^{n}\pr[x_{i}\neq z_{1i}].\label{eq : ineqperr}\ ] ] therefore ,
we have @xmath225\leq\epsilon$ ] , which concludes the proof . an alternative proof of achievability can be given by using the idea of codetrees and extending the notions of typicality introduced in @xcite .
the proof discussed above is based on a conceptually and algorithmically simpler approach , albeit its applicability is limited to lossless compression ( see next subsection ) . from the inequality ( [ eq : ineqperr ] ) , it follows that the optimality of the scheme above can be proved also under the more stringent block error probability constraint ( see also ( * ? ? ?
* ; * ? ? ?
3.6.4 ) ) . here
, we obtain a characterization of the rate - distortion function @xmath104 , for @xmath27 and @xmath41 .
the proof follows as a special case of that of proposition [ pro : for - any - delay ] to be discussed in the next section , and is based on similar arguments as for proposition [ prop : for - any - delay 1 ] .
[ cor : for - any - delay]for any delay @xmath11 and distortion @xmath226 , the following rate is achievable for the setting of fig .
[ fig0 ] @xmath227 with mutual informations evaluated with respect to the joint distribution @xmath228 and where minimization is done over all conditional distributions @xmath229 such that @xmath230\leq d_{1}.\label{eq : dist const r1d1d2 - 1 - 1}\ ] ] moreover , rate ( [ eq : rda])-([eq : dist const r1d1d2 - 1 - 1 ] ) is the rate - distortion function , i.e. , , for @xmath27 and @xmath41 . the optimality of the conditional codebook strategy for lossless compression shown in proposition [ prop : for - any - delay 1 ] hinges on the following fact : conditioned on the context ( _ @xmath231 _ @xmath232 ) , the samples _
@xmath4 _ are independent of the past samples @xmath219 by the hidden markov model assumption .
recall that the fact that the decoder has available the past source samples ( @xmath233@xmath232 ) since its estimates are correct with high probability . due to this independence property , and to the availability of the side information also at the encoder
, the latter need not use multi - letter compression codes and can instead use simple single - letter entropy codes conditioned on the values of ( _ _ @xmath231__@xmath232 ) without loss of optimality . in the lossy case considered in proposition [ cor : for - any - delay ] , instead , even for the point - to - point model , the independence condition discussed above does not hold for delays @xmath16 strictly larger than 1 .
in fact , at each time @xmath14 , the decoder has available the delayed side information @xmath17 only , conditioned on which the source samples @xmath4 are not independent of the past samples @xmath219 .
but , for @xmath41 , the independence condition at hand does apply and thus the optimality of single - letter codes can be proved as done in proposition [ cor : for - any - delay ] .
in this section , we consider the problem of lossy compression for the set - up of fig . [ fig2 ] .
note that the asymptotically lossless case follows from proposition [ prop : for - any - delay 1 ] , since , in order to guarantee lossless reconstruction also at the decoder with delayed side information , rate @xmath9 must satisfy the conditions in proposition [ prop : for - any - delay 1 ] . here , we obtain an achievable rate region @xmath234 for all delays @xmath11 for the model in fig .
[ fig2 ] , and show that such region coincides with the rate - distortion region , i.e. , @xmath235 , for @xmath27 and @xmath41 . to streamline the discussion ,
we start by consider the special case where @xmath32 and obtain a characterization of the rate - distortion function @xmath108 for @xmath27 and @xmath41 . [ pro:2]for any delay @xmath11 and distortion pair @xmath236
, the following rate is achievable for the set - up of fig .
[ fig2 ] with @xmath32 @xmath237 with mutual informations evaluated with respect to the joint distribution @xmath238 and where minimization is done over all conditional distributions @xmath239 such that @xmath240\leq d_{j}\text { , for } j=1,2.\label{eq : dist const r1d1d2 - 1}\ ] ] moreover , rate ( [ eq : r1d1d2a])-([eq : r1d1d2 - 1 ] ) is the rate - distortion function , i.e. , , for @xmath27 and @xmath41 .
[ rem : rate inter-1]rate ( [ eq : r1d1d2a ] ) can be easily interpreted in terms of achievability . to this end
, we remark that variable @xmath241 plays the role of the delayed side information @xmath17 at decoder 1 .
the coding scheme achieving rate ( [ eq : r1d1d2a ] ) operates in two successive phases . in the first phase ,
the encoder encodes the reconstruction sequence @xmath223 for decoder 1 . since decoder 1 has available delayed side information , using a strategy similar to the one discussed in sec .
[ sub : proof - of - achievability ] , this operation requires @xmath242 bits per source sample , as further detailed in sec .
[ sub : proof - of - achievability-2 ] .
note that decoder 2 is able to recover @xmath223 as well , since decoder 2 has available side information @xmath29 , and thus also the delayed side information @xmath17 . in the second phase
, the reconstruction sequence @xmath243 for decoder 2 is encoded .
given the side information available at decoder 2 , this operation requires rate @xmath244 , using again an approach similar to the one discussed in sec .
[ sub : proof - of - achievability ] .
the converse proof is in appendix [ sub : proof - of - converse-2 ] .
[ rem : for - memoryless - sources-1]for memoryless sources @xmath5 and @xmath6 , obtained by setting the transition probability @xmath245 to be independent of @xmath246 , it can be seen that the achievable rate ( [ eq : r1d1d2a])-([eq : r1d1d2 - 1 ] ) is the rate - distortion function for the scenario of fig .
[ fig2 ] with @xmath32 _ for all delays _ @xmath11 .
this observation extends lemma 1 to the more general set - up of fig .
[ fig2 ] with @xmath32 . to see this , note that for @xmath158 , rate ( [ eq : r1d1d2a])-([eq : r1d1d2 - 1 ] ) is given by @xmath247 with mutual informations evaluated with respect to the joint distribution @xmath248 and where minimization is done over all conditional distributions @xmath239 such that the distortion constraints ( [ eq : dist const r1d1d2 - 1 ] ) are satisfied .
rate ( [ eq : rem memoryless-1 ] ) recovers the rate - distortion function derived by @xcite for the case where decoder 1 has _ no _ side information .
therefore , rate ( [ eq : rem memoryless-1 ] ) is achievable even without any state information at decoder 1 .
we then conclude that _ _ delayed side information is not useful for memoryless sources__. _ _ note also that @xcite assumes non - causal availability of the side information at decoder 2 .
the equality of the rate derived in @xcite and the one in proposition [ pro:2 ] thus demonstrates that causal and non - causal side information lead to the same performance in terms of rate - distortion function .
while ( [ eq : r1d1d2a ] ) is easier to interpret in terms of achievability as done in remark [ rem : rate inter-1 ] , the equivalent expression ( [ eq : r1d1d2 - 1 ] ) highlights the rate loss due to the possible delay of the side information .
in fact , the mutual information @xmath249 accounts for the rate that would be needed to convey both @xmath223 and @xmath243 only to decoder 2 , which has non - delayed side information . therefore , the additional term @xmath250 can be interpreted as the extra rate that needs to be expended to enable transmission of @xmath223 also to decoder 1 , which has delayed side information .
we now consider the general model in fig .
[ fig2 ] .
[ pro : for - any - delay]for any delay @xmath11 and any distortion pair ( @xmath97 ) , define @xmath251 as the union of all rate pairs ( @xmath98 ) that satisfy @xmath252 for some joint distribution @xmath253 where minimization is done over all conditional distributions @xmath254 such that @xmath240\leq d_{j}\text { , for } j=1,2.\ ] ] we have that @xmath255 for any @xmath11
. moreover , equation ( [ eq : ach region fig2 ] ) holds with equality , and thus @xmath251 is the rate - distortion region , for @xmath27 and @xmath41 .
let us interpret the rate region @xmath251 in terms of achievability .
first , from remark [ rem : rate inter-1 ] , we observe that ( [ eq : ineq fig1 ] ) is the rate necessary to convey @xmath223 to both decoder 1 and decoder 2 , and an auxiliary codeword @xmath256 only to decoder 2 .
this auxiliary codeword @xmath256 carries information to decoder 2 that is then refined via message @xmath257 in particular , rewriting ( [ eq : ineq fig2 ] ) as @xmath258 , by comparison with ( [ eq : ineq fig1 ] ) , we see that the extra rate @xmath259 is needed to transmit sequence @xmath243 to decoder 2 , thus refining the information available therein due to message @xmath7 . and @xmath20 , which leads to the sum - rate constraint ( [ eq : ineq fig2 ] ) .
] the considerations in remark [ rem : for - memoryless - sources-1 ] can be also easily extended to the scenario of proposition [ pro : for - any - delay ] with @xmath33 .
( achievability ) we first prove achievability of rate ( [ eq : r1d1d2a ] ) in proposition [ pro:2 ] .
the proof extends the ideas discussed in sec .
[ sub : proof - of - achievability ] , to which we refer for details .
in particular , here we do not detail the calculations of the encoding `` error '' events and distortion levels , as they follow in the same way as in sec .
[ sub : proof - of - achievability ] . to encode sequence ( @xmath260 ) , the encoder partitions the interval @xmath174 $ ] into @xmath261 subintervals , namely @xmath262 for each @xmath178 , so that ( cf .
( [ eq : i(xy ) ] ) ) @xmath263\text { and } y_{i - d}=\tilde{y}\}.\label{eq : i(xy)-1}\ ] ] similar to sec .
[ sub : proof - of - achievability ] , a different compression codebook is used for each such interval @xmath262 , and thus for each pair of `` demultiplexed '' subsequences @xmath264 . the compression of each pair of sequences @xmath264 is based on a test channel @xmath265 specifically , the corresponding codewords @xmath223 are generated i.i.d . according to the marginal distribution @xmath266 @xmath267 and compression
is done based on standard joint typicality arguments . by the covering lemma @xcite , compression of sequences @xmath268 into the corresponding reconstruction sequence @xmath269
requires rate @xmath270 bits per source symbol in each interval @xmath262 , and thus an overall rate @xmath271 following the same considerations as in sec .
[ sub : proof - of - achievability ] .
in particular , the encoder multiplexes the compression indices corresponding to the @xmath261 intervals @xmath262 to produce message @xmath7 .
therefore , the latter only carries information about the individual sequences @xmath272 but not about the ordering of each entry within the overall sequence @xmath223 . based on the sequence @xmath273 produced in the first encoding phase described above , the encoder then performs also a finer partition of the interval @xmath174 $ ] into @xmath274 intervals @xmath275 with @xmath276 , so that @xmath277\text { and } y_{i - d}=\tilde{y},\mbox { } y_{i}=y,\mbox{\mbox { } and } z_{i}=z\}.\label{eq : i(xy)-1 - 1}\ ] ] compression of sequence @xmath278 into the corresponding reconstruction @xmath279 is carried out according to test channel @xmath280 as per the discussion above , requiring an overall rate of @xmath281 the compression indices for all sets @xmath282 are concatenated in message @xmath7 following the compression indices obtained from the sets @xmath262 . upon reception of message @xmath7 , decoder 1 and 2
can both recover the sequences @xmath269 and @xmath279 for all @xmath283 @xmath50 and @xmath284 via simple demultiplexing .
moreover , following the same reasoning as in sec .
[ sub : proof - of - achievability ] , decoder 1 can reconstruct sequence @xmath223 in the correct order in a causal fashion , using a decoder ( [ eq : decoder ] ) , which depends on message and delayed side information , since the value of @xmath80 can be obtained from sequences @xmath269 by knowing the value of @xmath285 similarly , decoder 2 can reorder sequence @xmath243 in a causal fashion using a decoder of the form ( [ eq : decoder2 ] ) .
this concludes the proof of achievability for proposition [ pro:2 ] .
we now turning to the proof of achievability proposition [ pro : for - any - delay ] .
for a fixed distribution ( [ sr : pmf ] ) , we need to prove that the rate region in fig .
[ figregion ] is achievable . to do this
, it is enough , by standard time - sharing arguments , to prove that corner points a and b are achievable .
corner point b corresponds to rate pair @xmath286 and @xmath32 .
but achievability of this region follows immediately from proposition [ pro:2 ] by setting @xmath287 in ( [ eq : r1d1d2a ] ) .
instead , corner point a corresponds to the rate pair @xmath288 this rate pair can be achieved by using a strategy similar to the one discussed above . in this strategy , when encoding the message @xmath20 , which is received only at decoder 2 , the encoder leverages the fact that the latter knows @xmath289 and @xmath80 , by appropriately partitioning the interval @xmath174 $ ] and using different test channels in each subinterval .
.,width=345 ] in this section , we consider two specific examples relative to the scenario in fig . [ fig0 ] .
the first example consists of binary - alphabet sources , while the second applies the results derived above to ( continuous - alphabet ) gaussian sources .
we focus on a distortion metric of the form @xmath290 that does not depend on @xmath291 in other words , the decoder is interested in reconstructing @xmath5 within some distortion @xmath226 .
we note that , under this assumption , the rate ( [ cor : for - any - delay ] ) equals the simpler expression @xmath292 with mutual informations evaluated with respect to the joint distribution @xmath293 where minimization is done over all distributions @xmath294 such that @xmath295\leq d_{1}.$ ] note that this simplification is without loss of optimality because the distortion constraint does not depend on the correlation between @xmath296 and @xmath297 therefore , we can impose the markov condition @xmath298 as in ( [ eq : simpler ] ) without changing the distortion , while reducing the mutual information in ( [ eq : rda ] ) . in the first example , we assume that @xmath0 is a binary markov chain with symmetric transition probabilities @xmath299 .
therefore , we have @xmath300 and @xmath63-step transition probabilities @xmath301 , which can be obtained recursively as @xmath302 and @xmath303 for @xmath304.=\left[\begin{array}{cc } 1-\varepsilon & \varepsilon\\ \varepsilon & 1-\varepsilon \end{array}\right]^{k}$ ] , well known from markov chain theory ( see , e.g. , @xcite ) . ]
note that this is a logistic map such that @xmath305 for large @xmath63 .
we also set @xmath306 , consistently with the convention adopted in the rest of the paper .
finally , we assume that @xmath307 with `` @xmath308 '' being the modulo-2 sum and @xmath309 being i.i.d .
binary variables , independent of @xmath6 , with @xmath310 , @xmath311 .
we adopt the hamming distortion @xmath312 .
we start by showing in fig .
[ fig4 ] the rate @xmath139 obtained from proposition 1 corresponding to zero distortion ( @xmath131 ) versus the delay @xmath16 for different values of @xmath313 and for @xmath314 .
note that the value of @xmath313 measure the `` memory '' of the process @xmath0 : for @xmath313 small , the process tends to keep its current value , while for @xmath315 , the values of @xmath0 are i.i.d .. for @xmath27 , we have @xmath316 , irrespective of the value of @xmath313 , where we have defined the binary entropy function @xmath317 .
instead , for @xmath16 increasingly large , the rate @xmath139 tends to the entropy rate @xmath164 .
this can be calculated numerically to arbitrary precision following ( * ? ? ?
* sec . 4.5 ) . note that a larger memory , i.e. , a smaller @xmath313 leads to smaller required rate @xmath139 for all values of @xmath16 .
[ fig5 ] shows the rate @xmath139 for @xmath318 versus @xmath319 for different values of @xmath16 . for reference
, we also show the performance with no side information , i.e. , @xmath164 . for @xmath320 ,
the source @xmath5 is i.i.d . and delayed side information is useless in the sense that @xmath321 ( remark [ rem : is - delayed - side ] ) .
moreover , for @xmath322 , we have @xmath35 , so that @xmath4 is a markov chain and the problem becomes one of lossless source coding with feedforward . from remark
[ rem : is - delayed - side ] , we know that delayed side information is useless also in this case , as @xmath323 . for intermediate values of @xmath319 ,
side information is generally useful , unless the delay @xmath16 is too large .
we now turn to the case where the distortion @xmath226 is generally non - zero . to this end , we evaluate the achievable rate ( [ eq : simpler ] ) in appendix [ sec : proof - of-()- ( ) ] obtaining @xmath324 for @xmath325 and @xmath326 otherwise .
in ( [ eq : ex1])-([eq : ex2 ] ) we have defined @xmath327 . recall that rate @xmath328 has been proved to coincide with the rate - distortion function @xmath104 only for @xmath329 ( corollary [ cor : for - any - delay ] ) . as a final remark ,
we use the result derived above to discuss the advantages of delayed side information . to this end ,
set @xmath322 so that @xmath35 and the problem becomes one of source coding with feedforward .
for @xmath41 , result ( [ eq : ex1])-([eq : ex2 ] ) recovers the calculation in ( * ? ? ?
* example 2 ) ( see also @xcite ) , which states that the rate - distortion function for the markov source @xmath5 at hand with feedforward ( @xmath41 ) is @xmath330 for @xmath331 and @xmath332 otherwise . from @xcite
( see also @xcite ) , it is known that the rate - distortion function of a markov source @xmath5 without feedforward , i.e. , @xmath333 , is equal to ( [ slb ] ) only for @xmath226 smaller than a critical value , but is otherwise larger .
this demonstrates that feedforward , unlike in the lossless setting discussed above , can be useful in the lossy case for distortion levels @xmath226 sufficiently large , as first discussed in @xcite . for lossless reconstruction for the set - up of fig .
[ fig0 ] with binary sources versus delay @xmath16 ( @xmath314).,width=432 ] for lossless reconstruction for the set - up of fig .
[ fig0 ] with binary sources versus parameter @xmath319 ( @xmath318).,width=432 ] we now assume that @xmath6 is a gauss - markov process with zero - mean , power @xmath334=1 $ ] and correlation @xmath335=\rho$ ] ( so that @xmath336=\rho^{d}$ ] ) .
moreover , @xmath4 is related to @xmath0 as @xmath337 where samples @xmath309 are i.i.d .
zero - mean gaussian with variance @xmath338 and independent of @xmath6 .
we concentrate on the mean square error distortion metric @xmath339 . using standard arguments
, we can apply the achievable rate ( [ eq : simpler ] ) to the setting at hand , although the result was derived for discrete alphabet ( see ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ch . 3.8 ) ) . by doing so , as shown in appendix [ sec : proof - of- ( ) ] , we get that the following rate is achievable for @xmath11 @xmath340 if @xmath341 and @xmath326 otherwise . as also discussed above , this rate coincides with the rate - distortion function for @xmath27 and @xmath41 .
similar to the discussion in the previous section for a binary hidden markov model , we remark that for @xmath342 , the problem becomes one of lossy source coding with feedforward of a gauss - markov process @xmath5 . in this case
, it is known that the rate - distortion function without feedforward , @xmath333 , equals @xmath343 only for distortions @xmath226 smaller than a critical value @xcite and is otherwise larger . by comparison with ( [ eq : gauss r ] )
, it then follows that feedforward , for sufficiently large distortion levels , can be useful in decreasing the rate - distortion function .
the problem of compressing information sources in the presence of delayed side information finds application in a number of scenarios including sensor networks and prediction / denoising . a general information - theoretic characterization of the trade - off between rate and distortion for this problem can be generally given in terms of multi - letter expressions , as done in @xcite .
such expressions are proved by resorting to complex achievability schemes that operate in increasingly large blocks , and generally require involved numerical evaluations . in this work ,
we have instead focused on a specific class of sources , which evolve according to hidden markov models , and derived single - letter characterizations of the rate - distortion trade - off .
such characterizations are established based on simple achievable scheme that are based on standard `` off - the - shelf '' compression techniques . moreover ,
the analysis has focused not only for the conventional point - to - point setting of @xcite , but also on a more general set - up in which side information may or may not be delayed . the value of the derived characterization is demonstrated by elaborating on two examples , namely binary sources with hamming distortion and gaussian sources with minimum mean square error distortion .
various extensions of the results presented here are possible .
for instance , the optimal strategy for a cascade model with three nodes in which the intermediate node has causal side information @xmath29 and the end decoder has delayed side information @xmath344 can be identified by applying the result in proposition [ pro:2 ] in a manner similar to @xcite .
the authors wish to thank associate editor and reviewers for their thoughtful comments that have helped us improve the quality of the paper .
for @xmath99 , fix a code @xmath345 as defined in sec . [
sec : system - model ] . using the definition of encoder ( [ eq : encoder ] ) , we have the equalities @xmath346 the first term in ( [ eq : app11 ] ) cam be written , using the chain rule for entropy , as @xmath347\nonumber \\ & + \sum_{i = n - d+1}^{n}h(y_{i}|y^{i-1}x^{n})\nonumber \\ & = a+\sum_{i = d+1}^{n}\left[h(y_{i - d}|y^{i - d-1}x^{i-1})+h(x_{i}|y_{i - d}x_{i - d+1}^{i-1})\right]\label{eq : app12}\end{aligned}\ ] ] where @xmath348 is a finite constant that does not increase with @xmath349 moreover , in the last line we have used the markov chain @xmath350 , which follows from ( [ eq : hidden markov ] )
. the second term in ( [ eq : app11 ] ) can be similarly written as @xmath351\nonumber \\ & \leq b+\sum_{i = d+1}^{n}\left[h(y_{i - d}|y^{i - d-1}x^{i-1})+h(x_{i}|y^{i - d}m)\right],\label{eq : app13}\end{aligned}\ ] ] where @xmath352 is a finite constant that does not increase with @xmath349 the inequality in ( [ eq : app13 ] ) follows from conditioning reduces entropy .
note also that we have the inequality @xmath353 by conditioning reduces entropy . by definition ,
a code @xmath345 must satisfy ( cf .
( [ dist constraints-1 ] ) ) @xmath354 where we have defined @xmath355 $ ] .
it follows that @xmath356 the first inequality ( [ eq : app13a ] ) follows from the fact that @xmath80 is a function of @xmath357 by ( [ eq : decoder ] ) and by conditioning reduces entropy ; the second inequality ( [ eq : app14 ] ) follows from fano s inequality and the third from ( [ eq : app13b ] ) .
finally , from ( [ eq : app11]),([eq : app12]),([eq : app13]),([eq : app16 ] ) we obtain @xmath358\\ & -b-\sum_{i = d+1}^{n}\left[h(y_{i - d}|y^{i - d-1}x^{i-1})+n\delta(\epsilon)\right]\\ & = a - b+\sum_{i = d+1}^{n}h(x_{i}|y_{i - d}x_{i - d+1}^{i-1})+n\delta(\epsilon),\end{aligned}\ ] ] which concludes the proof .
we prove the converse for proposition [ pro : for - any - delay ] , since proposition [ pro:2 ] follows as a special case .
we focus on @xmath41 , since the proof for @xmath27 can be obtained in a similar fashion . to this end , fix a code @xmath359 as defined in sec .
[ sec : system - model ] .
using the definition of encoder ( [ eq : encoder ] ) and decoder ( [ eq : decoder ] ) we have @xmath360 where we have defined @xmath361 $ ] .
all equalities above follow from standard properties of the entropy and mutual information , while the inequality ( [ eq : app31 ] ) follows by conditioning reduces entropy . following the similar steps ,
we obtain @xmath362 the proof is concluded by introducing a time - sharing variable @xmath363 uniformly distributed in @xmath174 $ ] and defining random variables @xmath364 @xmath365 , @xmath366 , @xmath367 and @xmath368 , and by leveraging the convexity of the mutual informations in ( [ eq : app32 ] ) and ( [ eq : app4 ] ) with respect to the distribution @xmath369 .
here we prove that ( [ eq : ex1])-([eq : ex2 ] ) equals ( [ eq : simpler ] ) for the binary hidden markov model of sec .
[ sub : binary - hidden - markov ] .
first , for @xmath370 , we can simply set @xmath371 to obtain @xmath372 and @xmath373\leq d_{1}$ ] , which , from ( [ eq : ex1 ] ) and the non - negativity of mutual information , leads to @xmath326 .
similarly , for @xmath374 , we can set @xmath375 to prove that @xmath326 . for the remaining distortion levels @xmath376 , under the constraint that @xmath373\leq d_{1}$ ] , we have the following inequalities @xmath377 where the third line follows by conditioning decreases entropy and the last line from the fact that @xmath378 is increasing in @xmath379 for @xmath380 .
this lower bound can be achieved in ( [ eq : simpler ] ) by choosing the test channel @xmath294 so that @xmath119 can be written as @xmath381 where @xmath382 is binary with @xmath383 and independent of @xmath296 and @xmath241 , and @xmath296 is also independent of @xmath241 . to obtain @xmath384
, we need to impose that the joint distribution @xmath385 is preserved by the given choice of @xmath294 . to this end , note that the joint distribution @xmath385 is such that we can write @xmath386 , where @xmath387 is binary and independent of @xmath241 , with @xmath388 .
therefore , preservation of @xmath385 is guaranteed if the equality @xmath389=p_{z_{1}}(1)*d_{1}=\varepsilon^{(d)}*q$ ] holds .
this leads to @xmath390 we remark that @xmath391 , due to the inequality ( [ eq : ex2 ] ) on the distortion @xmath226 .
this concludes the proof .
here we prove that ( [ eq : gauss r ] ) equals ( [ eq : simpler ] ) for the hidden gauss - markov model of sec .
[ sub : gaussian - hidden - markov ] .
this follows by using analogous arguments as done above for the binary hidden markov model .
the only non - trivial adaptation of the proof given above is the choice of the test channel for the case where @xmath392 .
this must be selected so that @xmath119 can be written as @xmath393 where @xmath382 is zero - mean gaussian with @xmath394=d_{1}$ ] and independent of @xmath296 and @xmath241 , and @xmath296 is also zero - mean gaussian and independent of @xmath241 . to obtain @xmath395 $ ] ,
we need to impose that the joint distribution of @xmath119 and @xmath241 is preserved by the given choice of the test channel . to this end , note that the joint distribution of @xmath119 and @xmath241 is such that we can write @xmath396 , where @xmath387 is zero - mean gaussian and independent of @xmath241 and @xmath397 , with @xmath398=1-\rho^{2d}$ ] .
therefore , preservation of the joint distribution of @xmath119 and @xmath241 is guaranteed if the equality @xmath395+d_{1}=1-\rho^{2d}+\sigma_{n}^{2}$ ] holds .
this leads to @xmath399=1-\rho^{2d}+\sigma_{n}^{2}-d_{1}.\ ] ] we remark that @xmath400\leq1 $ ] , due to the assumed inequality on the distortion @xmath226 .
references r. venkataramanan and s. s. pradhan , source coding with feed - forward : rate - distortion theorems and error exponents for a general source , _ ieee trans .
inform . theory _
2154 - 2179 , jun . 2007 .
r. venkataramanan and s. s. pradhan , directed information for communication problems with side - information and feedback / feed - forward , in _ proc .
of the 43rd annual allerton conference _ , monticello , il , 2005 .
r. venkataramanan and s. s. pradhan , `` on computing the feedback capacity of channels and the feed - forward rate - distortion function of sources , '' _ ieee trans .
58 , no . 7 , pp . 18891896 , jul .
2010 .
s. s. pradhan , `` on the role of feedforward in gaussian sources : point - to - point source coding and multiple description source coding , '' _ ieee trans .
inform . theory _
1 , pp . 331 - 349 , jan .
2007 .
h. permuter , y .- h .
kim and t. weissman , `` interpretations of directed information in portfolio theory , data compression , and hypothesis testing , '' _ ieee trans .
inform . theory _
57 , no . 6 , pp . 3248 - 3259 , jun .
2011 .
d. vasudevan , c. tian , and s. diggavi , lossy source coding for a cascade communication system with side - informations , in communication , control , and computing , 2006 44th annual allerton conference on , sept . 2006 | for memoryless sources , delayed side information at the decoder does not improve the rate - distortion function .
however , this is not the case for sources with memory , as demonstrated by a number of works focusing on the special case of ( delayed ) feedforward . in this paper , a setting is studied in which the encoder is potentially uncertain about the delay with which measurements of the side information , which is available at the encoder , are acquired at the decoder . assuming a hidden markov model for the source sequences , at first , a single - letter characterization is given for the set - up where the side information delay is arbitrary and known at the encoder , and the reconstruction at the destination is required to be asymptotically lossless .
then , with delay equal to zero or one source symbol , a single - letter characterization of the rate - distortion region is given for the case where , unbeknownst to the encoder , the side information may be delayed or not , and additional information can be received by the decoder when the side information is not delayed . finally , examples for binary and gaussian sources are provided .
rate - distortion function , hidden markov model , markov gaussian process , multiplexing , strictly causal side information , causal conditioning . |
chiral theory has caught up as an important tool to deal with hadron reactions at low and intermediate energies , accounting for the underlying basic qcd dynamics . by introducing chiral lagrangians where the explicit fields are mesons and baryons , a perturbative expansion in powers of momenta is done , leading to chiral perturbation theory ( @xmath3 ) which has allowed great progress in hadron physics .
a very important development in this direction is done with the introduction of unitarity in coupled channels which has allowed to extend the predictions of chiral theory at higher energies than is possible with chiral perturbation theory .
a review of the basic ideas can be found in @xcite .
one of the important issues of this unitarized theory is that it allows to deal with resonances , some of which appear necessarily as a consequence of the dynamics of the chiral lagrangians , in a similar way as bound states or resonances appear in potential theory in quantum mechanics .
we call these resonances dynamically generated and by now about 10 - 15 percent of the particles in the pdg can qualify for this nature , meaning that the meson- meson cloud , or meson - baryon cloud , becomes the basic part of the wave function overriding the relevance of the primary building blocks : constituent @xmath4 or three @xmath5 for the case of mesons or baryons , respectively .
i shall mention here some of these cases .
the findings about the nature of some resonances have also consequences in nuclear physics , since basic features associated to the chiral dynamics of the resonances show up in interesting ways in nuclear processes which can then be used to stress the nature of these resonances .
i shall also give some example of that in this talk .
a more extensive description of the ideas exposed here can be found in @xcite , but we concentrate here on the recent developments not discussed in @xcite .
we skip here details on the chiral lagrangians which can be found in @xcite and sketchhow unitarity enters the framework .
one can find a systematic and easily comprehensible derivation of the ideas of the n / d method applied for the first time to the meson baryon system in @xcite , which we reproduce here below and which follows closely the similar developments used before in the meson meson interaction @xcite .
one defines the transition @xmath6matrix as @xmath7 between the coupled channels which couple to certain quantum numbers .
for instance in the case of @xmath8 scattering studied in @xcite the channels with zero charge are @xmath9 , @xmath10 , @xmath11,@xmath12 , @xmath13 , @xmath14 , @xmath15 , @xmath16 , @xmath17 , @xmath18 .
unitarity in coupled channels is written as @xmath19 where @xmath20 , with @xmath21 the modulus of the c.m .
three momentum , and the subscripts @xmath22 and @xmath23 refer to the physical channels .
this equation is most efficiently written in terms of the inverse amplitude as @xmath24 the unitarity relation in eq .
( [ uni ] ) gives rise to a cut in the @xmath25matrix of partial wave amplitudes , which is usually called the unitarity or right hand cut .
hence one can write down a dispersion relation for @xmath26 @xmath27 where @xmath28 is the value of the @xmath29 variable at the threshold of channel @xmath22 and @xmath30 indicates other contributions coming from local and pole terms , as well as crossed channel dynamics but _ without _ right hand cut .
these extra terms are taken directly from @xmath3 after requiring the _ matching _ of the general result to the @xmath3 expressions .
notice also that the curled bracket , that we call the @xmath31 function , is the familiar scalar loop integral .
one can further simplify the notation by employing a matrix formalism . introducing the matrices @xmath32 , @xmath25 and @xmath33 ,
the latter defined in terms of the matrix elements @xmath34 and @xmath35 , the @xmath25-matrix can be written as : @xmath36^{-1}\cdot { \mathcal{t}}(w)~.\ ] ] which can be recast in a more familiar form as @xmath37 now imagine one is taking the lowest order chiral amplitude for the kernel @xmath33 as done in @xcite .
then the former equation is nothing but the bethe salpeter equation with the kernel taken from the lowest order lagrangian and factorized on shell , the same approach followed in @xcite , where different arguments were used to justify the on shell factorization of the kernel .
the on shell factorization of the kernel , justified here with the n / d method , renders the set of coupled bethe salpeter integral equations a simple set of algebraic equations .
the amplitudes of the channels discussed above are extrapolated to the complex second riemann sheet and we search for poles there . here
i discuss recent finding concerning the two poles corresponding to the @xmath0 resonance . in fig .
[ fig : tracepole ] we can see , from the work of @xcite , the trajectories of the poles when we start from an su(3 ) symmetric situation and go to the real world where this symmetry is broken .
this appears because of the different masses of the mesons and the baryon in the study of the interaction of the octet of the pion and the octet of the nucleon respectively .
the su(3 ) symmetric case corresponds to the poles in the real axis where we have two degenerate octets and a singlet .
as soon as su(3 ) symmetry starts being broken the degeneracy of the two octets disappears and two branches for i=0 and for i=1 develop which , at the end of the trajectories that represent the real world , correspond to well know resonances like the @xmath38 etc . with @xmath39 running from 0 to 1 .
the results are from @xcite . ]
the interesting thing to note is that in the region of the @xmath0 there are two poles , one from the original singlet and another one from one of the branches of the original octet . in practice
it is impossible to see two peaks in any reaction because the two resonances overlap , but the fact that the two poles couple differently to @xmath40 and @xmath41 ( the narrower pole couples strongly to @xmath8 and the wider one to @xmath42 ) has as a consequence that the @xmath0 should show up with quite different shapes in different reactions depending on their particular dynamics .
this is indeed the case as could be demonstrated recently with the performance of the @xmath43 @xcite experiment .
this reaction has been studied in @xcite and the basic mechanism for the reaction is given by fig .
[ two_exp ] left , where we see that it is dominated by the @xmath44 amplitude which has a strong contribution from the narrow pole at higher energies .
we should then expect that the peak of the @xmath0 appears at higher energies in this reaction and with a smaller width , and this is the case as one can see in fig .
[ two_exp ] right by comparing with the standard @xmath0 extracted from the @xmath45 reaction which was studied in @xcite .
the two experimental figures for the shapes of the @xmath0 in both reactions can be seen in fig .
[ two_exp ] right , giving a strong support to the existence of the two @xmath0 states .
( left ) and two experimental shapes of @xmath0 resonance ( right ) .
see text for more details.,title="fig : " ] ( left ) and two experimental shapes of @xmath0 resonance ( right ) .
see text for more details.,title="fig : " ]
given the success of the chiral unitary approach in generating dynamically low energy resonances from the interaction of the octets of stable baryons and the pseudoscalar mesons , in @xcite the interaction of the decuplet of @xmath46 with the octet of pseudoscalar mesons was studied and shown to lead to many states which were associated to experimentally well established @xmath47 resonances .
one of these would correspond to the @xmath2 .
in @xcite a refinement of the approach discussed above has been done including the @xmath8 and @xmath42 decay channels of the @xmath2 and finetuning the subtraction constants in the @xmath48 function , such that a good agreement with the position and width of the @xmath2 is attained . with this new information one
can face the study of the reaction @xmath49 by using the mechanisms of @xmath50 which provides the dominant contribution to the reaction at energies close to the @xmath2 . at higher energies of the experiment of
@xcite one finds other background mechanisms ( dashed line in fig .
[ sigfig2 ] ) providing a contribution that helps bring the theory and experiment in good agreement , as we can see in fig .
[ sigfig2 ] .
one can see in the figure that up to 575 mev / c of momentum of the @xmath51 the mechanism based on the strong coupling of the @xmath2 to the @xmath52 channel is largely dominant and provides the right strength of the cross section .
obviously it would be most interesting to investigate the region of lower energies of the @xmath51 in order to see if the predictions done by the theory are accurate .
this could be done for a different reaction , the @xmath53 @xcite , where the cross section is double than for @xmath54 . as we can see in fig .
[ sigfig2 ] , the predictions of the theory are in good agreement with the data .
( left ) and @xmath53 @xcite right.,title="fig : " ] ( left ) and @xmath53 @xcite right.,title="fig : " ]
the @xmath55 interaction in a nuclear medium has been the subject of many studies . in @xcite a selfconsistent calculation of the @xmath55 nucleus optical potential
is done and the results are consistent with data of kaonic atoms as shown in @xcite .
this potential has a strength of about 50 mev attraction at normal nuclear matter and an imaginary part of about 50 mev .
this leads to deeply bound states in nuclei of about 30 mev binding and widths of around 100 mev .
it was hence surprising to find other potentials where , with only three nucleons , the @xmath55 potential reaches a strength of about 650 mev at the center for the nucleus @xcite .
a closer inspection shows that the huge results come from several approximations among them the lack of selfconsistency in the calculations that , according to @xcite , leads to meaningless results , plus the shrinking of the nucleus to attain ten times nuclear matter density at the center of the nucleus .
based on this potential , some peaks seen at kek in the proton spectra following the absorption of @xmath51 on @xmath56 were interpreted as a signal of deeply bound kaonic states .
recently , it was shown in @xcite that the peaks of the experiments could be easily interpreted in terms of @xmath51 absorption by pairs of nucleons leading to @xmath57 or @xmath58 without further interaction of the baryons with the nucleus
. this claim has been recently reinforced by an experiment at finuda @xcite where the proton spectra is also studied following @xmath51 absorption in different nuclei and the same peaks are seen .
an additional measurement of pions in coincidence shows that they come from the decay of the @xmath59 for the peak at lower proton momentum , hence , confirming the claims of @xcite .
the finuda collaboration has another paper @xcite in which a wider peak has been seen in the invariant mass of @xmath60 measured back to back following @xmath51 absorption in different nuclei and which was claimed to correspond the a @xmath61 bound state . in another recent paper @xcite
it has been shown that such a peak appears unavoidably as a consequence of the mechanism of @xmath51 absorption by a pair of nucleons going to @xmath60 , followed by rescattering of the @xmath62 or the @xmath63 in the nucleus .
thus , the claims made for deeply bound kaons in nuclei with binding energies of about 200 mev and narrow widths are unfounded and this was the conclusion of a recent meeting explicitly devoted to this issue in @xmath64 trento @xcite .
within these techniques there is work done in @xmath65 selfenergy in the medium , @xmath66 and @xmath67 renormalization in nuclei and others , but i shall concentrate here on the latest issue which has connections with section 4 , where we studied the interaction of the decupled of baryons with the octet of nucleons .
the @xmath2 is one of the resonances dynamically generated with the channels @xmath52 and @xmath68 .
actually , this picture is only qualitative since the @xmath2 couples also to the @xmath42 and @xmath8 channels to which it decays .
the study done in @xcite including these channels finds that the coupling of the resonance to the @xmath52 is the strongest of all , as if it kept memory of what would happen in a simplified ideal world .
selfnergy in the nuclear medium ] even if the @xmath2 couples strongly to the @xmath52 channel , it does not decay into this state because it is below the threshold of @xmath52 .
however , in a nucleus the pion can easily excite a @xmath69 which has energy starting from zero .
thus we gain 140 mev phase space for the decay and this leads to a width in the medium from only this channel which is already bigger than the 14 mev free width of the @xmath2 .
this is not the only source of medium modification for the resonance , since the other decay channels are also modified in the medium .
a detailed study has been done @xcite ( see diagrams in fig . [ diagrams ] ) and we present the results in fig .
[ resultslam ] .
selfenergy as a function of the nuclear density and @xmath2 momentum . ]
the use of chiral lagrangians for the meson baryon interaction and the unitary extensions of chiral perturbation theory have allowed to face a large amount of problems which were barred to standard perturbation techniques .
it has opened the door to the study of many baryonic resonances which qualify neatly as dynamically generated resonances , or quasibound states of meson baryon .
thanks to this , a quantitative description of the meson baryon interaction at intermediate energies is now possible and with this , an important systematics has been introduced in the many body problem to face issues on the renormalization of hadron properties in the nuclear medium .
the constructed scheme is rather powerful and allows to make predictions which consecutive experiments are proving right .
two of these predictions , the two states for the @xmath0 , and the nature of the @xmath2 roughly as a quasibound state of @xmath52 , have found strong support from two very recent experiments .
the spectacular change of the width of the @xmath2 in the medium is partly due to the strong coupling of the resonance to the @xmath52 as predicted by chiral unitary dynamics .
finally , we also discussed that chiral theory , implementing selfconsistency in the many body calculation , leads to a moderate attraction for @xmath55 in nuclei , and that recents claims for deeply bound kaon states , which would require a huge @xmath55 optical potential , were a consequence of an incorrect identification of some peaks for which a natural conventional explanation has been just found .
altogether , one is seeing through all this work that chiral dynamics is a key ingredient that allows a unified description of much of the hadronic world at low and intermediate energies .
this work is partly supported by the spanish csic and jsps collaboration , the dgicyt contract number bfm2003 - 00856 , and the e.u .
euridice network contract no .
hprn - ct-2002 - 00311 .
this research is part of the eu integrated infrastructure initiative hadron physics project under contract number rii3-ct-2004 - 506078 .
99 j. a. oller , e. oset and a. ramos , prog .
* 45 * ( 2000 ) 157 .
e. oset , d. cabrera , v. k. magas , l. roca , s. sarkar , m. j. vicente vacas and a. ramos , lectures at puri hadron workshop , pramana * 66 * ( 2006 ) 731 [ arxiv : nucl - th/0504033 ] .
u. g. meissner , rep .
56 ( 1993 ) 903 ; v. bernard , n. kaiser and u. g. meissner , int .
e4 ( 1995 ) 193 .
d. jido , j.a .
oller , e. oset , a. ramos , u.g .
meissner , nucl .
phys . a * 725 * ( 2003 ) 181 .
s. prakhov _ et al . _ [ crystall ball collaboration ] , phys .
c * 70 * ( 2004 ) 034605 .
v. k. magas , e. oset and a. ramos , phys .
* 95 * ( 2005 ) 052301 .
y. akaishi , a. dote and t. yamazaki , phys .
b * 613 * ( 2005 ) 140 .
m. lutz , phys .
lett . b426 ( 1998 ) 12 .
e. oset and h. toki , phys .
c , in print .
arxiv : nucl - th/0509048 . m. agnello _ et al .
_ , arxiv : nucl - ex/0606021 . m. agnello _ et al .
_ [ finuda collaboration ] , phys . rev .
* 94 * ( 2005 ) 212303 .
v. k. magas , e. oset , a. ramos and h. toki , arxiv : nucl - th/0601013 . | in this talk i present recent developments in chiral dynamics of hadrons and hadrons in a medium addressing the following points : interaction of the octet of pseudoscalar mesons with the octet of baryons of the nucleon , showing recent experimental evidence on the existence of two @xmath0 states , the interaction of the octet of pseudoscalar mesons with the decuplet of baryons of the @xmath1 , with particular emphasis on the @xmath2 resonance , dynamically generated by this interaction .
then i review the interaction of kaons in a nuclear medium and briefly discuss the situation around the claims of deeply bound states in nuclei .
the large renormalization of the @xmath2 in the nuclear medium is shown as another example of successful application of the chiral unitary techniques . |
a main theme of this paper is the relation between the _ topological _ structure of a dynamical system and its _ measure - theoretic _ and _ ergodic _ properties . throughout the paper
we shall be concerned with the class @xmath2 of @xmath3 interval maps @xmath4 with negative schwarzian derivative and a finite number of non - flat critical points with possibly different orders , we give the precise definitions in section [ defs ] on page below .
our main result is to show that a sub - class of maps in @xmath5 admit equivalent characterizations in terms of some topological and metric properties of the critical orbits , which in turn imply several ergodic properties such as the existence of an absolutely continuous invariant measure .
this subclass is quite abundant in @xmath2 in the sense that general one - parameter families of maps will belong to it for a positive lebesgue measure set of parameters .
a _ physical measure _ for @xmath0 is any ergodic invariant measure @xmath6 with a basin @xmath7 ) of positive lebesgue measure , where @xmath8 , and the limit is in the weak-@xmath9 topology .
maps in @xmath2 can admit various kinds of physical measures such as dirac-@xmath10 measures supported on attracting fixed points or absolutely continuous measures .
there also exist examples for which @xmath0 admits no physical measures at all @xcite .
it was shown in @xcite that the number of physical measures for maps in @xmath2 is finite and bounded by the number of critical points .
we say that @xmath4 is _
( non - uniformly ) expanding _ if there exists a finite number @xmath11 , @xmath12 , of ergodic absolutely continuous invariant probability measures such that the union of their basins has full lebesgue measure and such that @xmath13 for each @xmath14 , @xmath15 . for simplicity
we assume that @xmath16 and so there is only one absolutely continuous invariant probability measure whose basin has full measure and which satisfies . in the general case our results hold for the restriction of @xmath0 to the basins @xmath17 of each ergodic measure @xmath18 .
in fact our results hold for the restriction of @xmath19 to the basin of any ergodic absolutely continuous invariant probability measure satisfying with no additional assumptions on the dynamics outside this basin .
we recall that @xmath4 and @xmath20 are _ topologically conjugate _ , or @xmath21 , if there exists a homeomorphism @xmath22 such that @xmath23 . a topological conjugacy preserves topological properties of maps such as dense and periodic orbits but in principle there is no reason for any measure - theoretic properties such as the existence of absolutely continuous invariant measures to be preserved .
we want to address the following question : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ suppose that @xmath0 is non - uniformly expanding and that @xmath24 .
does this imply that @xmath1 is non - uniformly expanding ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ a counterexample of bruin @xcite shows that the answer is negative in general even in the unimodal case
. however we prove the following [ generictheorem ] suppose that @xmath0 is non - uniformly expanding , @xmath24 , and that all critical points of @xmath25 are generic with respect to @xmath6 in the sense that @xmath26 then @xmath1 is non - uniformly expanding .
notice that holds for @xmath6 almost all points and in fact , by the absolute continuity of @xmath6 and the assumption that @xmath27 has full lebesgue measure , for lebesgue almost all points .
this justifies thinking of it as a `` generic '' condition .
we remark that the proof does not imply that the conjugacy @xmath28 is absolutely continuous .
rather , it is based on showing that @xmath1 satisfies certain conditions which can be shown by independent arguments to imply that @xmath1 is non - uniformly expanding .
the only existing positive results on the question of the topological invariance of non - uniform expansivity are in the unimodal setting .
the first example of a ( topological class of a ) map with an absolutely continuous invariant measure which is preserved by topological conjugacy was constructed by bruin @xcite , another such class is defined in @xcite .
sands @xcite formulated a ( topological ) condition on the kneading sequence of the map , i.e. the symbolic itinerary of the critical point , and showed that this implied the following condition @xmath29 for the ( single ) critical point @xmath30 ( and this implies non - uniform expansivity , see the next section ) .
notice that is strictly weaker than .
it follows from our results that sands topological condition is actually equivalent to a metric condition which we will call tsujii s condition , to be defined below , and which is strictly stronger than and strictly weaker than .
it was then shown in @xcite that itself is topologically invariant , by bringing together results of previous papers @xcite and observing that these results formed a closed chain of implications which included a condition which is clearly invariant by topological conjugacy .
these result do not however generalize to the multimodal setting : a counterexample was recently constructed in @xcite exhibiting a pair of quasi - symmetrically ( in particular topologically ) conjugate multimodal maps of which one satisfies and the other does not .
we shall show nevertheless that the slightly stronger tsujii condition mentioned above is topologically invariant for completely general multimodal maps .
this condition is satisfied by generic maps .
first we give a more in - depth discussion of condition and its relation to .
condition is more often formulated as @xmath31 it is traditionally known as the collet - eckmann condition since it was first introduced in @xcite and shown , together with some additional conditions which were later shown to be unnecessary @xcite , to imply non - uniform expansivity for unimodal maps .
the generalization of these results to the multimodal setting has been somewhat problematic for various technical reasons and only recently it has been shown that condition assumed to hold for _ every _ critical point does indeed imply non - uniform expansivity , see @xcite for the case when all critical points have the same order and @xcite for the general case .
we remark that in the general multimodal case condition is always satisfied by lebesgue almost every point if @xmath0 is non - uniformly expanding since it is equivalent to which follows from which holds for lebesgue almost every point by and the assumptions on @xmath6 .
this justifies the use of the term _ expanding _ for such a map @xmath19 .
the _ non - uniformity _ follows from the fact that the convergence in is not necessarily uniform and thus in general we have that the constant @xmath32 in is actually a function @xmath33 which depends on @xmath34 and such that @xmath35 : it might be necessary to wait for an arbitrary long time before the exponential behaviour becomes apparent .
the difference between and is quite subtle and does not usually play an important role in many argument which take advantage of the exponential growth property . in our results
however , this distinction will play a very important role .
we shall define below a set of conditions which together fall strictly between and .
the statement in theorem [ generictheorem ] is somewhat unsatisfactory due to the asymmetry between the assumptions and the conclusions : it is not the case in general that the critical points for @xmath1 are generic with respect to the corresponding absolutely continuous invariant measure for @xmath1 .
moreover it does not include some simple , albeit exceptional , cases , such as when the critical points are non - recurrent , which are clearly topologically invariant conditions and are known to imply non - uniform expansivity .
we shall therefore isolate the two main properties of genericity which are actually used in the proof and show that these properties are indeed topologically invariant .
one of these properties is the collet - eckmann condition ( or ) .
the other is a condition on the recurrence to the critical set .
let @xmath36 denote the standard euclidean distance in @xmath37 , let @xmath38 denote the set of critical points of @xmath0 and let @xmath39 for any @xmath40 , let @xmath41 denote a _
metric neighbourhood _ of the critical set .
we say that a ( critical ) point @xmath30 satisfies the _ slow recurrence _ condition if @xmath42 almost every point satisfies since @xmath43 and @xmath44 are uniformly comparable by the non - flatness of the critical points , birkhoff s
ergodic theorem implies that @xmath45 and the integrability condition implies that @xmath46 as @xmath47 .
it is therefore in some sense also a generic condition .
we say that a map satisfies condition ce if all its critical points satisfy the collet - eckmann condition and sr if all its critical points satisfy the slow recurrence conditions .
we are now ready to state our main result .
suppose that @xmath0 satisfies both ce and sr .
then every map @xmath1 topologically conjugate to @xmath0 also satisfies both ce and sr .
thus the simultaneous occurrence of both conditions ce and sr is topologically invariant .
we stress the fact that neither ce nor sr on their own are topologically invariant .
the measure - theoretical abundance of multimodal maps satisfying conditions ce and sr simultaneously was proved by tsujii @xcite for quite general families of multimodal maps , building on a theme of proving the abundance of maps satisfying various kinds of expansivity and recurrence conditions , all aimed at ultimately implying some form of non - uniform expansivity @xcite .
we mention also that non - uniformly expanding maps are in general very structurally unstable in the sense that arbitrarily small perturbations which do not preserve the topological structure of the map can destroy the expansivity , see @xcite .
we conjecture that tsujii s conditions might play a similar role in the multimodal setting to that played by the collet - eckmann conditions in the unimodal case .
several recent counterexamples show that some key properties of unimodal ce maps do not extend to the multimodal setting ; perhaps these counterexamples may be overcome by assuming tsujii s conditions instead .
our main theorem is precisely an example in which this is the case .
our arguments also indicate possible strategies for proving other technical properties for maps satisfying tsujii s conditions , such as _ backward collet - eckmann _ which is equivalent to ce in the unimodal case @xcite but does _ not _ follow from straightforward ce in the multimodal case @xcite , _ bounded backward contraction _ and _ exponential decay of correlations _ proved in @xcite and @xcite respectively for ce multimodal maps under the additional assumption that all critical points have the same order .
counterexamples exist showing that bounded backward contraction may fail for ce maps with critical points of different order @xcite .
these properties will be investigated in future papers .
our approach is quite distinct from the proof in @xcite of the invariance of the ce condition in the unimodal case .
the present strategy is more direct and in the spirit of sands thesis @xcite in which , in the unimodal setting , a topological condition is formulated in terms of the kneading sequence of the critical point and this condition is shown to imply the collet - eckmann condition .
the relation between sands topological condition and our slow recurrence condition was studied in @xcite in the unimodal case , and a generalization of the topological condition given in the bimodal case in @xcite . here
we push this whole approach significantly further by giving a more general definition of this condition which in particular applies to the multimodal setting , we call it a topological slow recurrence ( tsr ) condition , see on page . by construction
this condition is topologically invariant and thus our main result follows from proving the double implication for any @xmath48 we have @xmath49 in section [ combinatorial ] we give all the details of the combinatorial structure and define condition tsr .
we give a brief summary of the key ideas here .
the critical set @xmath38 defines a partition @xmath50 of @xmath51 which allows us to define a symbolic itinerary for all points which are not preimages of @xmath38 and to define cylinder sets made up of points which share the same itinerary up to some finite time .
the whole structure of cylinder sets is topological .
there is then a natural _ topological _ separation time : any point @xmath34 near @xmath52 shadows the orbit of the critical point for a time defined by the number of iterates for which @xmath34 and @xmath30 have the same symbolic itinerary .
this shadowing time tends to @xmath53 as @xmath34 tends to @xmath30 . in particular the symbolic sequence associated to any given point
contains information about its pattern of recurrence to the critical set : if it contains long finite blocks which coincide with long initial blocks of the symbolic sequences of one of the critical point then it must have come quite close to the critical set .
this simple idea is used in a crucial way in the definition of the condition tsr . in the remaining three sections we proceed to prove the three steps required to show .
we first show that ce + sr implies tsr , then that tsr implies sr and finally that tsr implies ce .
a particular effort has been made here to make the presentation as self - contained as possible and as much as possible independent of specialized technical notation and constructions familiar mainly to established specialists in one - dimensional dynamics . before ending this section
we give the formal definition of the class @xmath2 of maps under consideration and fix some fairly standard notation and definitions .
we shall consider _ s - multimodal maps with non - flat critical points _ by which we mean maps @xmath0 satisfying the following conditions .
let @xmath54 be a @xmath55 map , where @xmath56 is a closed interval .
we shall assume without loss of generality that @xmath57 $ ] .
we call @xmath58 a critical point of @xmath59 if @xmath60
. denote by @xmath61 the set of all critical points .
@xmath59 has finite number ( say @xmath62 ) of _ non - flat critical points _
@xmath63 if there exist @xmath64 , @xmath65 and a neighbourhood @xmath66 of @xmath67 in which @xmath68 .
@xmath59 has _ negative schwarzian derivative _ , i.e. @xmath69 for @xmath70 .
we remark that this condition implies in particular the non - existence of any inflection points .
thus all critical points are turning points , i.e. points at which the maps fails to be a local homeomorphism .
as part of the argument we shall need to obtain some derivative estimates about the iterates of @xmath0 along the forward orbit of the critical values . if these orbits map to a periodic point then the estimates are easily obtained .
otherwise we need to consider a two - sided neighbourhood of each critical value and the forward images of this neighbourhood .
we assume therefore that the forward orbits of all the critical points are disjoint from the boundary of @xmath37 or , if not , that the boundaries of @xmath37 are fixed points .
this is not restrictive since , if it is not the case , we can extend @xmath0 to a map @xmath71 defined on neighbourhood @xmath72 of @xmath37 such that the forward orbits of all the critical points are disjoint from the boundary of @xmath72 and the dynamics on @xmath37 is unaffected .
in this section we describe in detail the combinatorial structure associated to any s - multimodal map with non - flat critical points and having no neutral or attracting periodic orbits .
we emphasize that everything we describe here relies on no other assumptions .
some of the definitions are closely related to the so - called _ cutting / co - cutting times _ used in @xcite for unimodal maps ; we give a generalization to the multimodal setting and , we believe , a simplification of the way those concepts are introduced and applied .
we fix for the rest of the paper a constant @xmath73 sufficiently small so that the critical neighbourhood @xmath74 has precisely @xmath75 disjoint connected components ; in particular any @xmath76 for @xmath77 is associated unambiguously to one particular critical point . for any @xmath34 we shall use the notation @xmath78 , especially , @xmath79 for a critical point @xmath80 .
the critical points define a natural partition @xmath81 of @xmath37 into @xmath82 subintervals .
this allows us to associate a symbolic _ itinerary _ to any point not in the preimage of a critical point by letting @xmath83 for convenience we associate to each critical point two sequences which differ only in the first term , so that we think of the partition elements as closed intervals and think of each critical point as belonging to both of the adjacent intervals .
we define the symbolic _ separation time _ between two points as @xmath84 as long as the itineraries @xmath85 and @xmath86 are both defined .
if @xmath87 is a critical point and @xmath34 is in one of the adjacent partition elements then we naturally consider the symbolic itinerary for @xmath30 for which @xmath88 .
in particular we define the time it takes for a point @xmath34 to `` separate '' from the critical set @xmath52 by @xmath89 notice that @xmath90 as @xmath91 .
we define _ topological neighbourhoods _
@xmath92 of the critical set @xmath93 by @xmath94 we fix a constant @xmath95 sufficiently large so that @xmath96 recall , and such that every point @xmath34 with @xmath97 is associated to some particular critical point @xmath30 and @xmath98 and @xmath99 . for
any given finite sequence @xmath100 with @xmath101 we define the _ cylinder set _ @xmath102 notice that such a cylinder set may be empty .
for a given point @xmath34 with a symbolic itinerary @xmath103 we define the @xmath104th order cylinder set of @xmath105 as @xmath106 this is always a two sided neighbourhood of @xmath107 . we shall often write @xmath108 to denote the part of @xmath109 to the left and right respectively of the point @xmath34
. clearly @xmath110 for any @xmath111 .
moreover , by the non - existence of wandering intervals @xcite the preimages of the critical set are dense and therefore we have @xmath112 in particular the cylinder sets @xmath113 define a nested sequence of neighbourhoods of the point @xmath107 .
we now fix some arbitrary point @xmath34 not contained in any preimage of the critical set .
we fix some notation related to the partition @xmath114 of a neighbourhood of @xmath34 defined naturally by the cylinder sets .
more precisely , consider a generic cylinder set @xmath115 . by definition
the images @xmath116 all belong to the same element of @xmath117 for all @xmath118 . then there are two possibilities : 1
. @xmath119 does not contain any critical points in its interior . in this case
all points continue to share the same combinatorics and we have @xmath120 , and thus in particular @xmath121 2 .
@xmath122 contains one or more critical points in its interior . in this case
the cylinder set @xmath123 is given precisely by those points of @xmath124 which fall into the same element of @xmath125 as @xmath126 and @xmath127 in this case @xmath128 may have either one or two connected components .
we index the times at which case 2 above occurs by two sequences @xmath129 where @xmath130 and @xmath131 for a general point @xmath34 , the two sequences @xmath132 and @xmath133 are independent even though there may well be one or more pairs @xmath134 such that @xmath135 . for the special case of the partition @xmath136 associated to one of the critical points @xmath137
, we actually have @xmath138 .
for each such time @xmath139 we define the intervals @xmath140 which define precisely the elements of the partition @xmath141 of a neighbourhood of @xmath34 , refer to figure 1 .
captypefigure we call the elements of the sequences @xmath132 and @xmath133 respectively the _ left and right shadowing times _ associated to @xmath34 because every point in @xmath142 `` shadows '' the point @xmath34 for @xmath143 iterations : @xmath144 the partition @xmath145 ( and in particular @xmath146 for some critical point @xmath30 ) can be thought of as topological versions of the partitions defined by the _ binding periods _ of benedicks and carleson @xcite . here
we lose some good properties of binding periods such as uniformly bounded distortion but gain in other ways , as shall become clear below .
first we state three relatively straightforward but not immediately obvious properties of this partition which will play an important role .
first of all notice that @xmath147 indeed , by construction the boundary points of each partition element @xmath148 are preimages of some critical point , and one of the boundary points of each cylinder set @xmath149 is a preimage of some critical point of order @xmath150 .
moreover , there are no points in the interior of @xmath151 which map to a critical point before time @xmath152 . thus follows .
secondly , for any @xmath153 close to @xmath34 we have @xmath154 for @xmath155 and so in particular @xmath156 .
therefore , for any @xmath157 the relation gives @xmath158 which implies .
thirdly , for every @xmath159 such that @xmath160 we have @xmath161 to see this consider the two intervals @xmath162 : = \hat i^{(n_{i+1}^{\pm})}_{\pm } \subset \hat i^{(n_{i}^{\pm})}_{\pm}= : [ z , x ] $ ] . by construction
we have @xmath163 for some critical point @xmath164 and thus @xmath165 = [ \tilde c , x^{n_{i}^{\pm } } ] $ ] and @xmath166 , i.e. the interval @xmath167 $ ] contains the point @xmath168 in its interior . by
, @xmath169 is monotone on @xmath170 which implies in particular that @xmath171 can not contain any critical point in its interior before time @xmath172 , i.e. @xmath173 for all @xmath174 . at this time we have @xmath175 for some ( other ) critical point @xmath176 .
therefore @xmath177 is exactly the first time at which the iterates of @xmath178 and of @xmath179 fall on different sides of some critical point , and therefore is exactly the separation time : @xmath180 .
if @xmath181 this implies that @xmath182 as in .
see figure 2 .
captypefigure we are now ready to formulate the _ topological slow recurrence
_ condition for an arbitrary point @xmath34 : @xmath183 we say that the map @xmath0 satisfies condition
tsr if all the critical points satisfy notice that this depends only on the orbit of the critical points with respect to the partitions @xmath136 for @xmath184 .
it is therefore invariant under topological conjugacy .
[ lemma : srtsr ] let @xmath0 satisfy condition ce .
then there exists a constant @xmath185 such that @xmath186 we shall use the following @xcite [ nowprz ] there exists constants @xmath187 such that for all intervals @xmath188 and integers @xmath189 such that @xmath190 is monotone we have @xmath191 we only need to prove the result for @xmath192 as the choice of the constant @xmath193 can take into account smaller values .
then there is a well - defined critical point @xmath178 `` closest '' to @xmath194 i.e. @xmath195 , and @xmath196 .
the map @xmath197 is monotone on the interval @xmath198 $ ] and therefore sublemma [ nowprz ] implies @xmath199 .
[ cor : tsr ] let @xmath0 satisfy conditions ce and sr .
then it satisfies condition tsr . by condition sr , for every @xmath200 there
exist @xmath201 such that @xmath202 therefore , choosing @xmath203 sufficiently large so that @xmath204 and using we have @xmath205 since @xmath206 is arbitrary , this implies tsr .
[ tsrsr ] let @xmath0 satisfy condition tsr .
then there exists a constant @xmath207 such that @xmath208 arguing as in the proof of corollary [ cor : tsr ] we then get let @xmath0 satisfy condition tsr .
then it satisfies condition sr .
we first give two preliminary results which will be used in the proof .
[ exp ] there exists constants @xmath209 such that for any interval @xmath188 and any positive integer @xmath104 such that @xmath188 is a one - side neighbrouhood of a critical point @xmath30 , @xmath210 contains a one - sided neighbourhood of the same critical point @xmath30 and @xmath211 , we have @xmath212 ( we do not assume that @xmath213 is monotone ) .
by the non flatness of the critical points , there exists a constant @xmath214 depending only on @xmath215 such that in either case we have @xmath216 .
moreover , letting @xmath217 , the mean value theorem implies @xmath218 , and , using the non - flatness of the critical points again , we have @xmath219 for some constants @xmath220 depending only on @xmath0 . combining these estimates give @xmath221 and therefore @xmath222 . [ integer ]
let @xmath59 satisfy condition tsr .
then for any @xmath223 there exists @xmath224 such that @xmath225 from tsr we have @xmath226 as @xmath227 . by this immediately implies the statement .
we first define in an iterative process which may stop according to one of two possible stopping rules .
we then show that in both cases the conclusions of the lemma are satisfied .
we fix some @xmath228 and let @xmath229 be the corresponding integer from sublemma [ integer ] .
let @xmath230 _ if @xmath231 go to step 3 .
_ otherwise we argue as follows .
since @xmath232 , there is a well defined critical point @xmath233 such that @xmath234 and @xmath235 .
by there exists some @xmath236 with @xmath237 and @xmath238 here we omit the subscripts @xmath239 not to overload the notation , and let @xmath240 or @xmath241 according to the relative positions of @xmath194 and @xmath242 .
by construction @xmath243 for some critical point @xmath244 . _
if @xmath245 go to step 2 .
_ otherwise we repeat the algorithm with @xmath246 playing exactly the role of @xmath194 above .
more precisely , we consider the separation time @xmath247 since @xmath248 , and give @xmath249 _ if @xmath250 go to step 3 .
_ otherwise there exists some @xmath251 such that @xmath252 and @xmath253 again , by construction @xmath254 for some critical point @xmath255 . _
if @xmath256 equals either @xmath242 or @xmath244 go to step 2 .
_ otherwise we repeat the process again with @xmath257 playing the role of @xmath258 and continue in this way until we go to either step 2 or step 3 . in the general case
we have @xmath259 and @xmath260 _ if @xmath261 go to step 3 .
_ otherwise there exists some @xmath262 with @xmath263 such that @xmath264 for some critical point @xmath265 . _
if @xmath266 equals any one of @xmath267 go to step 2 .
_ the process has to stop in a maximum of @xmath82 steps as by that time the above condition is necessarily satisfied .
see figure 3 .
captypefigure we suppose here that the procedure described above gives constants @xmath268 such that @xmath269 and therefore we have one - sided neighbourhoods @xmath270 and @xmath271 of the same critical point @xmath272 .
moreover , by construction we have that @xmath273 and therefore we are in a position to apply the estimates of sublemma [ exp ] to the interval @xmath274 if we can show that @xmath275 to see that this is the case we recall that @xmath276 satisfies for all @xmath277 ( for otherwise we would have gone straight to step 3 ) and thus in particular the separation time @xmath278 is strictly less than @xmath279 .
this implies that the separation time @xmath280 is strictly less than @xmath281 where both @xmath282 and @xmath283 are one - sided neighbourhood ( on the same side ) of the critical value @xmath284 .
clearly this implies that @xmath285 which is precisely .
therefore , by sublemma [ exp ] we have @xmath286 by construction we also have @xmath287 and therefore by the mean value theorem we have @xmath288 combining and we get that there exist constants @xmath289 such that @xmath290 finally , from we have that there exists a uniform constant @xmath291 such that @xmath292 which then gives @xmath293 this completes the proof in this case .
we now consider the situation in which the procedure described in step 1 leads to the existence of some @xmath294 such that @xmath295 .
since @xmath296 we have that @xmath297 where @xmath298 is some constant depending only on @xmath0 , @xmath95 and @xmath299 .
therefore a similar argument to that used in the final part of step 2 gives the result in this case also .
we now show that for s - multimodal maps the topological slow recurrence condition implies the collet - eckmann condition .
we shall be concentrating here on the cylinder sets associated to the _ critical values _ rather than those associated to the critical points used in the arguments given in the previous section .
consider a critical value @xmath300 with shadowing times @xmath150 and corresponding intervals @xmath301 and @xmath302 as in section [ combinatorial ] .
for any ( large ) @xmath303 we say that the gap @xmath304 , respectively @xmath305 , between two left , respectively right , shadowing times is _ short _ if @xmath306 , respectively @xmath307 .
we say that a left short gap and a right short gap are _ simultaneous _ if they are contained in an interval @xmath308 $ ] of times of length @xmath309 . in section [ density ]
we show that there exist values of @xmath310 for which there are many intervals of simultaneous short shadowing times . in section [ step1 ]
we show that each such interval implies a uniform estimate on the exponential shrinking of the length of intervals associated to some particular cylinder sets at the corresponding critical value . finally , in section [ global ] we combine these two results to obtain the collet - eckmann property .
[ posden ] for every small @xmath311 there exists @xmath312 , @xmath313 and @xmath314 such that the number of disjoint simultaneous short ( with respect to @xmath315 ) shadowing time intervals associated to any critical values , and occurring before time @xmath104 , is @xmath316 for every @xmath317 .
suppose that a point @xmath34 satisfies condition .
then for all @xmath311 sufficiently small , there exist @xmath318 such that @xmath319 for all @xmath320 and @xmath321 .
indeed , the equality follows by , the first inequality follows simply because the summation on the left is over a subset of times of the summation on the right , and the final inequality follows directly from .
in particular we have @xmath322 the calculation holds for both left and right shadowing times independently and therefore there exists at least @xmath323 iterates which belong to a small gap for both the left and the right sequence of shadowing times simultaneously .
each such iterate is therefore contained in an interval @xmath324 of simultaneous short shadowing times and thus it is possible to define at least @xmath325 disjoint such intervals . finally , notice that if some point @xmath34 satisfies condition then any other point on the orbit of @xmath34 also satisfies .
in particular these conclusions hold for all critical values as required .
[ expshr ] for any integer @xmath326 , there exists a constant @xmath327 such that for any critical value @xmath328 and any associated pair @xmath329 $ ] and @xmath330 $ ] of simultaneous left and right short gaps we have @xmath331 let @xmath332 $ ] be the interval containing @xmath329 $ ] and @xmath330 $ ] and satisfying @xmath333 .
we claim first of all that there exists a @xmath334 such that @xmath335 to see this , let @xmath336 denote the set of all images and preimages of the critical set between time @xmath337 and @xmath338 and define @xmath339 to be the minimum distance between any two points in this set .
now let @xmath340= \hat{i}_{-}^{(n_{i-1}^{-})}\setminus \hat{i}_{-}^{(n_{i}^{- } ) } $ ] and @xmath341= \hat{i}_{+}^{(n_{j-1}^{+})}\setminus \hat{i}_{+}^{(n_{j}^{+})}$ ] .
then , by construction we have @xmath342 and therefore , since @xmath343 are all within @xmath344 iterates of @xmath345 we have @xmath346 these are the endpoints of the intervals @xmath347 and @xmath348 and thus the claim follows .
( see figure 4 ) .
captypefigure we now choose two points @xmath349 and @xmath350 such that @xmath351|=|f^{k}[x',y]| \geq \delta/2 $ ] and @xmath352|=|f^{k}[w',w]| \geq \delta/2 $ ] . by we observe that the map @xmath353 is monotone on @xmath354 $ ] and therefore the two extreme intervals @xmath355 $ ] and @xmath356 $ ] form the _ koebe space _ which guarantees uniformly bounded distortion in @xmath357 $ ] ( see @xcite ) .
in particular the proportion between the lengths of @xmath358 $ ] and @xmath359 $ ] is uniformly comparable to the proportion between the length of @xmath360 $ ] and @xmath361 $ ] and therefore we have @xmath362|}{|[y , c^1]| } \geq \frac{|[x ' , c^{1}]| } { |[y , c^{1}]| } \\ & \geq 1 + \frac{|[x ' , y]| } { |[y , c^{1}]| } \geq 1 + \mathcal \gamma \frac { |f^{k}[x',y]| } { |f^{k}[y , c^{1}]| } \geq 1 + \frac{\gamma\delta}{2|i| } \end{aligned}\ ] ] for a distortion constant @xmath363 which depends only on @xmath364 .
this proves the first inequality in the statement of the lemma with @xmath365 .
exactly the same argument gives the second inequality .
we are now ready to show that all critical values admit exponentially growing derivative along their orbits .
more specifically we shall prove the following [ ceproof ] suppose that every critical value satisfies ( tsr ) . then there exist constants @xmath366 such that for every critical value and every @xmath367 we have @xmath368 the exponential growth of the derivative then follows from the so - called _ minimum principle _ for maps with negative schwarzian derivative : [ negsch ] @xcite let @xmath59 be an s - multimodal map and @xmath369\subset i$ ]
be a subinterval . for @xmath370 , if @xmath371}$ ] is a diffeomorphism , then @xmath372 for any @xmath373 .
we therefore get the following suppose that every critical value satisfies ( tsr )
. then there exist constants @xmath366 such that for every critical value @xmath328 and every @xmath367 we have @xmath374 we begin with a sublemma which follow easily from the estimates obtained above .
[ shrink ] for all @xmath375 we have @xmath376 follows from lemmas [ posden ] and [ expshr ] .
we start by considering a ( left or right ) shadowing time @xmath150 .
let @xmath377 $ ] and recall that then @xmath378 for some critical point @xmath379 .
by lemma [ tsrsr ] we have @xmath380 therefore @xmath381 for every critical value and every shadowing time @xmath382 .
we now let @xmath150 be the smallest ( left or right ) shadowing time larger than @xmath383 .
then @xmath384 and so @xmath385 .
by we know that @xmath386 is monotone , so we have by , @xmath387 then , by and sublemma [ shrink ] , we obtain @xmath388 it is therefore sufficient to show that we can choose constants @xmath389 so that @xmath390 since @xmath391 , we can choose @xmath392 sufficiently large such that @xmath393 when @xmath394 .
so @xmath395 we start by taking @xmath396 , fixing some arbitrary @xmath397 and writing @xmath398 and @xmath399
. then we have @xmath400 and therefore @xmath401 it remains to show that @xmath402 this follows by the crucial facts that @xmath403 and @xmath404 by tsr and .
indeed , we use which implies that the left side of is bigger than @xmath405 for all @xmath406 for some sufficiently large @xmath407 depending only on the map and the constants @xmath408 and @xmath409 and thus , again using , for all @xmath410 for some sufficiently large @xmath411 also depending only on the same quantities as @xmath412 . to take care of smaller values of @xmath104
it is sufficient to choose @xmath413 sufficiently small depending only on the value of @xmath414 for @xmath415 .
this completes the proof .
lanyu wang would like to thank his phd supervisor professor zhifen zhang for her advise and encouragement , and professors zhujun jing and lan wen for their help and encouragement for many years . | we show that a strengthened version of the collet - eckmann condition for multimodal maps is topologically invariant . in particular , if @xmath0 is non - uniformly expanding and the critical points are generic with respect to the absolutely continuous invariant measure , then any map @xmath1 topologically conjugate to @xmath0 is also non - uniformly expanding . |
the 4@xmath3 configuration of pr ions in intermetallic compounds had been considered to be quite stable in view of no observation of strongly correlated electron behaviors until recent discovery of heavy fermion ( hf ) behaviors in some of pr - based compounds .
such unprecedented examples are prinag@xmath4 @xcite and filled skutterudites prfe@xmath1p@xmath2 @xcite and pros@xmath1sb@xmath2 @xcite . in the latter group ,
not all pr - based compounds show such unconventional behavior as summarized in table [ table:1 ] and at this stage it is not clear whether there is a systematic trend or not . in this article , we report recent studies on the thermal properties in the two pr - based compounds and discuss the nature of the hf ground states .
c|ccc & & & + fe & .thermal properties of prt@xmath1x@xmath2 .
see text for abbreviation . roughly estimated @xmath5 ( or @xmath6 at phase transitions ) are given in the unit of j / k@xmath7mol .
[ cols= " < " , ] + [ table:1 ]
in prfe@xmath1p@xmath2 , hf behavior is evidenced by the largely - enhanced electronic specific - heat - coefficient ( @xmath8 j / k@xmath7mol ) @xcite and cyclotron masses in dhva studies @xcite in applied fields where an anomalous ordered state ( ods ) , appearing below @xmath9 k , is suppressed .
non - magnetic nature of the ods has been revealed by the @xmath10pr - nuclear specific heat and nuclear scattering studies @xcite .
a possible order parameter of the ods is an antiferro - quadrupole ( afq ) @xcite , which couples to the observed lattice distortion @xcite and consequently to the expected fermi surface instability @xcite . in this scenario ,
a gap opening suggested from the electrical resistivity @xmath11 and hall coefficient @xmath12 @xcite and a field - induced staggered magnetic component observed in neutron scattering experiments @xcite can be naturally understood .
this scenario might point to a possibility that the quadrupole fluctuation is essential for the hf behavior in the high fields .
we have extended our study to la - substituted system ( pr@xmath13la@xmath14)fe@xmath1p@xmath2 .
it is found that the ods smears out rapidly with the la doping and a new magnetic transition appears for @xmath15 at @xmath16 k ( depending on @xmath17 ) , suggesting an competition between the quadrupole and the magnetic interactions in this compound ( reported elsewhere ) .
interestingly , after the suppression of the ods by the la doping , the hf behavior is recovered even in zero field and the @xmath5 value below @xmath18 is @xmath19 j / k@xmath7mol - pr at @xmath20 @xcite , indicating the compatibility between the magnetic ordering and the hf behavior .
compelling evidence for the hf superconductivity ( hfsc ) in pros@xmath1sb@xmath2 was given by a large specific heat jump @xmath21 j / k@xmath7 mol at @xmath22 k on a pellet of compressed powdered single crystals @xcite .
the jump is superimposed on a schottky - like anomaly appearing at @xmath23 k , which could be interpreted as a crystalline - electric - field ( cef ) excitation or a strongly energy - dependent quasiparticle excitation . in order to study the nature of the anomalous normal state
, we have measured specific heat in magnetic fields using a single crystal of high quality , which has been ensured by an observation of the dhva oscillations @xcite .
the schottky - like anomaly was found to be drastically suppressed with increasing magnetic field , i.e. , the downward curvature in @xmath24 disappears in 5 t or higher @xcite , indicating that the low - lying excitation leading to the anomaly has a magnetic character .
another important finding in our @xmath24 data is a clear thermodynamical evidence for the existence of a field - induced ordered phase ( fiop ) above 4.5 t. as shown in fig .
[ fig ] ( a ) , a distinct anomaly evidencing a field - induced phase transition develops with increasing field .
the transition temperature @xmath25 is plotted in a @xmath26-vs-@xmath27 phase diagram of fig .
[ fig ] ( b ) . below 8 t
, @xmath28 is positive and the ehrenfest s theorem suggests that magnetization @xmath29 is enhanced in the fiop . this behavior is actually reflected in the field dependence of the @xmath10pr - nuclear contribution @xcite , which causes the upturn in @xmath24 below 0.5 k ( see fig . [ fig ] ( a ) ) .
we speculate that the order parameter is of an afq accompanied by a field - induced antiferromagnetic ( af ) component , as observed in ceb@xmath30 @xcite and tmte @xcite , since af correlations inferred from @xmath31 could energetically stabilize the fiop in magnetic fields leading to @xmath32 .
bauer _ et al .
_ pointed out a possibility that the hf behavior in pros@xmath1sb@xmath2 involves a quadrupole kondo interaction on pr ions based on their cef model , which has a non - kramers doublet ground state @xcite .
from this view point , it is essentially important to clarify whether any residual entropy is hidden below 0.1 k since , for example , a residual entropy of @xmath33 is predicted for a single - site quadrupole kondo model @xcite . to check this
, we performed quantitative measurements of magnetocaloric effect , which have worked satisfactorily to study the thermal properties of the metamagnetic - like anomaly in ceru@xmath4si@xmath4 @xcite .
no anomalous entropy ( @xmath34 ) has been detected in @xmath35 t @xcite .
we can not deny , however , a possibility that pr ions in pros@xmath1sb@xmath2 form a quadrupole kondo lattice with an effective singlet ground state .
determination of the order parameter of the fiop is essential to understand the anomalous nature of the normal state , which provides the background of the rare @xmath0-based hfsc in pros@xmath1sb@xmath2 . | rare @xmath0-based heavy - fermion behaviors have been revealed recently in pr - based filled skutterudites prfe@xmath1p@xmath2 and pros@xmath1sb@xmath2 .
recent studies on the thermal properties on both compounds are reported , putting emphasis on the field - induced ordered phase found in pros@xmath1sb@xmath2 . |
recently there has been a resurgence of interest in f - theory as a non - perturbative formulation of type iib superstring theory which allows for the decoupling of gravity and for the formulation of gut theories ( first proposed in @xcite ) based on the gauge group @xmath0 ( see e.g. @xcite and references therein ) .
although descending from a high energy @xmath0 group , most of the models in the literature @xcite focus on reproducing the minimal supersymmetric standard model ( mssm ) at low energies , making it difficult to obtain an experimental link to f - theory . in this paper
we explore f - theory models in which the low energy supersymmetric theory contains the particle content of three 27 dimensional representations of the underlying @xmath0 gauge group .
the resulting low energy models will resemble either the e6ssm @xcite or a generalised nmssm+ @xcite depending on whether an additional abelian gauge group does or does not survive .
however there are novel features compared to both these models which , if observed , would provide circumstantial evidence for f - theory .
the f - theory models considered in this paper all descend from a parent @xmath5 gauge theory @xcite .
a crucial question for model construction is whether a gauged @xmath6 from the @xmath5 gauge theory can survive down to low energies , where the gauged @xmath6 may arise from one of the cartan generators of the non - abelian gauge group .
a clear example of this is the case of hypercharge @xmath7 , arising from su(5 ) after flux breaking in many f - theory models @xcite .
more generally , if we begin with the case of an @xmath0 gut group , we can break @xmath0 down to the standard model gauge group by turning on fluxes in @xmath6s in the following sequence of rank preserving breakings : e_6 & & so(10)u(1 ) _ [ so(10 ) ] + so(10 ) & & su(5)u(1 ) _ [ su(5)0 ] + su(5 ) & & su(3 ) su(2 ) u(1)_y [ sm0 ] .
for example , the @xmath8 under which the right handed neutrinos have no charge is given in terms of these @xmath6s by , @xmath9 in terms of f - theory model building , the case of successive flux breaking starting with @xmath0 can be studied in the so called spectral cover formalism @xcite , where the model building choices relating to the particle content of the model amount to making choices about the flux breaking . in the f - theory models considered in this paper
, there will either be a surviving gauged @xmath8 , or it will be broken by flux breaking at the gut scale .
the f - theory models with a surviving abelian gauge group resembles the e6ssm @xcite which is a supersymmetric standard model in which precisely such an extra @xmath8 gauge symmetry survives down to the tev scale .
however in the f - theory model the gauge coupling of the @xmath8 may differ from that on the e6ssm and may be much weaker for example .
the matter spectrum is similar to that of the e6ssm , namely three 27s of e6 which ensures anomaly cancellation .
this implies light exotics with the quantum numbers of higgs doublets and colour triplets of exotic quarks , arising from three @xmath10 representations of su(5 ) , plus three su(5 ) singlets which are charged under @xmath8 .
the coupling one of these singlets to @xmath11 generates an effective @xmath12 term after the the singlet acquires a low scale vacuum expectation value ( vev ) .
the f - theory version of the e6ssm does not require any additional states in order to achieve unification , unlike the e6ssm which includes an additional pair of doublet states called @xmath13 and @xmath14 @xcite .
it more closely resembles the minimal e6ssm ( me6ssm ) proposed in @xcite .
however , in the f - theory model , the gauge couplings at the gut scale are split by flux effects , while in the me6ssm , unification is achieved via an intermediate pati - salam gauge group .
proton decay represents another important difference between the two models . in the f - theory model proton decay
is suppressed by the geometric coupling suppression of a singlet state , which effectively suppresses the coupling of the exotic charge @xmath1 colour triplet state @xmath2 to quarks and leptons , while in the me6ssm all proton decay couplings are allowed but with highly suppressed coefficients .
this tends to give long lived @xmath2 decays in the me6ssm , but prompt @xmath2 decays in the f - theory model , with large couplings to left - handed quarks and leptons , providing characteristic and striking signatures at the lhc .
if there is no surviving extra abelian gauge group then the f - theory model resembles the nmssm+ which also involves three compete 27 dimensional families @xcite .
however , whereas in the nmssm+ the @xmath8 is broken by an additional sector close to the gut scale , in the f - theory model it is simply broken by flux breaking .
another important difference is that the nmssm+ is a scale invariant theory , involving only trilinear couplings such as the trilinear singlet couplings , while in the case of f - theory there are in addition singlet mass terms arising from non - perturbative effects , giving rise to a generalised version of the nmssm+ .
the phenomenological comments in the preceding paragraph concerning unification , proton decay and the @xmath3 couplings at the lhc all apply to this case as well where the @xmath8 is broken .
the main advantage of the nmssm+ over the e6ssm is that the fine - tuning is lower due to the absence of @xmath8 d - terms which would introduce a term in the higgs potential proportional to the fourth power of the @xmath15 mass as discussed in @xcite .
the nmssm+ , involving three compete 27 dimensional families , has lower fine - tuning than the nmssm , which in turn has lower fine - tuning than the mssm @xcite , making it the lowest fine - tuned model consistent with perturbative unification .
@xmath0 based f - theory models have been discussed previously , for example , issues concerning the global resolution of @xmath0 guts in @xcite , and the models of @xcite .
it should be noted that here we use abelian fluxes , whereas @xcite uses non - abelian fluxes .
the layout of the rest of the paper is as follows .
since many readers will not be familiar with f - theory , in section 2 of the paper we give a basic review of the motivation and basics of f - theory , building up to analysing the flux breaking mechanism in this way . in section 3 ,
the model building strategies are applied to the case of @xmath0 models .
the d flatness conditions are considered in order to calculate the singlet vevs of the model .
this allows us to calculate the scale at which the exotics decouple .
quark , charged lepton and neutrino masses are also discussed . in section 4 , unification and proton decay
are studied in the f - theory models . in section 5 ,
the @xmath0 based f - theory models are compared to the known e6ssm and nmssm+ models .
section 6 concludes the paper .
a major motivation for string theory is that it provides a consistent formulation of quantum gravity , the effects of which are expected to become important at the planck scale . with this achievement
though , comes the drawback that it is very hard to predict anything about low energy physics , due to the vast numbers of consistent solutions to the string theory equations of motion . if , however , we follow the arguments of @xcite and impose the conditions of unification and decoupling on the search for realistic models , the possibilities are severely restricted .
unification refers to the existence of a gut structure whereby the strong , weak and electromagnetic forces are described by a single gauge group and a single coupling constant at some high energy scale .
the fact that gravity is observed to be much weaker than the other forces is linked to the term `` decoupling '' , which refers to the existence of a theoretical limit where @xmath16 .
a class of models which satisfy both the criteria of unification and decoupling are f - theory guts .
f - theory is a 12 dimensional , strongly coupled formulation of type iib superstring theory . before considering this strongly coupled case , we can note the case of perturbative type iib string theory , which refers to a 10 dimensional theory where 6 of the dimensions are compactified , and the string coupling constant @xmath17 ( which governs how strongly the strings interact with one another ) is small , @xmath18 . in this perturbative regime ,
the particles of the standard model ( sm ) are described by excitations of open strings , whereas the graviton and gravitino are related to closed strings .
motivated by the weakness of gravity , one could try and formulate the sm by just using open strings , the ends of which are attached to d - branes @xcite . in the effective field theory of such d - brane constructions , sm or
gut matter arises when strings are attached to pairs of d - brane stacks , and so these matter fields are localised along the intersections of branes called `` matter curves '' .
a problem with this perturbative setup in su(5 ) guts , however , is that whilst we can generate a tree level yukawa coupling for the bottom quark , we can not for the top quark since both its chiral components live in the same gut representation ( the @xmath19 of su(5 ) ) and the yukawa interaction terms with a higgs @xmath20 , namely @xmath21 , do not match up an equal number of fundamental and anti - fundamental indices @xcite .
guts based on so(10 ) or exceptional groups also have problems .
the spinor 16 of so(10 ) can not be realised in open string perturbation theory , and e - type gauge groups are not possible .
this all suggests the need for extra non - perturbative ingredients . with a non - perturbative string coupling constant @xmath22
, exceptional gauge groups such as @xmath0 , @xmath23 and @xmath5 can be realised , meaning that now , in the language of an su(5 ) gut , both the @xmath24 and the @xmath25 couplings can be realised at tree level , due to the presence of these exceptional structures which were nt present in the case of perturbative strings .
this , therefore , leads us to the case where @xmath22 and f - theory .
formally , f - theory can be defined on a background @xmath26 where @xmath27 is 4 dimensional space time , and x is a calabi - yau ( cy ) complex fourfold .
it is assumed that x is elliptically fibered with a section over a complex three - fold base , @xmath28 .
what this means is that each point of the base @xmath28 is represented by a two - torus , these tori being called the fibres .
the dimensions occupied by the base are the 6 compactified dimensions of type iib string theory , and the complex modulus of the torus fibre encodes the axion and dilaton at each point on the base @xcite :
@xmath29 it is a fact that the presence of d7-branes ( filling 7 spatial dimensions and 1 time dimension ) affects the profile of the axio - dilaton , @xmath30 @xcite . as such , the reason that f - theory can be viewed as a 12 dimensional theory is that two dimensions are geometric dimensions which allow us to keep track of the variation of @xmath30 over the other ten dimensions . in f - theory , the gut group is realised on a 7-brane which wraps some 2 complex dimensional surface s. we can learn a lot by studying so called semi - local models , where the complications of global f - theory ( as discussed in @xcite)are avoided by just looking at regions close to the gut surface s. the ideas of local f - theory focus on the submanifold s , where the gut symmetry is localised .
we can consider intersections of the gauge brane with other 7-branes wrapping surfaces @xmath31 and supporting gauge groups @xmath32 .
along these intersections matter will reside , and so they are known as matter curves , @xmath33 . along the matter curves ,
the local symmetry group is enhanced to @xmath34 .
we can go one step further than this and then study the intersections of matter curves at points in s. when we have an intersection of matter curves , we induce a yukawa coupling and there is a further enhancement of the local symmetry to @xmath35 . in order to study yukawa couplings in the local setup , we can gain information by just considering the local area around the point of intersection on the surface s @xcite .
the semi - local approach to f - theory assumes that we have a parent @xmath5 gauge theory which is broken by a position dependent vev for an adjoint higgs field @xcite .
all the interactions in the theory are assumed to come from a single @xmath5 point of enhancement . at this point ,
all the matter curves of the theory meet , and the local symmetry group is enhanced all the way to @xmath5 .
as an example , we can take the gut group on s to be su(5 ) .
the breaking of @xmath5 to the gut group occurs as @xmath36 where the commutant of the gut group inside @xmath5 is called the perpendicular group , and in this case is @xmath37 .
the nature of the matter curves of the theory is found by decomposing the adjoint representation of @xmath5 as follows @xmath38
this equation shows us that we have twenty four singlet curves ( @xmath39 ) , five 10 curves , and ten @xmath40 curves .
the equations of these curves can be written in terms of the weights @xmath41 ( @xmath42 , @xmath43 ) , of the 5 representation of @xmath37 as follows @xmath44 in general , there are non linear relations between the @xmath41 and the coefficients of the elliptic fibration , which have the effect of identifying some of the @xmath41 .
the way in which the @xmath41 can be identified is determined by the monodromy group @xcite . as we are working in the semi - local picture , the full calabi - yau geometry has been decoupled , and so we must choose the monodromy group by hand . by requiring a tree level top quark yukawa coupling , we need at least a @xmath45 monodromy identifying two of the weights .
this is because we need the @xmath25 coupling to be invarient under the perpendicular u(1 ) symmetries .
as the top and anti - top come from the same 10 representation , they both have charge @xmath41 , and the up type higgs has charge @xmath46 , meaning that to cancel the charges we must have @xmath47 .
this can only be the case for @xmath48 , and so we must have an identification of at least two of the weights . from now on this minimal @xmath45 case will be assumed at all times , and we will identify @xmath49 . in the model building section of the paper , fields will be labelled by which representations of @xmath0 , so(10 ) and
su(5 ) they transform under , and also their charge under the perpendicular u(1)s , given by the appropriate linear combination of weights , as above .
we can take the gauge symmetry on s to be @xmath0 , @xmath50 or @xmath51 ( although f - theory models with no unification group have also been studied @xcite ) .
starting from @xmath5 , there are three equivalent symmetry breaking chains that can end up with @xmath51 , namely : @xmath52 as we can see from the above breaking chains , even if we break via @xmath0 or so(10 ) as the gut group , we always end up with an @xmath53 structure , which subsequently breaks down to the standard model .
the only difference between the three pictures is which u(1)s originate from the gut group and which originate from the perpendicular group . throughout this paper
, we will assume that the gut group is broken down to @xmath54 via flux breaking .
there are two types of flux that can be turned on : there are fluxes in the u(1)s from the perpendicular group which preserve the chirality of complete gut representations , and there are fluxes that can be turned on in the worldvolume of the 7-brane which break the gut structure . whenever we utilise flux breaking we end up with splitting equations which tell us the net number of states in a particular representation , for example , breaking su(5 ) down to the standard model by turning on a flux in the hypercharge direction ( as discussed in @xcite ) gives the following equations for the 10 and 5 representations of su(5 ) 10= \ { llrep.&@xmath55 + n_3,2 ^ 1-n_|3,2 ^ 1&:m_10 + n_|3,1 ^ 1-n_3,1 ^ 1 & : m_10-n + n_|1,1 ^ 1-n_1,1 ^ 1 & : m_10+n + . & & 5= \ { llrep.&@xmath55 + n_3,1 ^ 1-n_|3,1 ^ 1&:m_5 + n_1,2 ^ 1-n_1,|2 ^ 1 & : m_5+n + .
we can see from these equations that the flux associated with the integer m respects the gut structure , and so is a flux in the perpendicular u(1)s .
the flux associated with the integer n is the hypercharge flux and leads to incomplete su(5 ) multiplets . as this breaking
is due to the hypercharge flux , the integer n is given by the flux dotted with the homology class of the matter curve in question .
as such , we can obtain relations between these n integers ( and similar integers for different fluxes ) by calculating the homology classes of the matter curves . in order to do this ,
the spectral cover formalism is used , and the results are summarised in @xcite .
we start by looking at the model of @xcite ( model 1 ) , which was motivated by the fact that if we build a model based on complete 27s of @xmath0 with no matter coming from the adjoint ( 78 ) representation , we automatically take care of anomaly cancellation . table [ 1 ] shows the model building freedom we have in choosing the m and n integers specifying the flux breaking , and how these choices determine the standard model particle content of the model . here
we make the same choices for the ms and ns as in @xcite and these choices are summarised in table [ 1 ] . in table [ 1 ] , arbitrary numbers of singlets are allowed in the spectrum for now , so that we can calculate the restrictions on these numbers later on .
the final column of table [ 1 ] shows the low energy spectrum of the e6ssm that we want to arrive at by eliminating the required exotics from the previous column , which shows the sm particle content after flux breaking . by comparing the final two columns of table [ 1 ]
, we can see that the exotics which we wish to remove are the vector pairs @xmath56 and @xmath57 .
large masses will be generated for these fields through their coupling to sm singlet fields which acquire large vevs . from the @xmath0 point of view
, the only @xmath0 allowed trilinear term in the superpotential is @xmath58 .
the vectorlike pairs which we wish to remove from the low energy particle content are those which have components in both the @xmath59 and @xmath60 multiplets . as such ,
they are removed by introducing @xmath61 , an @xmath0 singlet , with couplings : _
3127_t_1= _ 31q+ _ 31(2u^c)(2)+ _ 31d^c+ _
[ 31 ] if @xmath61 gets a large vev these vector states get large masses as required .
the difference between this case and model 1 @xcite is that in model 1 , @xmath62 also gets a large vev .
this singlet has the following couplings _ 345_1 = _
34[3d+2h_u][3 + 3h_d]= _ 34[3(d)]+_34[2(h_uh_d ) ] .
[ 34 ] in the e6ssm , these exotics are light , and so instead of getting a large vev , this singlet now must acquire a tev scale vev .
it needs to be checked that the f and d - flatness constraints are satisfied , and that rapid proton decay is forbidden for the realisation of the spectrum .
.complete @xmath63s of @xmath0 and their @xmath50 and @xmath51 decompositions .
the @xmath51 matter states decompose into sm states as @xmath64 and @xmath65 with right - handed neutrinos @xmath66 , while @xmath51 higgs states decompose as @xmath67 and @xmath68 , where @xmath69 are exotic colour triplets and antitriplets .
we identify rh neutrinos as @xmath70 .
arbitrary singlets are included for giving mass to neutrinos and exotics and to ensure f- and d- flatness . [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] the particle spectrum of the f - theory models is summarized in table [ 1 ] .
the models here may be compared to the f - theory model in @xcite in which the singlets @xmath62 acquired a string scale vev which gave large masses to the exotic states , yielding a low energy theory as in the mssm , which we can call an f - mssm .
the new models here have a singlet spectrum where the new flatness conditions allow the singlets @xmath62 to have small vevs resulting in a light exotic mass spectrum .
in addition the singlets @xmath71 are used to generate electroweak scale effective @xmath12 terms .
five right handed neutrinos , as well as other restrictions on the numbers of certain singlets in the spectrum , are required to make the model consistent with f and d - flatness conditions . if the gauged @xmath8 is broken by flux at the gut scale then we have either the f - mssm as discussed previously or the f - nmssm+ as investigated here , where non - perturbative corrections break all global @xmath6 symmetries via @xmath71 mass terms
. however if the gauged @xmath8 is unbroken then we are led to an f - e6ssm but with the phenomenological differences discussed above .
for example , we emphasise that unification in the f - mssm @xcite may be achieved approximately at the field theory level , since the exotic states occur at high energy and have a small mass splitting , while in the f - e6ssm and f - nmssm+ models discussed here the gauge coupling splitting due to flux in f - theory plays a crucial role .
the three different f - theory models are compared in table [ models table ] . in order for proton decay to be controlled
, the geometric suppression at the field theory level corresponds to the imposition of a discrete @xmath45 symmetry . to understand
the origin of this geometric suppression would require knowledge of the gut singlet matter curves , which in turn requires a knowledge of the global geometry . from our limited understanding of the global aspects of f
- theory this just corresponds to an assumption about the global completion of the model .
jcc and sfk would like to thank george leontaris and graham ross for very useful discussions , and for carefully reading the manuscript .
sfk acknowledges partial support from the stfc consolidated st / j000396/1 and eu itn grants unilhc 237920 and invisibles 289442 . c. beasley , j. j. heckman and c. vafa , `` guts and exceptional branes in f - theory - i , '' jhep * 0901 * ( 2009 ) 058 [ arxiv:0802.3391 [ hep - th ] ] .
r. donagi and m. wijnholt , `` breaking gut groups in f - theory , '' arxiv:0808.2223 [ hep - th ] . c. beasley , j. j. heckman and c. vafa , `` guts and exceptional branes in f - theory - ii : experimental predictions , '' jhep * 0901 * ( 2009 ) 059 [ arxiv:0806.0102 [ hep - th ] ] .
e. dudas and e. palti , `` froggatt - nielsen models from e(8 ) in f - theory guts , '' jhep * 1001 * ( 2010 ) 127 [ arxiv:0912.0853 [ hep - th ] ] .
e. dudas and e. palti , `` on hypercharge flux and exotics in f - theory guts , '' jhep * 1009 * ( 2010 ) 013 [ arxiv:1007.1297 [ hep - ph ] ] . s. f. king , g. k. leontaris and g. g. ross , `` family symmetries in f - theory guts , '' nucl .
b * 838 * ( 2010 ) 119 [ arxiv:1005.1025 [ hep - ph ] ] .
j. c. callaghan , s. f. king , g. k. leontaris and g. g. ross , `` towards a realistic f - theory gut , '' jhep * 1204 * ( 2012 ) 094 [ arxiv:1109.1399 [ hep - ph ] ] . c. ludeling , h. p. nilles and c. c. stephan , phys .
d * 83 * ( 2011 ) 086008 [ arxiv:1101.3346 [ hep - th ] ] .
s. f. king , s. moretti and r. nevzorov , `` theory and phenomenology of an exceptional supersymmetric standard model , '' phys .
d * 73 * ( 2006 ) 035009 [ arxiv : hep - ph/0510419 ] .
s. f. king , s. moretti and r. nevzorov , `` exceptional supersymmetric standard model , '' phys .
b * 634 * ( 2006 ) 278 [ hep - ph/0511256 ] .
s. f. king , s. moretti and r. nevzorov , `` gauge coupling unification in the exceptional supersymmetric standard model , '' phys .
b * 650 * ( 2007 ) 57 [ hep - ph/0701064 ] . r. howl and s. f. king , jhep * 0801 * ( 2008 ) 030 [ arxiv:0708.1451 [ hep - ph ] ] ; r. howl and s. f. king , phys .
b * 652 * ( 2007 ) 331 [ arxiv:0705.0301 [ hep - ph ] ] .
m. kuntzler and s. schafer - nameki , arxiv:1205.5688 [ hep - th ] .
m. cvetic , r. donagi , j. halverson and j. marsano , arxiv:1209.4906 [ hep - th ] .
c. m. chen and y. c. chung , `` on f - theory @xmath0 guts , '' jhep * 1103 * , 129 ( 2011 ) [ arxiv:1010.5536 [ hep - th ] ] .
s. katz , d. r. morrison , s. schafer - nameki and j. sully , jhep * 1108 * ( 2011 ) 094 [ arxiv:1106.3854 [ hep - th ] ] .
j. tate , `` algorithm for determining the type of a singular fiber in an elliptic pencil , '' in modular functions of one variable iv , lecture notes in math .
476 , springer - verlag , berlin ( 1975 ) .
m. bershadsky , k. a. intriligator , s. kachru , d. r. morrison , v. sadov and c. vafa , `` geometric singularities and enhanced gauge symmetries , '' nucl .
b * 481 * ( 1996 ) 215 [ arxiv : hep - th/9605200 ] .
m. cvetic , i. garcia - etxebarria and j. halverson , jhep * 1101 * ( 2011 ) 073 [ arxiv:1003.5337 [ hep - th ] ] .
s. cecotti , c. cordova , j. j. heckman and c. vafa , `` t - branes and monodromy , '' jhep * 1107 * ( 2011 ) 030 [ arxiv:1010.5780 [ hep - th ] ] . g. k. leontaris and g. g. ross , `` yukawa couplings and fermion mass structure in f - theory guts , '' jhep * 1102 * ( 2011 ) 108 [ arxiv:1009.6000 [ hep - th ] ] .
r. donagi and m. wijnholt , `` higgs bundles and uv completion in f - theory , '' arxiv:0904.1218 [ hep - th ] .
j. marsano , n. saulina and s. schafer - nameki , `` monodromies , fluxes , and compact three - generation f - theory guts , '' jhep * 0908 * ( 2009 ) 046 [ arxiv:0906.4672 [ hep - th ] ] .
k. s. choi , `` su(3 ) x su(2 )
x u(1 ) vacua in f - theory , '' nucl .
b * 842 * ( 2011 ) 1 [ arxiv:1007.3843 [ hep - th ] ] .
j. marsano , `` hypercharge flux , exotics , and anomaly cancellation in f - theory guts , '' phys . rev .
* 106 * ( 2011 ) 081601 [ arxiv:1011.2212 [ hep - th ] ] .
t. w. grimm and t. weigand , `` on abelian gauge symmetries and proton decay in global f - theory guts , '' phys .
d * 82 * ( 2010 ) 086009 [ arxiv:1006.0226 [ hep - th ] ] .
g. camara , e. dudas and e. palti , jhep * 1112 * ( 2011 ) 112 [ arxiv:1110.2206 [ hep - th ] ] .
l. aparicio , a. font , l. e. ibanez and f. marchesano , jhep * 1108 * ( 2011 ) 152 [ arxiv:1104.2609 [ hep - th ] ] .
s. cecotti , m. c. n. cheng , j. j. heckman and c. vafa , `` yukawa couplings in f - theory and non - commutative geometry , '' arxiv:0910.0477 [ hep - th ] . s. f. king , `` large mixing angle msw and atmospheric neutrinos from single right - handed neutrino dominance and u(1 ) family symmetry , '' nucl .
b * 576 * ( 2000 ) 85 [ hep - ph/9912492 ] .
s. f. king , `` neutrino mass models , '' rept .
prog . phys .
* 67 * ( 2004 ) 107 [ hep - ph/0310204 ] .
r. blumenhagen , `` gauge coupling unification in f - theory grand unified theories , '' phys .
* 102 * ( 2009 ) 071601 [ arxiv:0812.0248 [ hep - th ] ] .
j. p. conlon and e. palti , phys .
d * 80 * ( 2009 ) 106004 [ arxiv:0907.1362 [ hep - th ] ] .
g. k. leontaris and n. d. tracas , `` gauge coupling flux thresholds , exotic matter and the unification scale in f - su(5 ) gut , '' eur .
j. c * 67 * ( 2010 ) 489 [ arxiv:0912.1557 [ hep - ph ] ] .
g. k. leontaris , n. d. tracas and g. tsamis , `` unification , kk - thresholds and the top yukawa coupling in f - theory guts , '' eur .
j. c * 71 * ( 2011 ) 1768 [ arxiv:1102.5244 [ hep - ph ] ] . g. k. leontaris and n. d. vlachos , `` on the gut scale of f - theory su(5 ) , '' phys .
b * 704 * ( 2011 ) 620 [ arxiv:1105.1858 [ hep - th ] ] .
j. r. ellis , k. enqvist , d. v. nanopoulos and k. tamvakis , phys .
b * 155 * ( 1985 ) 381 .
h.georgi , `` lie algebras in particle physics ( 1999 ) '' h. murayama and d. b. kaplan , `` family symmetries and proton decay , '' phys .
b * 336 * ( 1994 ) 221 [ arxiv : hep - ph/9406423 ] .
k. s. babu , j. c. pati and f. wilczek , `` fermion masses , neutrino oscillations , and proton decay in the light of superkamiokande '' , nucl .
b * 566 * ( 2000 ) 33 [ arxiv : hep - ph/9812538 ] .
t. goto and t. nihei , `` effect of rrrr dimension five operator on the proton decay in the minimal su(5 ) sugra gut model , '' phys .
d * 59 * ( 1999 ) 115009 [ arxiv : hep - ph/9808255 ] .
h. murayama and a. pierce , `` not even decoupling can save minimal supersymmetric su(5 ) '' , phys .
d * 65 * ( 2002 ) 055009 [ arxiv : hep - ph/0108104 ] .
u. ellwanger , c. hugonie and a. m. teixeira , `` the next - to - minimal supersymmetric standard model , '' phys .
* 496 * ( 2010 ) 1 [ arxiv:0910.1785 [ hep - ph ] ]
. s. f. king and p. l. white , `` nonminimal supersymmetric higgs bosons at lep-2 , '' phys .
d * 53 * ( 1996 ) 4049 [ hep - ph/9508346 ] .
g. g. ross , k. schmidt - hoberg and f. staub , jhep * 1208 * ( 2012 ) 074 [ arxiv:1205.1509 [ hep - ph ] ] . | f - theory is a non - perturbative formulation of type iib superstring theory which allows for the decoupling of gravity and for the formulation of gut theories based on the gauge group @xmath0 . in this paper
we explore f - theory models in which the low energy supersymmetric theory contains the particle content of three 27 dimensional representations of the underlying @xmath0 gauge group , plus two extra right - handed neutrinos predicted from f and d flatness . the resulting tev scale effective theory resembles either the e6ssm or the nmssm+ , depending on whether an additional abelian gauge group does or does not survive .
however there are novel features compared to both these models as follows : ( i ) if the additional abelian gauge group is unbroken then it can have a weaker gauge coupling than in the e6ssm ; ( ii ) if the additional abelian gauge group is broken then non - perturbative effects can violate the scale invariance of the nmssm+ leading to a generalised model ; ( iii ) unification is achieved not at the field theory level but at the f - theory level since the gauge couplings are split by flux effects , negating the need for any additional doublet states which are usually required ; ( iv ) proton decay is suppressed by the geometric coupling suppression of a singlet state , a mechanism peculiar to f - theory , which effectively suppresses the coupling of the exotic charge @xmath1 colour triplet state @xmath2 to quarks and leptons ; ( v ) the @xmath3 decays as a chiral leptoquark with couplings to left - handed quarks and leptons , providing characteristic and striking signatures at the lhc . * e6 models from f - theory * + james c. callaghan@xmath4 and stephen f. king@xmath4 , + @xmath4 _ school of physics and astronomy , university of southampton , _ _
so17 1bj southampton , united kingdom _ |
it is well known that the anomalous diffusion processes in various real - world complex systems can be well characterized by using fractional order anomalous diffusion models ( @xcite ; @xcite ) after the introduction of continuous time random walks ( ctrws ) in + @xcite .
regarded as a natural extension of the brownian motions , the ctrws are proven to be useful in deriving the time or space fractional order diffusion system by allowing the incorporation of waiting time probability density function ( pdf ) and general jump pdf ( @xcite ; @xcite ; @xcite ) .
for example , if the particles are supposed to jump at fixed time intervals with a incorporating waiting times , the particles then undergo a sub - diffusion process and the time fractional diffusion system is introduced to efficiently describe this process . as stated in @xcite and + @xcite , instead of analyzing a system by purely theoretical viewpoint ( for example ,
see @xcite ) , using the notions of sensors and actuators to investigate the structures and properties of systems can allow us to understand the system better and consequently enable us to steer the real - world system in a better way .
this situation happens in many real dynamic systems , for example the optimal control of pest spreading ( @xcite ) , the flow through porous media microscopic process ( @xcite ) , or the swarm of robots moving through dense forest ( @xcite ) etc .
it is now widely believed that fractional order controls can offer better performance not achievable before using integer order controls systems ( @xcite ; @xcite ) .
this is the reason why the fractional order models are superior in comparison with the integer order models .
moreover , it is worth noting that in many real dynamic systems , the regional observation problem occurs naturally when one is interested in the knowledge of the states in a subregion of the spatial domain ( @xcite ; @xcite ; @xcite ) .
focusing on regional observations would allow for a reduction in the number of physical sensors and offer the potential to reduce computational requirements in some cases .
in addition , it should be pointed out that the concepts of regional observability are of great help to reconstruct the initial vector for those non - observable system when we are interested in the knowledge of the initial vector only in a critical subregion of the system domain . motivated by the argument above , in this paper , by considering the locations , number and spatial distributions of sensors , our goal is to study the regional gradient observability of the riemann - liouville time fractional order diffusion process , which is introduced to better characterize those sub - diffusion processes ( @xcite . ) more precisely ,
consider the problem @xmath2 below and suppose that the initial vector @xmath3 and its gradient @xmath4 are unknown and the measurements are given by using output functions ( depending on the number and structure of sensors ) .
the purpose here is to reconstruct the initial gradient vector @xmath4 on a given subregion of the whole domain of interest .
we also explore the characterizations of strategic sensors when the system is regional gradient observability . moreover , there are many applications of gradient modeling .
for example , the concentration regulation of a substrate at the upper bottom of a biological reactor sub - diffusion process , which is observed between two levels ( see fig . [ fig1 ] ) ; anther example the energy exchange problem between a casting plasma on a plane target which is perpendicular to the direction of the flow sub - diffusion process from measurements carried out by internal thermocouples ( @xcite ) . for richer background on gradient modeling
, we refer the reader to @xcite and @xcite . to the best of our knowledge ,
no results are available on this topic and we hope that the results obtained here could provide some insights into the control theory analysis of the fractional order diffusion systems and be useful in real - life applications .
the rest contents of the present paper are structured as follows .
the problem studied and some preliminaries are introduced in the next section and in section @xmath5 we focus on the characteristic of the strategic sensors .
an approach which enables us to reconstruct the initial gradient vector of the system under consideration in the considered subregion is addressed in section @xmath6 .
several application examples are worked out in the end for illustrations .
in this section , we formulate the regional gradient observability problems for the riemann - liouville time fractional order diffusion system and then introduce some preliminary results to be used thereafter .
let @xmath7 be a connected , open bounded subset of @xmath8 with lipschitz continuous boundary @xmath9 and consider the following abstract time fractional diffusion process : @xmath10,~0<\alpha\leq 1 , \\\lim\limits_{t\to 0^+ } { } _ 0i^{1-\alpha}_{t}y(t)=y_0\mbox { supposed to be unknown , } \end{array}\right\}\ ] ] where @xmath11 generates a strongly continuous semigroup @xmath12 on the hilbert space @xmath13 , @xmath14 is a uniformly elliptic operator , @xmath15 , @xmath16 and @xmath17 denote the riemann - liouville fractional order derivative and integral with respect to time @xmath18 , respectively , given by ( @xcite and @xcite ) @xmath19 the measurements ( possibly unbounded ) are given depending on the number and the structure of the sensors with dense domain in @xmath20 and range in @xmath21 as follows : @xmath22 where @xmath23 is the finite number of sensors .
let @xmath24 and both the initial vector @xmath3 and its gradient are supposed to be unknown .
the system @xmath2 admits a unique mild solution given by ( @xcite and @xcite ) : @xmath25,\end{aligned}\ ] ] where @xmath26 @xmath12 is the strongly continuous semigroup generated by @xmath11 , @xmath27 and @xmath28 is a probability density function defined by @xmath29 @xmath30 satisfying ( @xcite ) @xmath31 let @xmath32 be a given region of positive lebesgue measure and let @xmath33 then the regional gradient observability problem is concerned with the directly reconstruction of the initial gradient vector @xmath34 in @xmath35 . consider the following two restriction mappings @xmath36 their adjoint operators are ,
respectively , denoted by @xmath37 and @xmath38 moreover , by eq .
@xmath39 , the output function @xmath40 gives @xmath41 where @xmath42 . to obtain the adjoint operator of @xmath43
, we have + * case 1 .
@xmath44 is bounded ( e.g. zone sensors ) * + denote the adjoint operator of @xmath44 and @xmath45 by @xmath46 and @xmath47 , respectively .
since @xmath45 is a bounded operator ( @xcite ) , we get that the adjoint operator of @xmath43 can be given by @xmath48 * case 2 .
@xmath44 is unbounded ( e.g. pointwise sensors ) * + note that @xmath44 is densely defined , then @xmath46 exists . to state our results ,
the following two assumptions are needed : @xmath49 @xmath50 can be extended to a bounded linear operator @xmath51 in @xmath52 ; + @xmath53 @xmath54 exists and @xmath55 .
extend @xmath43 by @xmath56 one has @xmath57 based on the hahn - banach theorem , similar to the argument in @xcite , it is possible to derive the duality theorems as in @xcite and @xcite with the above two assumptions .
then the adjoint operator of @xmath43 can be defined as @xmath58 let @xmath59 be an operator defined by @xmath60 we see that the adjoint of the gradient operating on a connected , open bounded subset @xmath7 with a lipschitz continuous boundary @xmath9 is minus the divergence operator , i.e. , @xmath61 is given by ( @xcite ) @xmath62 where @xmath63 solves the following dirichlet problem @xmath64 similar to the discussion in @xcite ; @xcite and @xcite , it follows that the necessary and sufficient condition for the regional weak observability of the system described by @xmath2 and @xmath40 in @xmath35 at time @xmath65 is that @xmath66 and we see the following definition .
[ rgodef ] the system @xmath2 with output function @xmath40 is said to be regional weak gradient observability in @xmath35 at time @xmath65 if and only if @xmath67 [ proposition1 ] there is an equivalence among the following properties : @xmath68 the system @xmath2 is regional weak gradient observability in @xmath35 at time @xmath65 ; + @xmath69 @xmath70 + @xmath71 the operator @xmath72 is positive definite .
* by definition @xmath73 , it is obvious to know that @xmath74 as for @xmath75 in fact , we have @xmath76 let @xmath77 , which then allows us to complete the proof .
@xmath78 @xmath79 when @xmath80 , the system @xmath2 is deduced to the normal diffusion process as considered in @xcite , which is a particular case of our results .
@xmath81 a system which is gradient observable on @xmath35 is gradient observable on @xmath82 for every @xmath83 moreover , the definition @xmath73 is also valid for the case when @xmath84 and there exist systems that are not gradient observable but regionally gradient observable .
this can be illustrated by the following example .
let @xmath85\times [ 0,1 ] \subseteq \mathbf{r}^2 $ ] and consider the following time fractional order diffusion system of order @xmath86 $ ] .
@xmath87 , \\
y(\xi,\eta , t)= 0~\mbox { on } \partial\omega \times [ 0,b],\\ \lim\limits_{t\to 0^+}{}_0i^{1-\alpha}_{t}y(x_1,x_2,t)=y_0(x_1,x_2)~\mbox { in } \omega \end{array}\right.\end{aligned}\ ] ] with the output functions @xmath88 where @xmath89 and @xmath90 is the dirac delta function on the real number line that is zero everywhere except at zero . according to the problem @xmath2 , @xmath91
. then the eigenvalue , eigenvector and the semigroup @xmath92 on @xmath93 generated by @xmath11 are respectively @xmath94 , @xmath95 and @xmath96 moreover , one has ( @xcite ) @xmath97 where @xmath98 @xmath99 is the generalized mittag - leffler function in two parameters .
next , we show that there is a gradient vector @xmath100 , which is not gradient observable in the whole domain but gradient observable in a subregion @xmath101 .
let @xmath102 . by eq .
@xmath103 , we obtain that @xmath104 , then @xmath105 however , let @xmath106\times[0,1/6],$ ] we see that @xmath107 which means that @xmath108 is gradient observable in @xmath109 the following two lemmas play a significant role to obtain our results .
( @xcite)[integral by part ] for any @xmath110 $ ] , @xmath111 , the following formula holds @xmath112_{t = a}^{t = b}- \int_a^b{g(t){}_t^cd_b^{\alpha}f(t)}dt .
\end{aligned}\ ] ] ( @xcite)[lem1 ] let @xmath113 be an open set and @xmath114 be the class of infinitely differentiable functions on @xmath7 with compact support in @xmath7 and @xmath115 be such that @xmath116 then @xmath117 almost everywhere in @xmath118
this section is devoted to addressing the characteristic of sensors when the studied system is regionally gradient observable in a given subregion of the whole domain .
firstly , we recall that a sensor can be defined by a couple @xmath119 where @xmath120 is the support of the sensor and @xmath121 is its spatial distribution . for example , if @xmath122 with @xmath123 and @xmath124 where @xmath125 is the dirac delta function in @xmath35 at time @xmath65 that is zero everywhere except at @xmath126 the sensor is called pointwise sensor . in this case the operator @xmath44 is unbounded and the output function can be written as @xmath127 it is called zone sensor when @xmath128 and @xmath129 .
the output function is bounded and can be defined as follows : @xmath130 for more information on the structure characteristic and properties of sensors and actuators , we refer the reader to ( @xcite , @xcite , @xcite ) and the references cited therein . next , to state our results , it is supposed that the measurements are made by @xmath131 sensors @xmath132 , where @xmath133 and @xmath134 , @xmath135 then @xmath2 can be rewritten as @xmath136,\\ y(\eta , t)=0~~\mbox { on } \partial \omega \times [ 0,b],\\ \lim\limits_{t\to 0^+ } { } _ 0i^{1-\alpha}_{t}y(x , t)=y_0(x)~\mbox { in } \omega \end{array}\right.\ ] ] with the measurements @xmath137 where @xmath138 .
moreover , since the operator @xmath14 is a uniformly elliptic operator , for any @xmath139 @xmath140 , @xmath11 satisfies @xmath141}{\left[y_1(\eta , t)\frac{\partial y_2(\eta , t)}{\partial v_a}-y_2(\eta , t ) \frac{\partial y_2(\eta , t)}{\partial v_{a^*}}\right]}dtd\eta,\end{aligned}\ ] ] where @xmath142 is the adjoint operator of @xmath11 .
moreover , by @xcite , there exists a sequence @xmath143 , @xmath144 such that @xmath145 each @xmath146 @xmath147 is the eigenvalue of the operator @xmath11 with multiplicities @xmath148 and @xmath149 @xmath150 for each @xmath151 , @xmath152 is the orthonormal eigenfunction corresponding to @xmath146 , i.e. , @xmath153 where @xmath154 @xmath155 and @xmath156 is the inner product of space @xmath93 .
then it follows that the strongly continuous semigroup @xmath12 on @xmath93 generated by @xmath11 can be expressed as @xmath157 the sequence @xmath158 is an orthonormal basis in @xmath93 and for any @xmath159 , it can be expressed as @xmath160 a sensor ( or a suite of sensors ) is said to be gradient @xmath161strategic if the observed system is regionally gradient observable in @xmath35 .
[ lemma3.1 ] for any @xmath162 with @xmath163 @xmath164 , suppose that @xmath165 satisfies the following system @xmath166,\\ e(\eta , t)=0~~\mbox { on } \partial \omega \times [ 0,b ] , \\
e(x , b)=0 ~~\mbox { in } \omega , \end{array}\right.\ ] ] where @xmath142 is the adjoint operator of @xmath11 and @xmath167 denotes the right - sided caputo fractional order derivative with respect to time @xmath18 of order @xmath0 $ ] given by ( @xcite ; @xcite and @xcite ) @xmath168 then we obtain that @xmath169 * proof . *
replacing @xmath18 in @xmath170 by @xmath171 , we have @xmath172}ds\\ = -{}^c_0d^{\alpha}_{t}e(x , b - t ) . \end{array}\end{aligned}\ ] ] then the system @xmath173 is equivalent to @xmath174,\\ e(\eta , b - t)=0~~\mbox { on } \partial \omega \times [ 0,b ] , \\
e(x , b-0)=0 ~~\mbox { in } \omega . \end{array}\right.\ ] ] similar to the argument in @xcite and @xcite , the mild solution of @xmath173 can be given by @xmath175 on the other hand , it follows from the definition of the adjoint operator of @xmath43 that @xmath176 this allows us to complete the proof .
[ theorem3.1 ] for any @xmath177 , @xmath178 , given arbitrary @xmath179 , define the following @xmath180 matrices @xmath181 @xmath182_{p\times r_j},\ ] ] where @xmath183 , @xmath184 and @xmath185 . for
all @xmath186 let @xmath187 and @xmath188 .
then the necessary and sufficient condition for the gradient @xmath161strategic of the sensors @xmath132 is that @xmath189 in particular , when @xmath190 the sensors @xmath132 is gradient @xmath161strategic if and only if @xmath191
@xmath192 given arbitrary @xmath179 , by definition @xmath73 , the sensors @xmath132 are gradient @xmath161strategic if and only if for any @xmath193 , @xmath194 where @xmath195 with @xmath196 .
let @xmath197 .
by lemma @xmath198 we have @xmath199}{\partial x_s},p_{1\omega}^ * y_s\right)_{l^2(\omega)},\end{aligned}\ ] ] where @xmath200 is the solution of the system @xmath173 .
next , we explore the exact expression of @xmath201 consider the following problem @xmath202 , \\
\rho(\eta , t)=0~~\mbox{on } \partial\omega\times [ 0,b ] , \\
\lim\limits_{t\to 0^+}{}_0i^{1-\alpha}_t\rho(x , t)=p_{1\omega}^*y_s(x){\kern 5pt}~\mbox{in } \omega , \end{array}\right.\ ] ] where @xmath178 and the unique mild solution of system @xmath203 can be given by ( @xcite ) @xmath204 multiplying both sides of @xmath173 with @xmath205 and integrating the results over the domain @xmath206 $ ] , @xmath207\frac{\partial\rho(x , t)}{\partial x_s}}dtdx\\ = -\int_q{a^*e(x , t)\frac{\partial\rho(x , t)}{\partial x_s}}dtdx\\{\kern 8pt } + \int_0^b \int_\omega { \sum\limits_{i=1}^p{p_{d_i}f_i(x)z_i(t ) } \frac{\partial\rho(x , t)}{\partial x_s } } dxdt .
\end{array}\end{aligned}\ ] ] consider the fractional integration by parts @xmath208 in lemma @xmath209 , one has @xmath210 \frac{\partial\rho(x , t)}{\partial x_s}}dtdx\\= -\int_\omega{\left[\lim\limits_{t\to 0^+}{}_0i^{1-\alpha}_t\rho ( x , t)\right ] \frac{\partial e(x,0)}{\partial x_s}}dx\\{\kern 8pt}-\int_q{e(x , t)\left [ { } _ 0d^{\alpha}_t\frac{\partial\rho(x , t)}{\partial x_s}\right]}dtdx .
\end{array}\end{aligned}\ ] ] then the boundary condition gives @xmath211}{\partial x_s}\left[\lim\limits_{t\to 0^+}{}_0i^{1-\alpha}_te(x , t)\right]}dx\\\textcolor[rgb]{1.00,0.00,0.00}{= } \int_0^b \int_\omega { \sum\limits_{i=1}^p{p_{d_i}f_i(x)z_i(t ) } \frac{\partial\rho(x , t)}{\partial x_s } } dxdt , ~s=1,2,\cdots , n . \end{array}\end{aligned}\ ] ] thus , we have @xmath212}{\partial x_s},p_{1\omega}^ * y_s\right)_{l^2(\omega)}\\ & = & \sum\limits_{s=1}^n\left(\frac { \partial \left[-e(x,0)\right]}{\partial x_s},\left[\lim\limits_{t\to 0^+}{}_0i^{1-\alpha}_te(x , t)\right]\right)_{l^2(\omega)}\\ & = & \sum\limits_{s=1}^n\sum\limits_{j=1}^{\infty}\sum\limits_{k=1}^{r_j}\sum\limits_{i=1}^{p}\int_0^b { e_{\alpha,\alpha}(\lambda_j\tau^{\alpha})z_i(\tau)}d\tau \left(\frac{\partial \xi_{jk}}{\partial x_s},p_{d_i}f_i\right ) y_{jks}.\end{aligned}\ ] ] by lemma @xmath213 , since @xmath193 is arbitrary , we see that the system @xmath2 is regionally gradient observable in @xmath35 at time @xmath65 if and only if @xmath214 i.e. , for any @xmath215 , one has @xmath216 where @xmath217 is a vector in @xmath218
. finally , since @xmath219 for all @xmath220 , @xmath221 we then show our proof by using the reductio and absurdum .
@xmath222 necessity .
if @xmath223 and @xmath224 and @xmath225 there exists a nonzero element @xmath226 with @xmath227 such that @xmath228 then we can find a nonzero vector @xmath226 satisfying @xmath229 this means that the sensors @xmath132 are not @xmath161strategic .
@xmath230 sufficiency . on the contrary , if the sensors @xmath132 are not @xmath161strategic , i.e. , @xmath231 then there exists a nonzero element @xmath232 such that @xmath233 this allows us to complete the first conclusion of the theorem . in particular , when @xmath234 similar to the argument in @xmath222 , if @xmath223 and @xmath235 , there exists a nonzero vector @xmath226 satisfying @xmath236 then the sensors @xmath132 are not @xmath161strategic .
moreover , if the sensors @xmath132 are not @xmath161strategic , there exists a nonzero element @xmath237 satisfying @xmath238 then if @xmath223 , it is sufficient to see that @xmath239 for all @xmath177 .
the proof is complete .
this section is focused on an approach , which allows us to reconstruct the initial gradient vector of the system @xmath2 in @xmath109 the method used here is hilbert uniqueness method ( hums ) introduced by @xcite , which can be considered as an extension of those given in @xcite .
let @xmath240 be the set given by @xmath241 for any @xmath242 , there exists a function @xmath243 satisfying @xmath244 . consider the following system @xmath245 , \\
\lim\limits_{t\to 0^+ } { } _ 0i^{1-\alpha}_t\varphi(t)=\tilde{g}^ * , \end{array}\right.\ ] ] which admits a unique solution @xmath246\times \omega)$ ] given by @xmath247 then we consider the semi - norm on @xmath240 @xmath248 and we can get the following result .
[ lemma4.1 ] if the system @xmath2 is regionally gradient observable in @xmath35 at time @xmath65 , then @xmath249 defines a norm on @xmath240 .
@xmath250 if the system @xmath2 is regionally gradient observable , by definition @xmath73 , one has @xmath251 moreover , for any @xmath252 since @xmath253 which gives @xmath254 it then follows that @xmath249 defines a norm of @xmath240 and the proof is complete . for @xmath242 , consider the operator @xmath255 defined by @xmath256 where @xmath257 solves the following system @xmath258)$ ] @xmath259 controlled by the solution of the system @xmath260 .
we then conclude that the regional gradient reconstruction problem is equivalent to solving the equation @xmath261 .
[ theorem4.1 ] if @xmath2 is regionally gradient observable in @xmath35 at time @xmath65 , then @xmath261 has a unique solution @xmath242 and the initial gradient @xmath4 in subregion @xmath35 is equivalent to @xmath262 .
@xmath250 by lemma @xmath263 , we see that @xmath264 is a norm of the space @xmath240 provided that the system @xmath2 is regionally gradient observable in @xmath35 at time @xmath65 .
let the completion of @xmath240 with respect to the norm @xmath264 again by @xmath240 .
by the theorem 1.1 in @xcite , to obtain the existence of the unique solution @xmath242 of problem @xmath261 , we only need to show that @xmath265 is coercive from @xmath240 to @xmath266 i.e. , there exists a constant @xmath267 such that @xmath268 indeed , for any @xmath252 we have @xmath269 then @xmath265 is coercive and @xmath261 has a unique solution , which is also the initial gradient to be estimated in the subregion @xmath35 at time @xmath65 .
the proof is complete .
note that if the riemann - liouville fractional derivative in system @xmath270 is replaced by a caputo fractional derivative , its unique mild solution will be given by ( @xcite ) @xmath271.\end{aligned}\ ] ] we see that @xmath272 and @xmath273 then the lemma 7 fails .
new lemmas similar to lemma 4 and lemma 7 are of great interest .
besides , this challenge is also our interest now and we shall try our best to study it in our forthcoming papers .
let @xmath274\times[0,1]$ ] and @xmath275 . in this section ,
let us consider the following system @xmath276 , \\
y(\eta , t)= 0~\mbox { on } \partial\omega_2 \times [ 0,b],\\ \lim\limits_{t\to 0^+}{}_0i^{1-\alpha}_{t}y(x , t)=y_0(x)~\mbox { in } \omega_2 , \end{array}\right.\end{aligned}\ ] ] where @xmath277 is the elliptic operator given by @xmath278 for any @xmath279@xmath280 , let the gradient of initial vector be @xmath281 then our aim here is to present an approach to reconstruct the regional gradient vector : @xmath282 where the sensors may be zone , pointwise or filament ones .
+ + * case 1 . zone sensors * + suppose that @xmath283\times [ d_3,d_4]\subseteq \omega_2 $ ] and the output functions are @xmath284 where the system is observed by one sensor @xmath119 and @xmath44 is bounded .
moreover , we get that the eigenvalue , corresponding eigenvector of @xmath285 and the semigroup generated by @xmath285 are @xmath286 , @xmath287 and @xmath288 + @xmath289 respectively .
then the multiplicity of the eigenvalues is one and @xmath290 let @xmath291 .
we see that @xmath292 is not gradient observable on @xmath293 .
however , [ prop12 ] the sensor @xmath119 is gradient strategic in @xmath275 if and only if @xmath294 where @xmath295 and @xmath296 * proof . *
according to the argument above , we have @xmath297 .
it then follows that @xmath298_{1 \times 1 } $ ] and @xmath299_{1 \times 1}.$ ] let @xmath300 and @xmath301 by theorem @xmath302 , then the necessary and sufficient condition for the sensor @xmath119 to be gradient strategic in @xmath275 is that @xmath303 + the proof is complete .
let @xmath304 be a set defined by @xmath305 and for any @xmath306 , we see that + @xmath307 ^ 2}dt$ ] + defines a norm on @xmath304 provided that @xmath292 is regionally gradient observable in @xmath308 at time @xmath65 .
consider the system @xmath309,\\ \psi(\eta , t)=0~\mbox { on } \partial\omega_2 \times [ 0,b],\\ \lim\limits_{t\to 0^+}{}_0i^{1-\alpha}_{t}\psi(x , t)=0\mbox { in } \omega_2 .
\end{array}\right.\ ] ] by theorem @xmath310 , the equation @xmath311 has a unique solution in @xmath304 , which is also the initial gradient @xmath312 in @xmath313 + + * case 2 .
pointwise sensors * + in this part , we consider the problem @xmath292 with the following unbounded output function @xmath314 let @xmath315 be the location of the sensor and let @xmath316 , @xmath317 in eq @xmath318 , then one has @xmath319 since @xmath320\times[0,1]$ ] , @xmath321 is continuous and @xmath322 ( @xcite ) , we get that the assumption @xmath49 is satisfied .
further , for any @xmath323 one has @xmath324 then the assumption @xmath53 holds . by theorem @xmath302 ,
similar to proposition @xmath325 , let @xmath316 and @xmath317 , we see that , there exists a subregion @xmath275 such that the sensor @xmath326 is gradient @xmath327strategic if and only if @xmath328 can imply @xmath329 .
further , for any @xmath306 , by lemma @xmath330 if @xmath292 is regionally gradient observable , then @xmath331 ^ 2}dt\ ] ] defines a norm on @xmath304 .
consider the following system @xmath332,\\ \psi(\eta , t)=0,~~(\eta , t)\in \partial\omega_2 \times [ 0,b],\\ \lim\limits_{t\to 0^+}{}_0i^{1-\alpha}_{t}\psi(x , t)=0 , ~~x\in \omega_2 .
\end{array}\right.\ ] ] it follows from theorem @xmath310 that the equation @xmath311 has a unique solution in @xmath304 , which is also the initial gradient @xmath312 on @xmath313 + + * case 3 .
filament sensors * + consider the case where the observer @xmath333 is located on the curve @xmath334\times \{\sigma\}\subseteq \omega_2 $ ] and the output functions are @xmath335 for example , let @xmath336 .
then the example @xmath292 is not gradient observable in @xmath293 at time @xmath65 . by theorem @xmath302 ,
let @xmath337 and @xmath338 in proposition @xmath325 , we get the following results .
the sensor @xmath339 is gradient strategic in a subregion @xmath275 if and only if @xmath340 where @xmath341 and @xmath342 for all @xmath343 .
let @xmath304 be defined by @xmath344 and for any @xmath306 , consider @xmath345,\\ \psi(\eta , t)=0~\mbox { on } \partial\omega_2 \times [ 0,b],\\ \lim\limits_{t\to 0^+}{}_0i^{1-\alpha}_{t}\psi(x , t)=0\mbox { in } \omega_2 , \end{array}\right.\ ] ] where @xmath346 ^ 2}dt $ ] is a norm on @xmath304 and by theorem @xmath310 , the equation @xmath311 has a unique solution in @xmath304 and @xmath347 on @xmath313
in this paper , we investigate the regional gradient observability problem for the time fractional diffusion system with riemann - liouville fractional derivatie , which is motivated by many real world applications where the objective is to obtain useful information on the state gradient in a given subregion of the whole domain .
we hope that the results here could provided some insights into the control theoretical analysis of fractional order systems .
moreover , the results presented here can also be extended to complex fractional order dpss and various open questions are still under consideration .
for example , the problem of state gradient control of fractional order dpss , regional observability of fractional order system with mobile sensors as well as the regional sensing configuration are of great interest . for more information on the potential topics related to fractional dpss
, we refer the readers to @xcite and the references therein | this paper for the first time addresses the concepts of regional gradient observability for the riemann - liouville time fractional order diffusion system in an interested subregion of the whole domain without the knowledge of the initial vector and its gradient .
the riemann - liouville time fractional order diffusion system which replaces the first order time derivative of normal diffusion system by a riemann - liouville time fractional order derivative of order @xmath0 $ ] is used to well characterize those anomalous sub - diffusion processes .
the characterizations of the strategic sensors when the system under consideration is regional gradient observability are explored .
we then describe an approach leading to the reconstruction of the initial gradient in the considered subregion with zero residual gradient vector . at last , to illustrate the effectiveness of our results
, we present several application examples where the sensors are zone , pointwise or filament ones . ,
@xmath1 , regional gradient observability ; gradient reconstruction ; time fractional diffusion process ; strategic sensors . |
the origin of mass , or the electroweak symmetry breaking ( ewsb ) is one of the most urgent problems of the particle physics today .
there are some dynamical approaches toward this problem such as technicolor @xcite , top - quark condensate @xcite , etc .
@xcite . in the top - quark condensate or the `` top mode standard model '' ( tmsm ) ,
the scalar bound state of @xmath5 plays the role of the higgs boson in the sm and the top quark naturally acquires the dynamical mass of the order of the ewsb scale .
however , the original version of the tmsm @xcite has some problems : the 4-fermion interaction of the top - quark was introduced by hand in order to trigger the ewsb .
the top - quark mass @xmath6 was predicted somewhat larger than the experimental value , @xmath7 , even if we take the cutoff to the planck or the gut scale .
@xcite such a huge cutoff also causes a serious fine - tuning problem .
recently , arkani - hamed , cheng , dobrescu , and hall ( acdh ) @xcite proposed an interesting version of tmsm in extra dimensions , where the sm gauge bosons and some generations of quarks and leptons are embedded in @xmath8 dimensions .
since bulk gauge interactions become naively non - perturbative in the high - energy region , the bulk qcd may trigger the top - condensation without introducing ad hoc 4-fermion interactions .
moreover , all condensates of kaluza - klein ( kk ) modes of the top quark contribute to the ewsb , thereby suppressing the predicted value of @xmath6 . however , we have found that the bulk qcd coupling has an upper bound .
@xcite it is thus nontrivial whether the ewsb dynamically takes place due to the bulk qcd or not .
we thus studied the dynamics of bulk gauge theories , based on the ladder schwinger - dyson ( sd ) equation .
@xcite by this we estimated the critical value of the binding strength @xmath9 , which particularly disfavored the top - quark condensate in the @xmath10 case of the acdh version .
( see sec . 2 .
) now , in order for the acdh version to work phenomenologically , the top - condensate should be the most attractive channel ( mac ) @xcite and its binding strength @xmath11 at the cutoff @xmath12 should exceed @xmath9 , whereas those of other bound states such as the tau - condensate ( @xmath13 ) should not : @xmath14 comparing our estimations of @xmath9 with the binding strengths obeying the renormalization group equations ( rges ) of the gauge couplings , we determine the cutoff @xmath12 satisfying eq .
( [ top - cond ] ) and thereby predict the top mass as well as the higgs mass .
@xcite in fact we can obtain a certain region of the effective cutoff @xmath12 satisfying eq .
( [ top - cond ] ) for @xmath1 , while the top - condensation is unlikely to occur for @xmath10 .
( see sec . 3 .
) this is in sharp contrast to the earlier approaches of acdh @xcite and kobakhidze @xcite where the cutoff @xmath12 is treated as an adjustable parameter . in sec . 4
, we predict masses of the top quark and the higgs boson in the formulation la bardeen , hill , and lindner ( bhl ) @xcite based on rges and compositeness conditions .
the value of @xmath6 around the universal compactification scale @xmath15 is governed by the quasi infrared - fixed point ( ir - fp ) for the top - yukawa coupling @xmath16 @xcite .
the behavior of @xmath16 is approximately given by @xmath17 , where @xmath18 , @xmath19 , @xmath20 , and @xmath21 is the conventional qcd coupling .
thus , the prediction of @xmath6 can be suppressed as @xmath22 increases .
we predict numerically the top mass , @xmath23 gev , and the higgs mass , @xmath3 gev , for @xmath1 , where we took @xmath24 tev .
we consider that the sm gauge group and the third generation of quarks and leptons are put in the @xmath8-dimensional bulk , while the first and second generations are confined in 3-brane ( 4-dimensions ) .
we assume that four of @xmath25-dimensions are the usual minkowski spacetime and extra @xmath22 spatial dimensions are compactified at a universal scale @xmath26 . before analyzing the ladder sd equation
, we study running effects of dimensionless bulk gauge couplings @xmath27 @xmath28 ) . above the compactification scale @xmath15
, we should take into account contributions of kaluza - klein ( kk ) modes .
we find approximately the total number of kk modes @xmath29 below the renormalization point @xmath30 , @xmath31 with the orbifold compactification @xmath32 , where we take @xmath33 and @xmath34 projections for @xmath10 and @xmath1 , respectively .
the dimensionfull bulk gauge coupling @xmath35 is related to the conventional 4-dimensional gauge coupling @xmath36 as @xmath37 . combining the definition of @xmath38 with rges for @xmath39 and eq .
( [ n_kk ] ) , we obtain approximately rges for @xmath27 , @xmath40 with @xmath41^{-1}$ ] .
rge coefficients @xmath42 and @xmath43 are given by @xmath44 where @xmath45 @xmath46 denotes number of generations ( composite higgs bosons ) in the bulk .
hereafter , we assume @xmath47
. we note @xmath48 for @xmath49 and @xmath50 .
we find that the bulk qcd coupling with @xmath48 has the ultraviolet fixed - point ( uv - fp ) @xmath51 , @xmath52 by using eq .
( [ rge_ed3 ] ) .
we can also show that the uv - fp is the upper bound of @xmath53 .
we thus need to investigate whether the bulk qcd coupling can be sufficiently large to trigger the ewsb or not .
let us investigate the condition for the ewsb triggered by the bulk qcd . the @xmath25-dimensional ladder sd equation is given by fig .
[ fig : sd ] : @xmath54 with the mass function @xmath55 and euclidean momenta @xmath56 , where we took the landau gauge in order to make the wave function renormalization identically unity .
@xcite for simplicity , we incorporate running effects of the bulk qcd as @xmath57 , which is closely related to the mac coupling , just on the uv - fp .
this simplification obviously makes the critical point lower .
we then obtain the critical binding strength @xmath9 :
@xmath58 for @xmath10 and @xmath1 , respectively .
@xcite these are minimal values within uncertainties of the ladder sd equation .
@xcite the critical points are unlikely to be smaller than the above values , even if we take into account the effect that the cutoff @xmath12 is not so large in fact as compared with the compactification scale @xmath15 .
@xcite we use most conservatively the values of eq .
( [ kd_sd1 ] ) in the following analysis .
the upper bounds of the bulk qcd coupling in the acdh scenario are found as @xmath59 for @xmath60 .
@xcite thus , the top - condensation is unlikely to occur for @xmath10 .
we analyze the mac at the cutoff @xmath12 by using rges of bulk gauge couplings . in the one - gauge - boson - exchange approximation
, we can easily obtain @xmath61 : @xmath62 @xmath63 the mac is the top ( tau)-condensate , when the bulk qcd ( hypercharge ) is dominant .
now , we are ready to compare @xmath64 and @xmath65 with the critical values eq .
( [ kd_sd1 ] ) . unless the mac coupling exceeds at least @xmath9 estimated in eq .
( [ kd_sd1 ] ) , any condensates can not be generated in the bulk . when eq .
( [ top - cond ] ) is satisfied , the top - quark acquires the large dynamical mass , whereas the tau - lepton still remains massless .
we show our results in fig .
[ fig1 ] . actually , eq . ( [ top - cond ] )
can be satisfied for @xmath1 .
we can confirm that the top - condensation is not favored for @xmath10 .
we also note that behaviors of @xmath61 are almost unchanged for @xmath24 tev .
in the same way as the approach of bhl , we can expect to reproduce the sm in the bulk in the energy scale between @xmath15 and @xmath12 : @xmath67 with @xmath68 and the kinetic term @xmath69 for the top quark and gauge bosons .
since we find the rge for the top - yukawa coupling @xmath70 , @xmath71 , \label{rge_y}\end{aligned}\ ] ] and that for the higgs - quartic coupling @xmath72 , @xmath73 , \label{rge_lam}\end{aligned}\ ] ] we can predict @xmath6 and @xmath66 by using compositeness conditions @xcite , @xmath74 since the running effect of the qcd coupling @xmath21 is almost negligible around @xmath15 , the top - yukawa coupling in @xmath75 is attracted toward the quasi ir - fp @xmath16 @xcite , whose behavior is approximately found as @xmath76 neglecting effects of the electroweak interactions .
this value obviously decreases as @xmath22 increases . as a result
, the problem of @xmath77 gev in @xmath78 is suppressed in our scenario .
now , we predict @xmath6 and @xmath66 .
( see figs .
[ fig1 ] and [ fig6 ] . )
we obtain numerically the top - quark mass @xmath6 as @xmath79 and the higgs boson mass @xmath66 as @xmath80 for @xmath1 , where we took @xmath24 tev . for details , see ref . @xcite .
we have investigated the idea that the ewsb dynamically occurs due to the top - condensation in the bulk , with emphasis on the dynamics of bulk gauge theories .
we estimated the critical binding strength @xmath9 , based on the ladder sd equation .
we also analyzed the mac by using rges for bulk gauge couplings . combining our mac analysis with @xmath9
, we showed that the top - condensation is favored for @xmath1 , while it is not for @xmath10 .
we have predicted @xmath6 and @xmath66 in the approach of bhl .
we then obtained @xmath2 gev , and @xmath81 gev for @xmath1 .
y. nambu , enrico fermi institute report no .
89 - 08 , 1989 ; in _ proceedings of the 1989 workshop on dynamical symmetry breaking _ , edited by t. muta and k. yamawaki ( nagoya university , nagoya , japan , 1990 ) .
m. hashimoto , prog .
. phys . * 100 * , 781 ( 1998 ) .
n. arkani - hamed , h. c. cheng , b. a. dobrescu and l. j. hall , phys . rev . *
d62 * , 096006 ( 2000 ) .
m. hashimoto , m. tanabashi and k. yamawaki , phys .
rev . * d64 * , 056003 ( 2001 ) .
v. gusynin , m. hashimoto , m. tanabashi and k. yamawaki , phys . rev . *
d65 * , 116008 ( 2002 ) .
s. raby , s. dimopoulos and l. susskind , nucl .
b169 * , 373 ( 1980 ) .
m. hashimoto , m. tanabashi and k. yamawaki , in preparation . | we study the dynamical electroweak symmetry breaking ( dewsb ) in the @xmath0-dimensional bulk with compactified extra dimensions . we identify the critical binding strength for triggering the dewsb , based on the ladder schwinger - dyson equation . in the top mode standard model with extra dimensions , where the standard model gauge bosons and the third generation of quarks and leptons are put in the bulk
, we analyze the most attractive channel ( mac ) by using renormalization group equations ( rges ) of ( dimensionless ) bulk gauge couplings and determine the effective cutoff where the mac coupling exceeds the critical value .
we then find that the top - condensation can take place for @xmath1 .
combining rges of top - yukawa and higgs - quartic couplings with compositeness conditions , we predict the top mass , @xmath2 gev , and the higgs mass , @xmath3 gev , for @xmath1 , where we took the universal compactification scale @xmath4 tev .
presented at + _ 2002 international workshop on _ + _ " strong coupling gauge theories _ + _ and effective field theories " ( scgt 02 ) _ + _ nagoya , japan _ + _
1013 december 2002 _ |
the purpose of the work we present here is to obtain a model for the membrane potential of a single cell which is reasonably realistic , and yet so simple that it can be used in practice to simulate numerically single cells or several coupled cells . for simplicity ,
experimentally observed currents ( boyett et al . 1993 ) are omitted if they seem too small to have a significant influence on the intracellular ion concentration , or at least too small to change the dynamics of the cell . on the other hand
, we try to make the theory realistic by using equations that are compatible with , or can be derived from , basic physical principles .
it is a basic assumption of most models ( wilders 1993 ) for the electrical activity of cells that only the motion of positive ions , and specifically those of potassium , calcium and sodium , influence the membrane potential .
this assumption is usually expressed as a differential equation for the time dependence of the potential .
we observe that this differential equation can be integrated exactly , and argue that the integration constant is given by the requirement that the potential is zero when the ion concentrations on both sides of the membrane are equal , as the density of negative charge happens to be the same on both sides .
then it follows that the potential is directly proportional to the excess number of positive ions inside the cell , a formula which is nothing but the one for an electric capacitance that follows from gauss s law in electrostatics .
we derive equations for ionic currents flowing through channels , exchangers and electrogenic pumps .
these are based on the boltzmann distribution law ( boltzmann 1868 ) , which states that a particle in thermal equilibrium spends less time in states of higher energy than in states of lower energy , the markov assumption ( markov 1906 ) which says that the transition probabilities of a stochastic system ( of markov type ) is only dependent on its present state , and the principle of detailed balance ( onsager 1931 ) which says that the microscopic laws of physics are invariant with respect to the reversal of time .
our equations were inspired by ehrenstein and lecar s model of channel gating ( 1977 ) , nonner and eisenberg s model for channel current ( 1998 ) , mullins model of the @xmath0 exchanger ( 1977 ) , and chapman s model of the @xmath1 pump ( 1978 ) .
in particular the book of lorin john mullins ( 1981 ) `` ion transport in heart '' has been a major source of inspiration for us .
the theory is illustrated with a simple model of spontaneously active cells in the rabbit sinoatrial node .
the observable parameters in the model are based on the experiments of shibasaki ( 1987 ) , hagiwara et al .
( 1988 ) , muramatsu et al .
( 1996 ) and sakai et al .
the non observable parameters in the model are determined numerically , in the same way as in an earlier study ( endresen 1997a ) , by comparing the action potentials generated by the model with the shape of the action potentials recorded by baruscotti et al .
( 1996 ) . by using an algebraic equation for the potential in place of the standard differential equation ,
as mentioned above , we obtain a model which is stable against a slow drift of the intracellular ion concentrations , sometimes seen in other models .
furthermore , by fixing the integration constant for the voltage we obtain from the model a prediction of the steady state ion concentrations in the cell .
it is even possible to predict these steady state concentrations by starting with an initial state having equal concentrations inside and outside the cell , and integrating the equations of motion over a long time interval . from the equations of motion we obtain an equation that explicitly demonstrates the energy balance in the process of moving ions in and out of the cell .
the energy to make the cell function comes from breakdown of atp that runs the @xmath2 pump .
part of this free ( or useful ) energy is dissipated while the rest enters the cell . in the cell
some of this energy is used to create a potential energy that depends upon the ionic concentrations in the cell , while the rest is dissipated by the currents in the ionic channels and the @xmath0 exchanger .
this potential energy function is thus such that the work associated with ionic currents balances exactly the change in potential energy . in a numerical integration of the differential equations
one may compute separately the work and potential energy , comparing the two in order to check ( and maybe control ) the accuracy of the numerical integration . in our long time integration
we observe indeed a balance between work and change in cell membrane potential energy .
there are two basic principles behind the average motion of particles .
the first is diffusion , which is general ; the second applies only to charged particles such as ions in solutions .
simple diffusion is described by the empirical law of fick ( 1855 ) , @xmath3 \;,\ ] ] where @xmath4 is the ionic flux , @xmath5 $ ] the concentration of ions and @xmath6 the ratio of the velocity to the force acting on a particle , known as the mobility .
@xmath7 is the absolute temperature and @xmath8 is boltzmann s constant .
the empirical law of ohm ( 1827 ) describes the net motion of charged particles in an electric field , @xmath9 \nabla u \;,\ ] ] where @xmath10 is the valence , @xmath11 the elementary charge and @xmath12 the electrical potential .
since we assume that the temperature is constant , we can neglect the thermal flux given by fourier s empirical law .
the fact that the mobility in fick s law must be identical to the mobility in ohm s law was first noticed by einstein ( 1905 ) . if we combine eqs .
( [ eq1 ] ) and ( [ eq2 ] ) , the total flux of ions due to diffusion and electric forces is @xmath13 \exp\left(\frac{zeu}{kt}\right)\right ] \;.\ ] ] the equilibrium potential for which the flux is zero , is @xmath14_{\rm e}}{[{\rm s}]_{\rm i}}\right)\;.\ ] ] it can be found by setting @xmath15 in eq .
( [ eq3 ] ) and integrating from the extracellular ( e ) to the intracellular ( i ) side of the membrane .
here @xmath16 , @xmath17 , @xmath5_{\rm i}$ ] and @xmath5_{\rm e}$ ] are the intracellular and extracellular potentials and concentrations .
this equation , first stated by nernst ( 1888 ) is based only on the empirical laws of ohm and fick and the relation of einstein .
the same formula can be derived in a more general way using the boltzmann factor ( boltzmann 1868 ) . the relative probability at equilibrium that an ion is at the intracellular or extracellular side of a cell membrane is @xmath18_{\rm i}}{[{\rm s}]_{\rm e } } = \exp\left(-\frac{ze(u_{\rm i } - u_{\rm e})}{kt}\right ) \;,\ ] ] where @xmath19 is the energy difference between the two positions of the ion .
solving ( [ eq5 ] ) for @xmath20 gives ( [ eq4 ] ) . with the definition @xmath21
the equilibrium potentials for the predominant cellular cations are then @xmath22_{\rm e}}{[{\rm k}]_{\rm i } } \ ; , \\
\label{eq8 } v_{\rm ca } & = & \frac{v_t}{2 } \ln \frac{[{\rm ca}]_{\rm e}}{[{\rm ca}]_{\rm i } } \ ; , \\ \label{eq9 } v_{\rm na } & = & v_t \ln \frac{[{\rm na}]_{\rm e}}{[{\rm na}]_{\rm
i } } \;.\end{aligned}\ ] ] imagine that ionic channels are either completely open or completely closed and randomly fluctuate between these states in a simple markov process ( markov 1906 ) , described by the first order kinetics ( ehrenstein and lecar 1977 ) @xmath23 where the rate constants @xmath24 and @xmath25 are functions of transmembrane voltage and control the transitions between the closed ( @xmath26 ) and the open ( @xmath27 ) states of the gate .
the rate for a closed channel to open is @xmath24 , and @xmath25 is the rate for an open channel to close .
let @xmath28 denote the average fraction of channels that are open , or , equivalently , the probability that a given channel will be open .
we may say that the ionic flux through an ensemble of channels is regulated by a sliding door whose position is @xmath28 .
this yields : @xmath29 where @xmath30 here @xmath31 denotes the steady state fraction of open channels and @xmath32 the relaxation time .
let us assume that the energy difference between the open and closed positions is given by @xmath33 where @xmath34 is a gating charge , usually @xmath35 , such that @xmath36 represents the change in electrical potential energy due to the redistribution of charge during the transition , and where the term @xmath37 represents the difference in mechanical conformational energy between the two states . at equilibrium , @xmath38 , and
the ratio of the probabilities for a single channel to be in the open state or the closed state is @xmath39 this relation is known as the principle of detailed balance ( onsager , 1931 ) .
the same ratio is given by the boltzmann distribution ( boltzmann 1868 ) , @xmath40 thus , from eqs .
( [ eq14 ] ) , ( [ eq15 ] ) , and ( [ eq16 ] ) , with @xmath41 @xmath42^{-1 } \;.\ ] ] the simplest possible choice for @xmath43 and @xmath44 is @xmath45 where @xmath46 is a constant . assuming @xmath46 to be constant gives the maximum relaxation time at the voltage where @xmath47 .
the relaxation time as a function of @xmath48 is then @xmath49}^{-1}\;.\ ] ] here we want to obtain the current @xmath50 through a one dimensional ionic channel of length @xmath51
. we will allow the cross sectional area @xmath52 to vary with position , i.e. , we take @xmath53 . by definition , @xmath54 is the inside and @xmath55 the outside of the membrane .
let @xmath56 be the @xmath28-component of the flux @xmath57 , the other components are negligible as long as the variation of @xmath52 with @xmath28 is smooth and slow .
this is the analogue of water flow in a pipe of varying cross section . by stationary flow
, the current @xmath50 must be the same through all cross sections , i.e. independent of @xmath28 .
thus the flux @xmath58 is inversely proportional to the area @xmath52 , by the relation @xmath59 we insert @xmath58 from this equation in the @xmath28 component of eq .
( [ eq3 ] ) , and multiply the resulting equation by @xmath60 , introducing a constant voltage @xmath61 chosen such that @xmath62 then we find that @xmath63 \exp\left(\frac{ze(u - u_0)}{kt}\right)\right ] \;.\ ] ] here @xmath12 , @xmath5 $ ] and @xmath52 are functions of @xmath28 , while all other quantities are constant .
( note however that the mobility @xmath6 may be reduced in a very narrow channel ; one possible way to take into account such an @xmath28 dependence of @xmath6 is to replace the area @xmath52 by an effective area @xmath64 which is smaller than @xmath52 ) .
integrating from the inside @xmath54 to the outside @xmath55 we obtain @xmath65_{\rm e } \exp\left(-\frac{zev}{2kt}\right ) - [ { \rm s}]_{\rm i } \exp\left(\frac{zev}{2kt}\right ) \right ] \;,\ ] ] where @xmath66 the concentrations are @xmath5_{\rm i}$ ] on the inside and @xmath5_{\rm e}$ ] on the outside . if we extract a factor @xmath67_{\rm e } [ { \rm s}]_{\rm i}}$ ] , and write the ratio of the concentrations in terms of the nernst potential defined in eq .
( [ eq4 ] ) , eq .
( [ eq24 ] ) can be written in the following way , @xmath68_{\rm e } [ { \rm s}]_{\rm i } } \left [ \sqrt{\frac{[{\rm s}]_{\rm i}}{[{\rm s}]_{\rm e } } } \exp\left(\frac{zev}{2kt}\right ) -
\sqrt{\frac{[{\rm s}]_{\rm e}}{[{\rm s}]_{\rm i } } } \exp\left(-\frac{zev}{2kt}\right ) \right ] \nonumber \\ & = & \frac{2zeukt}{i}\ ; \sqrt{[{\rm s}]_{\rm e } [ { \rm s}]_{\rm i } } \sinh\left(\frac{ze(v - v_{\rm s})}{2kt}\right ) \;.\end{aligned}\ ] ] eq .
( [ eq26 ] ) is our general result that follows from the combined ohm s and fick s law .
now the integral @xmath69 depends upon both the voltage @xmath70 and the cross section @xmath53 . to determine
the @xmath28 dependence of @xmath71 one would need poisson s equation for the electrostatic potential , taking into account the net charge distribution in the membrane , including both positive and negative ions .
however , this charge distribution will depend upon detailed properties of membranes and their channels that have been little known so far .
thus it seems a reasonable approach to make certain assumptions directly about @xmath71 .
a commonly used assumption is that @xmath71 is linear , i.e. that the electric field @xmath72 is constant , and that the cross section is constant , @xmath73 .
( [ eq26 ] ) takes the form @xmath74_{\rm e } [ { \rm s}]_{\rm i}}\ ; \frac{a_0v\sinh\left(\frac{ze(v - v_{\rm s})}{2kt}\right ) } { d\sinh\left(\frac{zev}{2kt}\right ) } \;.\ ] ] as should be expected , this relation simplifies to the usual ohm s law in the special case @xmath5_{\rm i}=[{\rm s}]_{\rm e}$ ] , since then @xmath75 by eq .
( [ eq4 ] ) .
( [ eq27 ] ) is known as the goldman constant field approximation .
goldman ( 1943 ) wrote : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we assume that the membrane contains a large number of dipolar ions near the isotonic point and that these can act to minimize distortion in the field especially at low currents .
we then approach a situation in which the field is constant and are led to a solution analogous to that given by mott ( 1939 ) for electronic conduction in the copper copper oxide rectifier . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ a more general case , perhaps somewhat more realistic , where the integral @xmath69 can still be calculated exactly , is that of an ion channel having a constant area @xmath76 , except for a short and narrow constriction or pore in its middle , with an area @xmath77 much smaller than @xmath76 .
an example is a cylindrical pore of radius @xmath78 and length @xmath79 , which is typical for ionic channels .
if we furthermore assume a constant electric field everywhere in the channel , and if the length of the pore is @xmath80 , then we have that @xmath81\;.\ ] ] the limit of this as @xmath82 is @xmath83 \approx \frac{\epsilon d}{a_p}\;.\ ] ] the last approximation holds when the contribution from the pore dominates the integral , which will be the case e.g. when the ratio of areas , @xmath84 , is of the order @xmath85 . for @xmath86 small but nonzero
the @xmath48 dependence of @xmath69 is only of second order in @xmath86 , thus it will be a good approximation in a finite voltage range to take @xmath69 to be constant , equal to @xmath87 .
the approximation @xmath88constant which is also valid under more general conditions than those assumed in the above oversimplified derivation , and it gives @xmath89 here @xmath90 is independent of @xmath48 , e.g. in the case considered above , @xmath91_{\rm e } [ { \rm s}]_{\rm i}}\ ; \frac { a_{\rm p}}{\epsilon d } \;.\ ] ] for na and k ions it is a good approximation to consider the square root of the concentrations @xmath67_{\rm e } [ { \rm s}]_{\rm i}}$ ] constant , while for ca ions the relative change in concentration is more significant during one action potential . in the present work we used @xmath92constant in all three cases , for simplicity . we have checked that this does not affect our numerical results significantly .
there is reason to ask whether the linear voltage profile @xmath71 can be a reasonable approximation in the presence of a pore .
indeed , it might seem natural to conclude that most of the voltage drop must be concentrated at the pore due to its large resistance .
however , with the combined ohm s and fick s law , the current is driven by gradients in both voltage and concentration , as follows from eq .
( [ eq3 ] ) .
a large current may be due to a large voltage drop over the pore , or it may be due to a large change in concentration , without any large voltage difference .
thus , in general one has to take into account the detailed properties of the channel in order to see which one of the gradients is the dominant driving force in a given situation . in a recent investigation by nonner and eisenberg ( 1998 ) , poisson s equation relating the net charge density and electrostatic potential
was included in a more extensive analysis for a specific model of a channel with a narrow pore . in their analysis
they indeed find that only part of the voltage drop is across the pore ( something like half of it ) . in their numerical simulations the voltage in the pore
is dominated by the presence of charged carboxyl groups , and thus almost independent of the transmembrane voltage .
this lends support to the approximation used here , that the integral @xmath69 , eq .
( [ eq24b ] ) , can be regarded as being constant .
thus our simple result ( [ eq29 ] ) has the characteristic features of the current voltage relationships obtained by nonner and eisenberg in their more extensive investigation .
one characteristic feature is that eq .
( [ eq29 ] ) shows inward rectification for large values of @xmath93_{\rm e}/[{\rm s}]_{\rm i}$ ] , i.e. increased conductance for large negative potentials .
indeed the curves in figure 3a in nonner and eisenberg ( 1998 ) have shapes of a hyperbolic sine .
such a behavior is not predicted by the goldman ( 1943 ) equation , eq .
( [ eq27 ] ) , but is seen in many excitable cells ( hille 1992 ) .
this is another reason to base our computations on eq .
( [ eq29 ] ) in order to see the consequences of its application . if the flux of ions is given by eq .
( [ eq29 ] ) and regulated by the fraction of open channels @xmath28 , the membrane current through potassium channels is where @xmath94 is the maximum value of the relaxation time , @xmath95 is the conductance parameter of eq .
( [ eq30 ] ) , @xmath96 is given by eq .
( [ eq7 ] ) , and the time dependence of @xmath28 is given by eq .
( [ eq11 ] ) with eqs .
( [ eq17 ] ) and ( [ eq21 ] ) for @xmath31 and @xmath97 respectively . here
we have used the identity @xmath98 the calcium and sodium channels have an inactivation mechanism in addition to the above activation mechanism .
we can view these mechanisms as two independent markov processes , or equivalently two independent sliding doors , which are both affected by voltage .
an ion can only go through if both sliding doors are at least slightly open . here
the activation mechanism is very fast , with a time constant of only a few milliseconds , so we use the steady state fraction of open channels , eq .
( [ eq17 ] ) , for this .
the maximum time constant of inactivation for calcium and sodium channels are of the same order of magnitude as the maximum time constant of the activation of the potassium channel ( typically a few hundred milliseconds ) , thus and , where @xmath99 and @xmath100 are the conductance parameters of the calcium and sodium currents respectively , @xmath101 and @xmath102 are given by eqs .
( [ eq8 ] ) and ( [ eq9 ] ) , @xmath103 and @xmath104 are the half
activation potentials , and @xmath105 and @xmath106 are the half inactivation potentials .
note that the activation and inactivation mechanisms work in the same way , and differ in two respects only .
first , the time constants differ experimentally by two orders of magnitude , and second , the gating charge @xmath34 , eq .
( [ eq14 ] ) , is @xmath107 in one case and @xmath108 in the other case .
the na , k atpase is found in the plasma membrane of virtually all animal cells and is responsible for active transport of sodium and potassium .
low sodium concentration and high potassium concentration in the cytosol are essential for basic cellular functions such as excitability , secondary active transport , and volume regulation . in our model ,
the @xmath2 pump is the only energy source .
we shall assume that the following equation is a complete macroscopic description of the pump reaction ( chapman 1978 ) , @xmath109 where atp , adp and @xmath110 are adenosine triphosphate , adenosine diphosophate and inorganic phosphate , while @xmath43 and @xmath44 are the rates for the forward and backward reactions .
the energy involved in the movement of 3 @xmath111 and 2 @xmath112 ions against their electrochemical gradients is @xmath113 where @xmath96 and @xmath114 are given by eqs .
( [ eq7 ] ) and ( [ eq9 ] ) .
this result is independent of the detailed interaction between ions , molecules and the atpase enzyme .
therefore , the total change in gibbs free energy is @xmath115 where @xmath116 is the energy associated with the breakdown of atp , and @xmath117 .
note that @xmath118 has to be negative , at least when averaged over time , but the sum @xmath119 may very well be positive , since @xmath116 is large and negative .
thus , part of the energy from atp breakdown goes into increasing the free energy of potassium and sodium ions , but much energy is dissipated , since the energy available is actually much larger than the energy required to translocate the potassium and sodium ions at small negative membrane potentials . in practice , such a pump or motorized swing door will quickly reach saturation .
we therefore choose the sum of the forward and backward rates to be constant , resembling the maximum possible speed of the swing door in the forward and backward directions , @xmath120 where @xmath46 is a constant . at equilibrium ,
the forward reaction must occur just as frequently as the reverse reaction , giving @xmath121 solving eqs .
( [ eq39 ] ) and ( [ eq40 ] ) for @xmath43 and @xmath44 gives @xmath122 the difference @xmath123 gives the net pump current for a cell with @xmath124 pumps as @xmath125 where @xmath126 . to maintain a steady state for the intracellular calcium concentration in for example heart cells , the amount of calcium that enters the cell via ionic channels must be extruded .
the @xmath0 exchanger is the major mechanism responsible for achieving a balance between calcium entry and extrusion in oscillating cells .
we assume that the rate for the forward ( @xmath43 ) and the backward ( @xmath44 ) exchange reaction given by ( mullins 1977 ) @xmath127 are governed largely by the electrochemical gradients for sodium and calcium , together with the membrane potential .
in other words , the energy produced when 3 extracellular sodium ions take the elevator down into the cytosol is used to elevate one calcium ion up from the cytosol into the extracellular space , @xmath128 where @xmath101 and @xmath114 are given by eqs .
( [ eq8 ] ) and ( [ eq9 ] ) .
the total work done in reaction ( [ eq45 ] ) is @xmath129 the ratio of @xmath43 to @xmath44 in eq .
( [ eq45 ] ) is again determined by @xmath118 like in eq .
( [ eq40 ] ) . however , in the present case saturation effects are not expected and furthermore @xmath118 will vary around zero , so we put @xmath130 where we make the assumption that @xmath46 is a constant ( mullins , 1981 ) .
for a cell with @xmath131 exchangers the net current is then @xmath132 where @xmath133 .
imagine that the electrical activity of a cell is described by the five currents discussed above , and that all the other currents ( boyett 1996 ) are of minor importance .
the standard differential equations for the voltage , and the conservation laws for intracellular ionic concentrations are then @xmath134_{\rm i } } & = & \frac { 2i_{\rm nak } - i_{\rm k}}{fv}\ ; , \\
\label{eq54 } \frac{d}{dt}{[\rm ca]_{\rm i } } & = & \frac { 2{i_{\rm naca}}- i_{\rm ca}}{2fv}\ ; , \\ \label{eq55 } \frac{d}{dt}{[\rm na]_{\rm i } } & = & \frac { -{i_{\rm na}}- 3i_{\rm nak}- 3i_{\rm naca}}{fv}\;,\end{aligned}\ ] ] where @xmath26 is cell capacitance , @xmath135 is faraday s constant , and we assume that the cell volume @xmath136 is constant . now eqs . ( [ eq53 ] ) , ( [ eq54 ] ) and ( [ eq55 ] ) can be solved for @xmath137 , @xmath138 , and @xmath139 , and we obtain @xmath140_{\rm i } } + 2i_{\rm nak } \ ; , \\
\label{eq57 } i_{\rm ca } & = & -2fv \frac{d}{dt}{[\rm ca]_{\rm i } } + 2i_{\rm naca } \ ; , \\
\label{eq58 } i_{\rm na } & = & -fv \frac{d}{dt}{[\rm
na]_{\rm i } } - 3i_{\rm nak}- 3i_{\rm naca } \;.\end{aligned}\ ] ] inserting this into eq .
( [ eq52 ] ) yields @xmath141_{\rm i}}+2{[\rm ca]_{\rm i}}+{[\rm na]_{\rm i } } \right)\;,\ ] ] since the remaining currents cancel .
this equation can also be written as @xmath142_{\rm i } + 2[\rm ca]_{\rm i } + [ \rm
na]_{\rm i } \right\ } \right ) = 0\;.\ ] ] this integrated gives @xmath143_{\rm i}+2[\rm ca]_{\rm i}+[\rm na]_{\rm i } \right ) = v_0 \;,\ ] ] where the integration constant @xmath144 has to be determined .
given that the voltage across a capacitor is zero when the net charge difference is zero , we must choose the integration constant @xmath145_{\rm e } + 2[\rm ca]_{\rm e } + [ \rm na]_{\rm e } \right ) \;,\ ] ] which gives @xmath146_{\rm i}-[\rm k]_{\rm e } + 2([\rm ca]_{\rm i}-[\rm ca]_{\rm e } ) + [ \rm na]_{\rm i}-[\rm na]_{\rm e } \right\ } \;.\ ] ] this choice of @xmath144 depends on the assumption that all other ions have the same concentrations on both sides , consistent with eq .
( [ eq52 ] ) where it is assumed that they do not contribute to the current .
this is also consistent with standard assumptions in the literature ( encyclopdia britannica 1997 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in the extracellular fluid , electroneutrality is preserved by a balance between a high concentration of @xmath147 on the one hand and a high concentration of @xmath148 as well as small quantities of impermeant anions such as bicarbonate , phosphate , and sulfate on the other . in the cytoplasm , where @xmath149 concentration is high ,
the concentration of @xmath148 is much below that necessary to balance the sum of the positive charges .
electroneutrality is maintained there by negatively charged impermeant proteins and phosphates .
osmotic balance is maintained between the extracellular fluid and the cytoplasm by movement of water through the plasma membrane when the total concentration of particles on one side is not equal to that on the other .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ eq . ( [ eq63 ] ) is nothing but the relation between electric potential and charge of a capacitor , which is actually the origin of eq .
( [ eq52 ] ) .
thus it is completely general and independent of the number of membrane currents in a model .
it means that : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the voltage across the membrane of a cell is caused by , and is directly proportional to , the surplus of charge inside the cell .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ since eq .
( [ eq63 ] ) is the explicit integral of eq .
( [ eq52 ] ) , it can be used instead of eq .
( [ eq52 ] ) ( or the equivalent of eq .
( [ eq52 ] ) ) in any model .
the differential equation , eq .
( [ eq52 ] ) , is needed only in models where the intracellular ionic concentrations are not tracked individually ( for example , the hodgkin
huxley equations ( 1952 ) ) .
there is a significant difference between eqs .
( [ eq52 ] ) and ( [ eq63 ] ) for use in numerical simulations , for the following reason .
there are two different ways to determine how many ions there are inside a cell .
the first method counts every ion entering or leaving ( eq .
( [ eq52 ] ) ) , while the second method counts all the ions inside the cell ( eq . ( [ eq63 ] ) ) .
both methods will give correctly the _ variation _ in the number of ions in the cell .
however , the observer of ions entering and leaving observes only the variations in the number , and if he wants to know the actual number , he must make an initial guess of the number of ions already inside . because his guess may differ significantly from the actual ion number , the results from the two methods may be contradictory . a variant of eq .
( [ eq63 ] ) has recently been derived by varghese and sell ( 1997 ) .
however , they did not identify the integration constant @xmath144 , which is related to the initial ionic concentrations and represents the initial net charge via the electric capacitance of the cell as shown in eq .
( [ eq63 ] ) .
there is reason to ask whether it is a reasonable approximation to omit the anions in eq .
( [ eq63 ] ) . this can be justified if the total concentration of cations is approximately the same on both sides .
indeed , this property is seen in most ionic models , like for instance in wilders ( 1993 ) , where the cation concentrations are @xmath150_{\rm e } = 5.4 \ , { \rm mm } & \quad [ { \rm ca}]_{\rm e } = 2 \ , { \rm mm } & \quad [ { \rm na}]_{\rm e } = 140 \ , { \rm mm } \\ { [ { \rm k}]}_{\rm i } = 140 \ , { \rm mm } & \quad [ { \rm ca}]_{\rm i } = 0.0000804 \ , { \rm mm } & \quad [ { \rm na}]_{\rm i } = 7.5 \ , { \rm mm } \;.\\ \end{array}\ ] ] the current @xmath50 in eq .
( [ eq22 ] ) may be written as @xmath151 where @xmath152 is the number of ions transferred from the inside to the outside of the membrane , and @xmath153 is the rate of transfer . the change in free energy when one ion is transferred , is @xmath154 , and the total change in free energy over a time interval is @xmath155 it follows from eq .
( [ eq29 ] ) that the current @xmath50 has the same sign as the voltage @xmath156 as required in general to have thermodynamic stability , so that the integrand in eq .
( [ eq64a ] ) is positive ( strictly speaking non - negative ) , and therefore @xmath157 ( or @xmath158 ) . by similar reasoning , taking into account all the reversal potentials and the free energy associated with the breakdown of atp
, we find that the total change in free energy due to the five currents in our model is @xmath159 dt\;.\end{aligned}\ ] ] each of the five terms in the integrand is positive ( non - negative ) , since each current has the same sign as the corresponding voltage .
in other words , energy is dissipated all the time by all the five currents , implying that @xmath158 . the main contribution to
the negative @xmath118 is the atp term , @xmath160 which in practice is negative all the time , and furthermore is large in magnitude compared to the other terms .
it should be noted that this term is the source of useful energy that is dissipated to maintain the activity of the cell and keep it away from equilibrium . keeping this term apart
, we may calculate the change in free energy of the ionic system , @xmath161 dt \;.\end{aligned}\ ] ] using eqs .
( [ eq52 ] ) , ( [ eq56 ] ) , ( [ eq57 ] ) , and ( [ eq58 ] ) to eliminate the currents , and assuming the capacitance @xmath26 and the volume @xmath136 to be constant , we get that @xmath162_{\rm e}}^{[\rm k]_{\rm i } } v_{\rm k}\,d([{\rm k}]_{\rm i } ) \nonumber\\ & & -2fv \int_{[\rm ca]_{\rm e}}^{[\rm ca]_{\rm i } } v_{\rm ca}\,d([{\rm ca}]_{\rm i } ) -fv \int_{[\rm na]_{\rm e}}^{[\rm na]_{\rm i } } v_{\rm na}\,d([\rm na]_{\rm i})\;,\end{aligned}\ ] ] where the reversal potentials @xmath96 , @xmath101 , and @xmath102 depend on the integration variables @xmath163_{\rm i}$ ] , @xmath164_{\rm i}$ ] , and @xmath165_{\rm i}$ ] according to eqs .
( [ eq7 ] ) , ( [ eq8 ] ) and ( [ eq9 ] ) . integrating from the equilibrium state @xmath166 , @xmath163_{\rm i}=[\rm k]_{\rm e}$ ] , @xmath167_{\rm i}=[\rm ca]_{\rm e}$ ] , and @xmath165_{\rm i}=[\rm na]_{\rm e}$ ] , using the indefinite integral @xmath168 we find in passing it may be noted that in the present case the equilibrium state is simply the one with equal concentrations of cations on both sides , as follows from our assumption of having the same concentration of anions or negative charge on both sides of the membrane .
more generally equilibria for ionic systems are described by the donnan ( 1911 ) equilibrium that can yield different concentrations on both sides of the membrane .
since @xmath169 is a function only of the state of the cell and is independent of the process by which the state is reached , it represents a potential energy for the cell , which we will call @xmath170 .
note that @xmath171 in the equilibrium state , whereas @xmath172 in all other states .
@xmath170 is the minimum energy needed to bring a thermal system away from equilibrium with its surroundings , or equivalently , the maximum work that can be performed by the system when returning to equilibrium .
the potential energy @xmath170 , as defined in eq .
( [ eq68 ] ) , contains three terms , each of which can be given a more direct physical interpretation .
the first term is simply the electrostatic energy of a capacitor , while the two temperature dependent terms are related to thermal properties .
in fact , since we assume ideal dilute solutions , the change in entropy due to changes of ion concentrations away from their equilibrium values is @xmath173_{\rm i}\ , \ln \left({\frac { [ \rm k]_{\rm e}}{[\rm k]_{\rm i}}}\right ) + [ \rm ca]_{\rm i}\ , \ln \left({\frac{[\rm ca]_{\rm e}}{[\rm ca]_{\rm i}}}\right ) + [ \rm na]_{\rm i}\ , \ln\left({\frac { [ \rm na]_{\rm
e}}{[\rm na]_{\rm i}}}\right)\right\}\;.\ ] ] under the same changes , the change in osmotic pressure inside the cell is equal to the difference in osmotic pressure across the membrane , which is , for ideal solutions , @xmath174_{\rm i}-[\rm k]_{\rm e } + [ \rm na]_{\rm i}-[\rm na]_{\rm e } + [ \rm ca]_{\rm i}-[\rm ca]_{\rm e } \right)\;.\ ] ] note that for fixed volume the anions will not contribute to the difference in osmotic pressure , but they will contribute if the volume is changed and the membrane is impermeable to them . in terms of the change in transmembrane voltage , @xmath48 , the change in entropy , @xmath175 , and the change in transmembrane osmotic pressure , @xmath176 , as compared to the equilibrium state , we may write @xmath177 equation ( [ eq79 ] ) is the vant hoff equation ( 1887 ) for the osmotic pressure across a solute impermeable barrier separating two ideal dilute solutions . in 1887
vant hoff noticed that the behavior of solutes in dilute solutions resembles the behavior of a perfect gas ( vant hoff , 1887 ) , and as quoted by arrhenius in a memoir edited by jones ( 1899 ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the pressure which a gas exerts at a given temperature if a definite number of molecules is contained in a definite volume , is equal to the osmotic pressure which is produced by most substances under the same conditions , if they are dissolved in any given liquid . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ rewriting eq .
( [ eq68 ] ) , using eqs .
( [ eq65ions ] ) and ( [ eq77 ] ) , we may summarize the energy balance in the following way , @xmath178dt&&.\end{aligned}\ ] ] the left hand side of this equation is the useful work performed upon the cell by the @xmath2 pumps , moving @xmath179 and @xmath149 ions against their potential gradients .
the energy supplied by the pumping of ions produces the following effects that either change the potential energy of the cell or cause energy loss by dissipation , 1 . a transmembrane voltage difference , @xmath48 ; 2 . a change in entropy , @xmath175 ; 3 . a transmembrane osmotic pressure difference , @xmath176 ; and 4 .
downhill ionic currents through the exchangers and channels , @xmath137 , @xmath180 , @xmath139 , and @xmath181 . in an oscillating cell , as described by the present model , the following two inequalities will hold , over a sufficiently long time interval , @xmath182 the first inequality is simply the inequality @xmath183 , which follows from eq .
( [ eq44 ] ) .
it means that the energy released by breakdown of atp is larger than the useful work performed by the pumps , as required from general principles , in other words , that energy is dissipated by the current @xmath184 produced by the pumps .
the second inequality must hold due to eq .
( [ eq80 ] ) , where the right hand side consists of three oscillating potential energy terms plus four positive terms that describe energy dissipation .
this inequality shows that the useful work performed by the pumps is positive , as required to maintain the dissipation due to the other currents of a working cell away from thermal equilibrium .
we may remark that the laws of ohm and fick , equations ( [ eq1 ] ) and ( [ eq2 ] ) , are consistent with the use of ideal solutions and osmotic pressure that assume independent ( non interacting ) particles .
arrhenius ( 1902 ) wrote about the relationship between osmotic pressure and diffusion : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ besides the electrical , other forces may be active in causing the movement of the ions .
of these the osmotic pressure is the most important . on account of this pressure a phenomenon called diffusion ( hydrodiffusion )
may be observed .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in the model considered the osmotic pressure @xmath176 has not been involved in the dynamics .
however in an extended model with variable volume @xmath136 it will be more important as it will determine the solute flux through the membrane .
in the above , a mathematical model of the membrane potential has been derived where eqs .
( [ eq7 ] ) , ( [ eq8 ] ) , and ( [ eq9 ] ) represent the equilibrium potentials , eqs .
( [ eq31 ] ) , ( [ eq33 ] ) , and ( [ eq34 ] ) the ionic currents , eqs .
( [ eq51 ] ) and ( [ eq44 ] ) the exchanger and the pump currents , eqs .
( [ eq53 ] ) , ( [ eq54 ] ) and ( [ eq55 ] ) the ionic concentrations , eq . ( [ eq63 ] ) the membrane voltage , and finally , eq . ( [ eq77 ] ) the osmotic pressure across the cell membrane .
the model has 6 time dependent variables @xmath28 , @xmath185 , @xmath186 , @xmath163_{\rm i}$ ] , @xmath167_{\rm i}$ ] and @xmath187_{\rm i}$ ] , and the equations are summarized in appendix a. akinori noma published in 1996 an excellent review of the ionic mechanisms of the cardiac pacemaker potential ( noma , 1996 ) . in this short paper ,
noma investigated the mechanisms that produce spontaneous activity in sinoatrial node cells , and introduced the following overview of the relevant ionic currents .
* deactivation of @xmath137 ( @xmath188 ) . * removal of inactivation of @xmath189 and @xmath190 . *
activation of the hyperpolarization activated current ( @xmath191 ) . * activation of l type @xmath192 current ( @xmath189 ) . * activation of t
type @xmath192 current ( @xmath193 ) .
@xmath194 : : : a cation current with reversal potential of about @xmath195 .
@xmath196 : : : spontaneous openings of the @xmath149 channels . @xmath184 : : : na / k pump current . @xmath181 : : : na / ca exchange current . @xmath197 : : : atp sensitive @xmath149 channels .
we will not try to determine the relative amplitude of the above current components here , but instead demonstrate that only five membrane currents is sufficient to ensure stable intracellular ionic concentrations . these are @xmath137 , @xmath180 , @xmath184 , @xmath181 , and @xmath139 . in our model
we assume that @xmath198 and @xmath199 are of minor importance ; i.e. when we talk about @xmath180 we mean @xmath200 , and when we talk about @xmath137 we mean @xmath201 .
the various parameters play different roles in the model , and we list them in different tables to distinguish between fundamental physical constants ( table [ table1 ] ) , experimentally observed constants ( table [ table2 ] ) , adjustable parameters ( table [ table3 ] ) and initial conditions ( table [ table4 ] ) in the model .
one parameter not listed is the gating charge @xmath34 , eq .
( [ eq14 ] ) , which is the origin of the factor @xmath202 in eqs .
( [ eq31 ] ) , ( [ eq33 ] ) and ( [ eq34 ] ) .
this corresponds to a slope factor for the activation and inactivation curves of @xmath203 at @xmath204 .
the observed slope factors are @xmath205 for activation of @xmath137 ( shibasaki 1987 ) , @xmath206 for activation of @xmath180 ( hagiwara et al .
1988 ) , @xmath207 for inactivation of @xmath180 ( hagiwara et al .
1988 ) , @xmath208 for activation of @xmath139 ( muramatsu et al .
1996 ) , and , finally , @xmath209 for inactivation of @xmath139 ( muramatsu et al .
hence , we see that @xmath210 , corresponding to a gating charge of @xmath35 , is a good approximation . the half activation and inactivation potentials in the model ( @xmath211 , @xmath103 , @xmath105 , @xmath104 and @xmath106 ) are based on the experiments of shibasaki ( 1987 ) , hagiwara et al .
( 1988 ) and muramatsu et al .
( 1996 ) , and we use a value of @xmath212 that gives a reversal potential for the sodium pump in good agreement with the experiments of sakai et al .
the maximum time constants in these experiments were @xmath213 for activation of @xmath137 ( shibasaki 1987 ) , @xmath214 for inactivation of @xmath180 ( hagiwara et al .
1988 ) and @xmath215 for inactivation of @xmath139 ( muramatsu et al .
1996 ) . in the model , however , we combine these and use a maximum time constant of @xmath216 for both @xmath217 , @xmath218 and @xmath219 .
finally , we use typical values for cell volume , cell capacitance , and extracellular ionic concentrations .
two of the differential equations in the model have almost identical but opposite dynamics .
if we modify the half
inactivation potentials of calcium from @xmath220 to @xmath221 , it is possible to relate the inactivation gating of calcium to the activation gating of potassium by the equation @xmath222 since the time constants for these two processes are equal . we have thus reduced the number of differential equations in the model by one .
this computational saving will be irrelevant for the computation of one action potential , but is important in an extended model with thousands of coupled cell , or in a long time integration of the one cell model .
the relative amplitude of the ionic currents that drive membrane depolarization during diastole is still a matter of debate .
difrancesco ( 1993 ) argues that the hyperpolarization activated current ( @xmath191 ) is the only current that can generate and control the slow depolarization of pacemaker cells .
the @xmath191 current is normally carried by @xmath111 and @xmath149 .
guo et al .
( 1995a ; 1995b ; 1996 ) reported another current , called the sustained inward current @xmath190 , where the major charge carrier is believed to be @xmath111 .
also a @xmath223 `` window '' current has been observed in rabbit sinoatrial node cells ( denyer & brown 1990 ) .
it is possible that any one of these currents , or a combination of them , is responsible for membrane depolarization during diastole .
however , the estimates of the net membrane current during diastole is so imprecise ( zaza et al .
1997 ) , that we could not form a judgment on the question . during diastole
the electrochemical driving forces produce outward @xmath149 currents and inward @xmath111 and @xmath223 currents , and the driving force for @xmath223 is much larger than that for @xmath111 .
these findings implies that a significant background influx of @xmath223 is possible during diastole , and that this current might be responsible for the pacemaker activity in sinoatrial node cells .
we denote the conductance for this current @xmath224 , and modify ( [ eq33 ] ) combined with ( [ eq85 ] ) to read as @xmath225\,\sinh\left(\frac{v - v_{\rm ca}}{v_t}\right ) \;,\ ] ] with @xmath226 . in our model @xmath224
is responsible for the slow diastolic depolarization , and it is thus our pacemaker current .
however , as suggested by guo et al .
( 1995b ) , it is indeed possible that the sustained inward current @xmath190 may largely replace the role of the @xmath223 currents , assumed here and in previous studies ( wilders 1993 ) . the density of ionic channels , exchangers and pumps ( i.e. @xmath227 , @xmath224
, @xmath100 , @xmath95 , @xmath228 and @xmath229 ) can vary significantly from cell to cell . in order to reproduce recorded action potentials ( fig .
7 a. in baruscotti et al .
( 1996 ) ) , we fit the adjustable parameters ( table [ table3 ] ) and the initial conditions ( table [ table4 ] ) numerically .
more details of the method are given in ( endresen 1997a ) .
many different combinations of @xmath99 , @xmath224 , @xmath100 , @xmath230 , @xmath228 , and @xmath229 resulted in good approximations to the experimentally recorded waveform , from which we conclude that different cells can produce the same action potential although they have different mixtures of ionic channels , exchangers , and pumps .
the five differential equations in the model were solved numerically using a fifth order runge
kutta method with variable steplength .
more details are given in ( endresen 1997b ) .
we computed the work @xmath231 , defined as minus the integral on the right hand side of ( [ eq70 ] ) , to check that the equation @xmath232 was satisfied numerically .
this could also be used for varying the steplength in an efficient way ( marthinsen et al .
1997 ) , since the solution of our differential equations must satisfy this constraint .
these `` checksum equations '' are shown in appendix b. in fig .
1 ( a ) the modeled action potential is shown together with the experimental curve of baruscotti et al .
the curves are identical in shape , but we adjusted the modeled curve somewhat ( we multiplied the voltage amplitude by a factor 1.25 , without changing the minimum value ) to obtain the same voltage amplitudes . at present
it is not clear to us which mechanism is needed in the model in order to avoid this factor .
1 ( b ) shows the five membrane currents in the model , @xmath180 , @xmath139 , @xmath137 , @xmath184 , and @xmath181 .
2 shows the spontaneous action potentials together with the intracellular ionic concentrations and the osmotic pressure across the cell membrane .
these computations used the initial conditions stated in table [ table4 ] .
cells must generate their membrane potential by actively transporting ions against the respective concentration gradients . to examine this process in our model , we ran a simulation starting with equal intracellular and extracellular ionic concentrations : @xmath233_{\rm i } } = { [ \rm k]_{\rm e } } = 5.4 \ , { \rm mm}$ ] , @xmath234_{\rm i } } = { [ \rm ca]_{\rm e } } = 2 \ , { \rm mm}$ ] , and @xmath235_{\rm i } } = { [ \rm na]_{\rm e } } = 140 \ , { \rm mm}$ ] .
the results are presented in fig .
3 , that shows the voltage and nernst potentials ( a ) , and the energies ( b ) in a long time simulation . after approximately 750 seconds ( 12.5 minutes ) the system reaches oscillations identical to the original oscillations shown in figs . 1 and 2 ( this can not be seen from fig .
3 since the time scale is very different ) .
this long time simulation is a numerical indication that the oscillations in fig . 2 and 3 indeed correspond to a stable limit cycle .
we have presented a simple model for the cells of the rabbit sinoatrial node .
our model involves only @xmath2 , and @xmath223 ions , their respective channels , the @xmath236 exchanger , and the @xmath2 pump .
the equations were derived using basic physical principles and conservation laws .
since the only source of energy in our model is the sodium potassium pump , we can easily track the flow of energy , and show that the pump works to generate a transmembrane voltage , osmotic pressure difference , and an entropy .
our equations also account for the energy lost due to downhill ionic fluxes through the exchanger and channels .
a prediction of osmotic pressure variations is a novel result of our energy analysis .
the intracellular ionic concentrations are dynamic variables in our model , governed by the conservation eqs .
( [ eq53 ] ) , ( [ eq54 ] ) , and ( [ eq55 ] ) .
this allows us to replace the standard differential equation for the voltage ( [ eq52 ] ) with the algebraic eq .
( [ eq63 ] ) . although a number of other ionic models also keep track of intracellular ionic concentrations ( see wilders ( 1993 ) ) , we are unaware of any other model using an algebraic equation for the membrane potential
. models that use the standard voltage differential eq .
( [ eq52 ] ) have a phase space with one superfluous extra dimension .
the initial condition for this extra differential equation can not be chosen independently of the initial conditions for the conservation eqs .
( [ eq53 ] ) , ( [ eq54 ] ) , and ( [ eq55 ] ) if it is , the computed membrane potential will be erroneous . for these reasons
, we suggest that our algebraic expression for the membrane potential should replace the standard voltage differential equation in models where intracellular ionic concentrations are dynamic variables .
our model does not include the funny current ( @xmath191 ) , atp sensitive channels , stretch - activated channels , or other ion channels that may be important ( boyett 1996 ) .
we also ignored the effect of calcium uptake and release from the sarcoplasmatic reticulum , which would affect the nernst potential of calcium , but not the membrane potential .
we have assumed that the ionic channels are governed by a markov process , that the maximum of the activation / inactivation time constant occurs at the same voltage as the inflection point of the sigmoidal steady state activation / inactivation curve , and that the steady state activation / inactivation curves were temperature independent .
also , we have assumed that the cell volume is constant .
while such assumptions reduce the number of parameters in the model , they may also result in discrepancies with experiment .
finally , we would like to point out that our model is based on experiments where some were conducted at room temperature ( 22@xmath237 ) ( baruscotti et al .
1996 ; muramatsu et al .
1996 ) , while others were performed at @xmath238 ( shibasaki 1987 ; hagiwara et al . 1988 ; sakai et al . , 1996 ) .
it is not clear what effect varying temperature has in our model as this was not checked out numerically .
the values of the parameters @xmath99 , @xmath100 , @xmath95 , @xmath228 and @xmath229 , given in table [ table3 ] , are only an estimate of the actual physiological parameters .
we did not systematically study the dynamics of the model for different values of the parameters , but we hope that future experiments will help to discriminate between different parameter sets that may reproduce the experimentally recorded action potentials .
we are most grateful to drs .
baruscotti , difrancesco , and robinson who supplied us with recordings of action potential waveforms from rabbit sinoatrial node .
lars petter endresen would like to thank professor per jynge for giving a fascinating introduction to the exciting field of cardiac electrophysiology .
discussions with per hemmer , kre olaussen , and jacques belair have been essential .
lars petter endresen was supported by a fellowship at ntnu , and has received support from the research council of norway ( programme for supercomputing ) through a grant of computing time .
kevin hall acknowledges support from the medical research council of canada .
baruscotti m , difrancesco d , robinson rb ( 1996 ) a ttx sensitive inward sodium current contributes to spontaneous activity in newborn rabbit sino atrial node cells .
journal of physiology ( london ) 492:2130 boyett mr , harrison sm , janvier nc , mcmorn so , owen jm , shui z ( 1996 ) a list of vertebrate cardiac ionic currents : nomenclature , properties , function and cloned equivalents .
cardiovascular research 32:455481 donnan fg ( 1911 ) theory of membrane equilibria and membrane potentials in the presence of non dialysing electrolytes .
a contribution to physical chemical physiology .
zeitschrift fr elektrochemie and angewandte physikalische chemie 17:572581 muramatsu h , zou ar , berkowitz ga , nathan rd ( 1996 ) characterization of a ttx sensitive @xmath111 current in pacemaker cells isolated from the rabbit sinoatrial node .
american journal of physiology 270:h2108h2119 | we give an explicit formula for the membrane potential of cells in terms of the intracellular and extracellular ionic concentrations , and derive equations for the ionic currents that flow through channels , exchangers and electrogenic pumps .
we demonstrate that the work done by the pumps equals the change in potential energy of the cell , plus the energy lost in downhill ionic fluxes through the channels and exchangers .
the theory is illustrated in a simple model of spontaneously active cells in the cardiac pacemaker .
the model predicts the experimentally observed intracellular ionic concentration of potassium , calcium , and sodium .
likewise the shapes of the simulated action potential and five membrane currents are in good agreement with experiments .
we do not see any drift in the values of the concentrations in a long time simulation , and we obtain the same asymptotic values when starting from the full equilibrium situation with equal intracellular and extracellular ionic concentrations . = _ = 11 # 1#2#3#4 to#1 0= # 2 minus1fil 0-.50 0 plus1fil -.500 minus1fil # 4 # 1#2 to#1 = 2=#2 0=to234 0=0 0 by0 012 minus1fil 0-.50 0 plus1fil -.500 minus1fil 567 # 1#2#3#4 to#1 0= # 2 0 plus1fil minus1fil # 4 # 1#2 to#1 = 2=#2 0=to234 012 0 plus1fil minus1fil 567 # 1#2#3#4 _
border#1#2#3 # 1#2_border#1#2 # 1#2_border#1#21 # 1#2#3#4 _
border#1#2#3 # 1#2_border#1#2 # 1#2_border#1#21 = manfnt # 1 2= # 1 0= to0to8 0= to00to8 0= to0width8height2 depth2 0= to0width8height2 depth2 0= to0 0= to00 2 _ txt _ box # 4 _ inbox = # 1 by-#2 by-#2 _ border _ border
_ txt= _ box=#4_txt#1 to_box _ box -_box
to_box_txt _ inbox _
txt= _ = = # 1#2#3 |
since the neutrino oscillation was first observed at super - kamiokande ( sk ) @xcite , many experiments have measured the physics parameters of the neutrino oscillation @xcite-@xcite .
some of these experiments observe the survival probability of @xmath6 and @xmath7 which are generated in the atmosphere by the cosmic ray @xcite .
accelerator based long baseline experiments @xcite also measure the @xmath6 survival probability . from the combined results of these experiments @xcite-@xcite ,
the mass - squared difference and the mixing angle are obtained as @xmath19 [ eq : atm_data ] @xmath20 the sign of the mass - squared difference , eq .
( [ eq : mass_atm ] ) , can not be determined from these experiments .
the sk collaboration also reported that the atmospheric neutrinos oscillate into active neutrinos @xcite .
the combined results of the solar neutrino observations @xcite , which measure the survival probability of @xmath21 from the sun , and the kamland experiment @xcite , which measure the @xmath22 flux from the reactors at distances of a few 100 km , find @xmath23 [ eq : sun_data ] @xmath20 where the sign of mass - squared difference has been determined by the matter effect inside the sun @xcite .
the sno experiment determined that @xmath21 from the sun changes into active neutrinos @xcite . for the reactor experiments , which measure the survival probability of @xmath22 from the reactor at distances of @xmath24 km @xcite , the chooz experiment observed no signal , giving @xmath25 [ eq : rct_data ] @xmath20 at the 90% confidence level @xcite .
more recently , the t2k collaboration reported several @xmath26 appearance candidate events giving @xmath27 for @xmath28 and @xmath29 with the normal ( inverted ) hierarchy at the 90% confidence level @xcite .
the minos collaboration reported @xmath30 also for @xmath28 and @xmath29 with the normal ( inverted ) hierarchy @xcite .
the double chooz collaboration , which is one of the new reactor experiments , found hints of reactor electron anti - neutrino disappearance consistent with neutrino oscillation and reported @xmath31 from analyzing both the rate of the prompt positrons and their energy spectrum @xcite . recently
another new reactor experiment , the dayabay experiment @xcite , announced that they have measured the neutrino mixing angle as @xmath32 which is more than @xmath33 away from zero . the reno collaboration , which also measure the reactor @xmath22 survival probability , shows the evidence of the non - zero mixing angle ; @xmath34 from a rate - only analysis , which is @xmath35 away from zero .
since the miniboone experiment @xcite did not confirm the lsnd observation of rapid @xmath36 oscillation @xcite , there is no clear indication of experimental data which suggests more than three neutrinos .
therefore the @xmath26 appearance analysis of t2k @xcite and minos@xcite presented above assume the 3 neutrino model , with the @xmath37 flavor mixing , the mns ( maki - nakagawa - sakata ) matrix @xcite @xmath38 relating the weak interaction eigenstates @xmath39 and the mass eigenstate @xmath40 with the mass @xmath41 ( @xmath42 ) .
the mass - squared differences that dictate the neutrino oscillation phase are then identified as @xmath43 where only the magnitude of the larger mass - squared difference , @xmath44 is determined in eq .
( [ eq : mass_atm ] ) .
the @xmath45 case is called normal , while @xmath46 case is called inverted neutrino mass hierarchy . with a good approximation @xcite
, the three independent mixing angles of the mns matrix can be related to the oscillation amplitudes in eqs .
( [ eq : mixing_atm ] ) , ( [ eq : mixing_sun ] ) , and @xmath47 in eqs .
( [ eq : mixing_rct ] ) , ( [ eq : t2k_s13 ] ) to ( [ eq : dayabay_s13 ] ) , with the mns matrix elements @xmath48 : @xmath49 [ eq : mns_exp ] @xmath20 in the last equations , the defining region of the three mixing angles @xmath50 can be chosen as @xmath51 @xcite , which is consistent with the convention of non - negative @xmath52 and @xmath53 @xcite .
the argument of @xmath54 gives the cp phase of the mns matrix , @xmath55 even after @xmath56 , the smallest of the three mixing angles , is measured by the accelerator based long baseline experiments @xcite and the reactor experiments @xcite , three unknowns remain in the 3 neutrino model , which are the sign of the larger mass - squared differences , normal ( @xmath57 ) or inverted ( @xmath58 ) , the leptonic cp phase ( @xmath10 ) , and the sign of @xmath59 if its magnitude differs from zero significantly .
the main purpose of the next generation neutrino oscillation experiments is to determine these three unknown parameters of the three neutrino model .
the accelerator based long baseline neutrino oscillation experiment with two - detectors for one - beam , such as tokai - to - kamioka - and - korea ( t2kk ) experiment @xcite-@xcite , is one of the promising experiments for measuring all the three unknowns .
when one measures the neutrino energy ( @xmath60 ) and the magnitude of the @xmath21 appearance probability at significantly different baseline lengths , the degeneracy between the sign of the larger mass - squared difference and the cp phase can be resolved @xcite-@xcite , since they affect the magnitude of the @xmath21 appearance probability and the neutrino energy at the first oscillation maximum differently at the two baseline lengths .
once the mass hierarchy and the cp phase are determined , the sign of @xmath59 can be determined @xcite since the @xmath26 oscillation probability is proportional to @xmath61 . in the previous works @xcite-@xcite , physics impacts of the t2kk experiment
have been studied systematically , and the following observations have been made : if a 100 kton water detector is placed in korea at @xmath62 km observing the @xmath13 off - axis beam ( oab ) during the t2k exposure time of @xmath63 pot ( protons on target ) , the mass hierarchy can be resolved at 3@xmath8 level for @xmath64 when the hierarchy is normal ( inverted ) , and the cp phase can be constrained uniquely , by measuring the ccqe ( charged - current quasi elastic ) events @xcite .
the sign of @xmath59 can also be determined for @xmath65 with @xmath66 level , if @xmath67 for the normal hierarchy @xcite .
when we take into account the smearing of reconstructed neutrino energy due to finite detector resolution and the fermi motion of target nucleons , resonance production , and the neutral current ( nc ) @xmath68 production background to the @xmath21 appearance signal , it is found that the mass hierarchy pattern can still be determined at @xmath66 level for @xmath69 when the hierarchy is normal ( inverted ) , but the cp phase can no longer be established at @xmath66 level with @xmath63 pot @xmath6 exposure @xcite . in ref.@xcite ,
matter distribution profile along the t2k and tokai - to - korea baselines have been studied , and merits of splitting the total exposure time into half @xmath6 and half @xmath7 beam have been reported .
studies on the impacts of the neutrino energy smearing and the nc @xmath68 background for the @xmath36 oscillation measurements are in progress @xcite .
oki island is placed between japan and korea along the t2k baseline and inside japanese territory .
the distance from j - parc to oki island is about @xmath70 km which is about two times longer than that of the t2k experiment ( @xmath0 km ) . in fig .
[ fig : t2oki1 ] , we show the surface map of the t2k , t2ko ( tokai - to - kamioka - and - oki ) , and t2kk experiments , in which the red lines denote the baselines for each experiment and the yellow blobs show the center of the neutrino beam from j - parc at the sea level , when the off - axis angle at sk is @xmath71 , @xmath72 , and @xmath73 .
the authors of ref.@xcite consider the physics performance of a 100 kton liquid argon time projection chamber placed at oki island . in ref.@xcite
, the authors studied the oki island site from the geological , geographic and infrastructure points of view for the possibility to construct a large detector .
they conclude that oki island is one of the good candidate sites for a large detector . in this paper , we study the physics potential of a 100 kton water detector placed in oki island instead of korea during the t2k experimental period , by using exactly the same setting assumed for the t2kk experiment in ref.@xcite , except for the location of the detector .
this allows us to compare the physics capability of the two proposals on the same footing .
since we do assume in this analysis that the @xmath74 ccqe events can be distinguished from the background , our results should equally be applied for a more advanced liquid argon detector @xcite .
this article is organized as follows . in the next section ,
we briefly review neutrino oscillation in the matter and give useful approximation formula for @xmath74 and @xmath75 oscillation probabilities . in section [ sec : far ]
, we show the matter profile between j - parc and oki island by using recent seismological measurements and give relations between the off - axis angle observable at oki and that at sk .
the expected numbers of @xmath74 @xmath76 ccqe events at the far ( oki ) and the near ( sk ) detectors are shown for typical parameters in section [ sec : event ] . in section [ sec : chi ] , we present the @xmath77 function with which we estimate the statistical sensitivity of the t2ko and t2kk experiments on the mass hierarchy and the cp phase . in section
[ sec : mass ] , we show the results on the mass hierarchy determination of the t2ko experiment and compare it with t2kk , and also with 2kn where the 100 kton detector is placed in the kamioka site making the total fiducial volume 122 kton at @xmath0 km . in section [ sec : cp ] , we show the cp phase sensitivity of t2ko , t2kk , and 2kn experiments .
we summarize our findings in the last section .
in long baseline neutrino oscillation experiments , neutrinos from the accelerator interact coherently with electrons and nucleons by charged and neutral current interactions .
these coherent interactions make an additional potential in the effective hamiltonian @xcite . because the potential from the neutral current interactions are flavor - blind , it does not affect the neutrino oscillation probability in the three neutrino model . on the other hand , the charged - current interactions contribute only to the effective potential of @xmath21 and @xmath22 .
the neutrino oscillation probabilities hence depend on the matter density profile along the baseline .
detailed studies on the matter distribution dependence of the oscillation probabilities for the t2k and t2kk experiments are found in ref .
@xcite . in this exploratory analysis
, we compare the physics capability of the t2ko , t2kk , and t2k experiments by using the average matter density along the baselines which is found to approximate the matter effects rather accurately @xcite .
the hamiltonian of a neutrino propagating in the matter is then expressed as @xmath78\ , , \label{eq : hamiltonian1}\\ & = & \dfrac{1}{2e } \tilde{u}^ { } \left ( \begin{array}{ccc } \lambda_1 & & \\ & \lambda_2 & \\ & & \lambda_3 \end{array } \right ) \tilde{u}^\dagger\ , , \label{eq : hamiltonian2}\end{aligned}\ ] ] [ eq : hamiltonians ] @xmath20 on the flavor space @xmath79 , with @xmath80 , where @xmath81 gives the matter effect with the electron number density @xmath82 , which is approximated by the average matter density @xmath83 ( g/@xmath84 ) along the baseline . by using the solution of eq .
( [ eq : hamiltonian2 ] ) , the oscillation probability that an initial flavor eigenstate @xmath85 is observed as a flavor eigenstate @xmath86 after traveling a distance @xmath87 along the baseline is @xmath88 [ eq : ppps ] @xmath20 where @xmath89 all our numerical results are based on eq .
( [ eq : ppps ] ) . however , we find the following analytic approximations @xcite useful for understanding the reason why and how the one - beam two - detectors long baseline experiments such as t2kk @xcite-@xcite and t2ko , with a far detector in oki island , can determine the neutrino mass hierarchy and the cp phase simultaneously .
they are obtained by expanding the oscillation probabilities in terms of the three small parameters ; the matter effects , which is small at energies below a few gev around the earth crust for @xmath90 g/@xmath84 and @xmath91 km , the oscillation phase @xmath92 , which is also small ( @xmath93 ) near the first oscillation maximum , @xmath94 , and the mixing factor @xmath95 , eq .
( [ eq : def_srct ] ) , which is about @xmath96 from the recent reactor measurements eqs .
( [ eq : dayabay_s13 ] ) and ( [ eq : reno_s13 ] ) .
the @xmath6 survival probability can then be approximated as @xmath97 around the first maximum of the main oscillation phase @xmath98 with [ eq : abmu ] @xmath99 we find that the above formula reproduce the survival probability with 1% accuracy throughout the parameter range explored in this analysis , which covers all the three neutrino model parameters in the @xmath66 allowed range of the present measurements eqs .
( [ eq : atm_data])-([eq : reno_s13 ] ) , for the neutrino energies 400 mev @xmath100 4 gev ( for @xmath6 , @xmath7 ) , and for the baseline lengths @xmath101 km @xmath102 1200 km , except where the probability is very small , ( @xmath103 ) @xcite .
the matter effects are proportional to the term @xmath104 and it appears with the opposite sign both in the amplitude correction term @xmath105 eq .
( [ eq : amu ] ) where @xmath106 for the normal while @xmath107 for the inverted hierarchy , and also in the phase shift in eq .
( [ eq : pmm ] ) , where @xmath108 for the normal , @xmath109 for the inverted hierarchy around the oscillation maximum @xmath110 . nevertheless , the effect is too small to be observed in the near future because the term @xmath111 is bounded as @xmath112 at @xmath113 c.l . from eq .
( [ eq : mixing_atm ] ) , and because both @xmath105 and @xmath114 are proportional to the product of the two small parameters , the matter effect term of @xmath115 and @xmath116 .
the coefficient of @xmath117 in @xmath114 is also small , only @xmath118 of the main oscillation phase @xmath119 .
in other words , the @xmath6 survival probability is very insensitive to the unknown parameters of the three neutrino model , and hence can be used to measure @xmath120 and @xmath121 accurately . under the same conditions that give eq .
( [ eq : pmm ] ) for the @xmath6 survival probability , the @xmath21 appearance probability can be approximated as @xcite @xmath122 where we retain both linear and quadratic terms of @xmath123 and @xmath124 .
the analytic expressions for the correction terms @xmath125 , @xmath126 and @xmath127 are found in ref.@xcite . for our semi - quantitative discussion below
, the following numerical estimates @xcite for @xmath128 and @xmath129 suffice : [ eq : abeapprox ] @xmath131 the term @xmath127 is relevant only when the @xmath26 oscillation probability is very small @xcite .
the first term in @xmath125 in eq .
( [ eq : aeaprox ] ) is sensitive not only to the matter effect but also to the mass hierarchy pattern , since @xmath132 for the normal while @xmath133 for the inverted hierarchy around the oscillation maximum . for the normal ( inverted ) hierarchy , the magnitude of the @xmath26 transition probability
is enhanced ( suppressed ) by about @xmath134 at kamioka , @xmath135 at oki island , and @xmath136 at @xmath137 km in korea , around the first oscillation maximum , @xmath138 .
when @xmath139 is fixed at @xmath140 , the difference between the two hierarchy cases grows with @xmath87 , because the matter effect grows with @xmath60 . within the allowed range of the model parameters , the difference of the @xmath125 between sk and a far detector at oki or korea becomes @xmath141 [ eq : ae_diffs ] @xmath20 where the upper sign corresponds to the normal , and the lower sign for the inverted hierarchy .
the hierarchy pattern can hence be determined by comparing @xmath142 near the oscillation maximum @xmath143 at two vastly different baseline lengths @xcite-@xcite , _ independently _ of the sign and magnitude of @xmath144 . in eq .
( [ eq : beaprox ] ) , it is also found that the first term in @xmath126 , which shifts the oscillation phase from @xmath119 to @xmath145 , is also sensitive to the mass hierarchy pattern . as in the case for @xmath125 ,
the difference in @xmath126 between sk and a far detectors is found to be @xmath146 [ eq : be_diffs ] @xmath20 where the upper sign is for the normal , and the lower sign for the inverted hierarchy .
this implies that the mass hierarchy pattern can also be discriminated by measuring the neutrino energy of the first oscillation maximum at two different baseline lengths . as in the case of the oscillation amplitude above
, this determination can be made independent of the unknown @xmath117 terms since they cancel in the differences , eq .
( [ eq : be_diffs ] ) .
once the sign of @xmath147 is fixed by the terms linear in @xmath83 , the terms linear in @xmath123 , which appear as those proportional to @xmath148 in eq .
( [ eq : abeapprox ] ) , allow us to constrain @xmath144 via the amplitude @xmath125 , and @xmath117 via the phase shift @xmath126 .
therefore , @xmath10 can be measured uniquely once the mass hierarchy pattern is determined . from the above discussions , we understand qualitatively why the mass hierarchy as well as both @xmath144 and @xmath117 can be determined uniquely by observing the @xmath26 oscillation probability around the first oscillation maximum at two vastly different baseline lengths .
therefore , in order to take advantage of this very efficient mechanism to determine all the main unknowns of the three neutrino model , one should arrange for high neutrino flux both at near and far detectors around the first oscillation maximum which appears at the same energy - to - baseline - length ratio , @xmath149 for @xmath150ev@xmath151 ; eq .
( [ eq : mass_atm ] ) .
this observation led to the t2kk proposal of choosing the @xmath73 off - axis beam ( @xmath152gev ) at sk and @xmath153 off - axis beam ( @xmath154gev ) at @xmath137 km @xcite .
since the baseline length of @xmath155 km to oki island is about a factor of two longer than @xmath156 km for t2k , we may also expect that a beam with smaller off - axis angle at oki island will enhance its physics capability ; see , however , discussions in section [ sec : mass ] .
the oscillation probabilities for the anti - neutrino , @xmath157 , are obtained from the expressions for @xmath158 by reversing the sign of the matter effect term , ( @xmath159 ) , and that of @xmath10 , ( @xmath160 ) .
the differences in the shift terms @xmath125 and @xmath126 for @xmath26 and @xmath161 and @xmath162 for @xmath36 oscillation probabilities are , respectively , @xmath163\ , , \label{eq : diff_be}\end{aligned}\ ] ] [ eq : diff_ae_and_be ] @xmath20 where the upper sign should be taken for the normal and the lower sign for the inverted hierarchy . from eq .
( [ eq : diff_ae ] ) , the difference of the amplitude between @xmath21 and @xmath22 appearance probability grows with the baseline length with the opposite sign for the normal and inverted hierarchies .
the dependence on @xmath144 is also strong since @xmath125 and @xmath161 changes in the opposite direction when we vary @xmath144 .
the phase - shift term @xmath126 and @xmath162 also change sign in the opposite direction for the normal and inverted hierarchies , whose sign is independent of @xmath117 because of the smallness of the @xmath117 dependence in eq .
( [ eq : diff_be ] ) . by the same token
, the use of both @xmath6 and @xmath7 beams does not improve significantly the measurement of @xmath117 .
because these sign changes occur independently of the @xmath87-dependence of the oscillation probabilities , the physics potential of the t2ko experiment will be improved by dividing the total exposure time to neutrino and anti - neutrino beams just as in t2kk @xcite .
it is worth noting here that all the formalism presented in this section should be useful for studying physics potential of t2k plus nova @xcite , whose baseline length of @xmath164 km lies between those of tokai to oki island ( @xmath165 km ) and to korea ( @xmath166 km ) .
in this section , we study the matter profile along the baselines between j - parc and kamioka , oki island , and korea by referring to recent seismological measurements @xcite- @xcite .
we also show the relation between the off - axis angles at sk and oki island , and the beam profiles for the relevant off - axis angles @xcite . because oki island is placed just at the middle point between sk and the east shore of korea along the t2k beam , as shown in fig .
[ fig : t2oki1 ] , the cross section view of the t2k , t2ko , and t2kk experiments along the baselines can be shown on one frame as in fig .
[ fig : xsec - t2oki ] .
the horizontal axis of fig . [ fig : xsec - t2oki ] gives the distance from j - parc along the arc of the earth surface and the vertical axis denotes the depth below the sea level .
three curves show the baselines from the j - parc to sk ( @xmath0 km ) , oki island ( @xmath70 km ) and to the korean east shore at @xmath12 km .
the numbers in the white boxes are the average matter density of the layer in units of g/@xmath84 .
@xcite-@xcite . ]
if we assume that the matter density in each layer of the earth crust as shown in fig . [ fig : xsec - t2oki ] has the value equals to the quoted average , we can estimate the average matter density @xmath83 along the three baselines : @xmath167 [ eq : matter ] @xmath20 the error of these average density can be estimated from the uncertainty of the matter density in each region and the location of the boundary of each layer .
they are measured by using the velocity of the seismic wave in most geophysical researches . from the uncertainty in the correlation between the matter density and the measured @xmath168-wave sound velocity @xcite
, we adopt @xmath169 overall error for the matter density .
small fluctuation in the matter density in each layer of the earth crust does not affect the neutrino oscillation probabilities significantly , because the contribution from the higher fourier modes of the matter density distribution is strongly suppressed @xcite .
the locations of the boundaries are measured rather accurately from the reflection point of the seismic wave . from fig .
[ fig : xsec - t2oki ] , we find that the neutrino beam for the tokai - to - oki baseline goes through the upper crust layer with @xmath170g/@xmath84 , except when it crosses fossa magna filled with sediment .
the uncertainty of the boundary depth between the sediment layer and the upper crust is only a few hundred meters .
the error from the uncertainty of the boundary depth can hence be safely neglected for the t2ko experiment .
the traveling distance in the mantle and the crust depends on the depth of the boundary between the lower crust and the upper mantle for the t2kk experiment .
the average matter density is sensitive to the boundary location for the baseline of @xmath171 km , because the baseline almost touches the mantle as can be seen from fig .
[ fig : xsec - t2oki ] .
the resulting uncertainty has been estimated and is found to be significantly smaller than the @xmath169 overall uncertainty @xcite .
we therefore use the average matter density along each baseline given in eq .
( [ eq : matter ] ) to estimate the oscillation probability and assign its overall uncertainty of @xmath169 in this study .
km , 1100 km , and 1200 km ) .
the horizontal axis gives the off - axis angle at sk and the vertical axis is that at the far detector locations .
the purple line shows the off - axis angle at oki island .
the red , blue , and black lines are the smallest off - axis angle at @xmath12 km , 1100 km and 1200 km , respectively .
the points corresponding to the purple circles and black squares are used in our numerical analysis . ]
figure [ fig : relation ] shows the relation among the off - axis angles of the neutrino beam from j - parc observable at sk , oki , and a far detector in korea ( @xmath12 km , 1100 km , and 1200 km ) . the horizontal axis gives the off - axis angle at sk and the vertical axis gives the corresponding off - axis angles at far detector locations .
the purple line shows the off - axis angle at oki island , which grows as that at sk is increased , because both sk and oki island is located in the east ( upper ) side of the beam center of the t2k neutrino beam ; see fig .
[ fig : t2oki1 ] .
the red , blue , and black lines are the smallest off - axis angle observable in korea at @xmath12 km , 1100 km , and 1200 km , respectively .
in contrast to the oki island case , the off - axis angle at korea decreases as that at sk grows , because korea is in the west ( lower ) side of the beam center , also as shown in fig .
[ fig : t2oki1 ] .
the points corresponding to the purple circles and black squares in the figure are used in our numerical analysis in the following sections . in fig .
[ fig : profile ] , the beam profiles of the three off - axis beams ( oab ) with @xmath2 , @xmath72 , and @xmath1 off - axis angles at sk are shown at sk , oki island ( three purple circles in fig .
[ fig : relation ] ) and at @xmath12 km in korea ( three black squares in fig .
[ fig : relation ] ) , together with the @xmath26 and @xmath36 oscillation probabilities as functions of the neutrino energy , @xmath60 in units of gev .
the left column panels are for the @xmath6 focusing beam and the right column ones are for the @xmath7 focusing beam .
the upper three rows , ( a1)-(a3 ) and ( b1)-(b3 ) , display the beam flux for the @xmath6 and @xmath7 focusing beam , respectively , at @xmath172 km in units of @xmath173/cm@xmath151/@xmath175pot ( protons on target ) .
the first row , ( a1 ) and ( b1 ) , gives the beam profile observable at each far detector for the @xmath2 off - axis beam ( oab ) at the sk ( red solid line ) , which gives the @xmath4 oab at oki island ( green dashed line ) , and the @xmath176 oab for a far detector in korea at @xmath12 km ( blue dotted line ) .
the second row , ( a2 ) and ( b2 ) , is for the @xmath72 oab at sk , which gives the @xmath177 oab at oki island , and the @xmath178 oab at @xmath12 km .
the case for the @xmath1 oab at sk , which gives the @xmath3 oab at oki island and the @xmath13 oab at @xmath12 km , is shown in the third row .
it is clearly seen that all fluxes at oki and korea are harder than those at sk .
the hardest flux with @xmath179 gev is found for the @xmath2 at sk beam at oki , and for the @xmath1 at sk beam in korea .
the @xmath72 at sk beam gives almost the same flux shape of @xmath180 oab at oki and at the @xmath12 km location in korea , as shown in fig .
[ fig : t2oki1 ] .
this may call upon the possibility of locating detectors both in oki and in korea , where both detectors observe exactly the same oab around the off - axis angle of @xmath178 . focusing beam and
the right ones are for the @xmath7 focusing beam .
the upper 6 panels , ( a1)-(a3 ) and ( b1)-(b3 ) , show the beam fluxes at @xmath172 km and the bottom 4 panels , ( p1)-(p4 ) , show the @xmath26 ( @xmath181 ) oscillation probability with the input parameters of eqs .
( [ eq : matter ] ) and ( [ eq : input ] ) for @xmath182 and @xmath29 .
the first row , ( a1 ) and ( b1 ) gives the beam profile at each far detector for the @xmath2 off - axis beam ( oab ) at sk , the second row , ( a2 ) and ( b2 ) is for the @xmath72 oab at sk , and the third row , ( a3 ) and ( b3 ) is for the @xmath1 oab at sk .
the fourth row , ( p1 ) and ( p3 ) , shows the @xmath26 and @xmath36 oscillation probability for the normal hierarchy , respectively , and the bottom row , ( p2 ) and ( p4 ) , is for the inverted hierarchy .
, title="fig : " ] focusing beam and the right ones are for the @xmath7 focusing beam .
the upper 6 panels , ( a1)-(a3 ) and ( b1)-(b3 ) , show the beam fluxes at @xmath172 km and the bottom 4 panels , ( p1)-(p4 ) , show the @xmath26 ( @xmath181 ) oscillation probability with the input parameters of eqs .
( [ eq : matter ] ) and ( [ eq : input ] ) for @xmath182 and @xmath29 .
the first row , ( a1 ) and ( b1 ) gives the beam profile at each far detector for the @xmath2 off - axis beam ( oab ) at sk , the second row , ( a2 ) and ( b2 ) is for the @xmath72 oab at sk , and the third row , ( a3 ) and ( b3 ) is for the @xmath1 oab at sk .
the fourth row , ( p1 ) and ( p3 ) , shows the @xmath26 and @xmath36 oscillation probability for the normal hierarchy , respectively , and the bottom row , ( p2 ) and ( p4 ) , is for the inverted hierarchy .
, title="fig : " ] the lower two rows in fig .
[ fig : profile ] show the @xmath26 ( @xmath36 ) oscillation probability in the left ( right ) column , for the normal hierarchy , ( p1 ) and ( p3 ) , and for the inverted hierarchy , ( p2 ) and ( p4 ) .
the oscillation probabilities are calculated for the average matter densities of eq .
( [ eq : matter ] ) , and the three - neutrino - model parameters : @xmath183 [ eq : input ] @xmath131 with @xmath184 [ eq : input_rct ] the plus ( minus ) sign in eq .
( [ eq : input_dm13 ] ) is taken for the normal ( inverted ) hierarchy , whose predictions are shown in the fourth ( bottom ) row of fig .
[ fig : profile ] .
the red solid line in these panels is the oscillation probability at the sk , the green dashed line is for oki island , and the blue dotted line is for a far detector in korea , at @xmath12 km . from the four plots ( p1)@xmath185(p4 ) in the bottom
two rows in fig .
[ fig : profile ] , we find that the first oscillation peak at sk appears at @xmath60 around @xmath186 gev for all the four cases , @xmath6 vs @xmath7 and for both hierarchies . since the oab peaks at around @xmath187 , 0.6 , and 0.5 gev for @xmath2 , @xmath72 , and @xmath1 , respectively , as shown by the red solid curves in the first three rows of fig . [
fig : profile ] , all the oab at sk between @xmath2 and @xmath1 can observe @xmath26 and @xmath36 oscillations with good efficiency . on the other hand ,
the oscillation probabilities at oki island are high at @xmath60 around @xmath188 gev from the green dashed curves in the bottom two rows , where the flux is slightly small for the @xmath1 oab at sk , as shown also by green dashed curves in the third row . finally , at @xmath12 km , the @xmath26 and @xmath189 oscillation probabilities are large at around @xmath190 gev , and hence the oab with @xmath1 at sk is most favorable @xcite-@xcite .
we will confirm the above observations for the whole region of the three neutrino model parameter space in the following sections .
in this section , we explain how we estimate event numbers at each far detector . in order to compare the physics capability of the t2ko experiment and that of t2kk experiment with a far detector at various baseline lengths and oab s
, we adopt the same conditions as those in refs .
@xcite : all the detectors at kamioka , oki island , and korea are assumed to have excellent detection and kinematical reconstruction capabilities for @xmath6 and @xmath21 charged current quasi - elastic ( ccqe ) events within the fiducial volumes of the 22.5 kton at kamioka ( sk ) and 100 kton at oki island or in korea .
we use the neutrino fluxes of j - parc beam at various off - axis angles @xcite and the ccqe cross sections off water target @xcite to compute event numbers as functions of ( reconstructed ) neutrino energy with the bin width of @xmath191 mev at @xmath192 mev .
the energy bin width of 200 mev is chosen to take account of kinematical reconstruction errors due to fermi motion , resonance production and detector resolutions @xcite .
we take account of background from secondary beams , such as @xmath21 , @xmath22 and @xmath7 fluxes for the @xmath6 focusing beam , but we do not consider other backgrounds including the single @xmath68 production from the neutral current ( nc ) events @xcite .
in other words , the results of our studies show what a perfect detector of a given fiducial volume can achieve with neutrino beams from j - parc when it is placed along the t2k beam line .
reconstruction efficiency and errors as well as the background rejection capabilities depend on specific detector designs and their studies are beyond the scope of this paper .
the number of @xmath85 ccqe events from the @xmath86 flux in the @xmath6-focusing beam in the @xmath193-th energy bin , @xmath194 , at each site are calculated as @xmath195 where the suffix `` d '' denotes the place of the far detector ( d = sk , oki , kr ) , @xmath196 is neutrino or anti - neutrino flavor @xmath197 , @xmath198 is the fiducial mass of each detector ( @xmath199 kton and @xmath200 kton ) , @xmath201 is the avogadro number , @xmath202 is the @xmath86 flux from j - parc @xcite which scales as @xmath203 , @xmath204 is the neutrino or anti - neutrino oscillation probability calculated by eq .
( [ eq : ppps ] ) with the average matter density of eq .
( [ eq : matter ] ) , and @xmath205 is the ccqe cross sections per nucleon in water for each flavor type @xcite . we also define the event number for @xmath7 focusing beam as @xmath206 where @xmath207 gives the @xmath208 flux in the @xmath7 focusing beam @xcite .
because we only consider the contribution from the secondary neutrino fluxes of the primary beam as the background in this analysis , each @xmath209- and @xmath210-like event numbers in the @xmath193-th bin are calculated as @xmath211 for @xmath212 at each far detector `` d '' ( d = sk , oki , kr ) .
we also define the event numbers with @xmath7 focusing beam as @xmath213 for @xmath212 at each far detector .
[ [ sk ] ] sk ~~ -like ccqe events for @xmath60 ( @xmath214 ) @xmath215 gev at sk ( 22.5kton ) with @xmath1 off - axis angle and @xmath5pot exposure : ( a1 ) and ( a2 ) are for @xmath6 focusing beam , and ( b1 ) and ( b2 ) are for @xmath7 focusing beam ; whereas ( a1 ) and ( b1 ) are for the normal hierarchy , and ( a2 ) and ( b2 ) are for the inverted hierarchy .
the results are for @xmath182 , @xmath29 , and @xmath216 g/@xmath84 and the other input parameters of eq .
( [ eq : input ] ) .
the red solid line in each panel denotes the total event numbers , the purple dotted line shows the @xmath217 contribution in ( a1 ) and ( a2 ) , @xmath218 in ( b1 ) and ( b2 ) , the blue short dashed line gives the sum of the contributions from @xmath36 and @xmath217 in ( a1 ) and ( a2 ) , that of @xmath26 and @xmath218 in ( b1 ) and ( b2 ) , and the green dashed line shows the total background contribution from the secondary beams .
, title="fig : " ] -like ccqe events for @xmath60 ( @xmath214 ) @xmath215 gev at sk ( 22.5kton ) with @xmath1 off - axis angle and @xmath5pot exposure : ( a1 ) and ( a2 ) are for @xmath6 focusing beam , and ( b1 ) and ( b2 ) are for @xmath7 focusing beam ; whereas ( a1 ) and ( b1 ) are for the normal hierarchy , and ( a2 ) and ( b2 ) are for the inverted hierarchy .
the results are for @xmath182 , @xmath29 , and @xmath216 g/@xmath84 and the other input parameters of eq .
( [ eq : input ] ) .
the red solid line in each panel denotes the total event numbers , the purple dotted line shows the @xmath217 contribution in ( a1 ) and ( a2 ) , @xmath218 in ( b1 ) and ( b2 ) , the blue short dashed line gives the sum of the contributions from @xmath36 and @xmath217 in ( a1 ) and ( a2 ) , that of @xmath26 and @xmath218 in ( b1 ) and ( b2 ) , and the green dashed line shows the total background contribution from the secondary beams .
, title="fig : " ] -like ccqe events for @xmath60 ( @xmath214 ) @xmath215 gev at sk ( 22.5kton ) with @xmath1 off - axis angle and @xmath5pot exposure : ( a1 ) and ( a2 ) are for @xmath6 focusing beam , and ( b1 ) and ( b2 ) are for @xmath7 focusing beam ; whereas ( a1 ) and ( b1 ) are for the normal hierarchy , and ( a2 ) and ( b2 ) are for the inverted hierarchy .
the results are for @xmath182 , @xmath29 , and @xmath216 g/@xmath84 and the other input parameters of eq .
( [ eq : input ] ) .
the red solid line in each panel denotes the total event numbers , the purple dotted line shows the @xmath217 contribution in ( a1 ) and ( a2 ) , @xmath218 in ( b1 ) and ( b2 ) , the blue short dashed line gives the sum of the contributions from @xmath36 and @xmath217 in ( a1 ) and ( a2 ) , that of @xmath26 and @xmath218 in ( b1 ) and ( b2 ) , and the green dashed line shows the total background contribution from the secondary beams .
, title="fig : " ] -like ccqe events for @xmath60 ( @xmath214 ) @xmath215 gev at sk ( 22.5kton ) with @xmath1 off - axis angle and @xmath5pot exposure : ( a1 ) and ( a2 ) are for @xmath6 focusing beam , and ( b1 ) and ( b2 ) are for @xmath7 focusing beam ; whereas ( a1 ) and ( b1 ) are for the normal hierarchy , and ( a2 ) and ( b2 ) are for the inverted hierarchy .
the results are for @xmath182 , @xmath29 , and @xmath216 g/@xmath84 and the other input parameters of eq .
( [ eq : input ] ) .
the red solid line in each panel denotes the total event numbers , the purple dotted line shows the @xmath217 contribution in ( a1 ) and ( a2 ) , @xmath218 in ( b1 ) and ( b2 ) , the blue short dashed line gives the sum of the contributions from @xmath36 and @xmath217 in ( a1 ) and ( a2 ) , that of @xmath26 and @xmath218 in ( b1 ) and ( b2 ) , and the green dashed line shows the total background contribution from the secondary beams .
, title="fig : " ] the typical event numbers of the @xmath209-like ccqe events for @xmath60 ( @xmath214 ) @xmath215 gev at sk ( @xmath219 kton ) with the @xmath1 oab and @xmath5 pot exposure is shown in fig .
[ fig : t2k ] : ( a1 ) and ( a2 ) are for the @xmath6 focusing , and ( b1 ) and ( b2 ) are for the @xmath7 focusing beam ; ( a1 ) and ( b1 ) are for the normal , and ( a2 ) and ( b2 ) are for the inverted hierarchy .
these results are for @xmath182 , @xmath29 and @xmath220g/@xmath84 the other input parameters of eq .
( [ eq : input ] ) . in each panel
, the red solid line denotes the total event number , which is the sum of the signal events , @xmath26 for ( a1 ) and ( a2 ) , or @xmath36 for ( b1 ) and ( b2 ) , and the total background from the secondary beams shown by the green dashed line .
the purple dotted line shows the @xmath217 contribution in ( a1 ) and ( a2 ) , @xmath221 in ( b1 ) and ( b2 ) .
the blue short dashed line gives the sum of the contributions from @xmath222 and @xmath223 in ( a1 ) and ( a2 ) , and those from @xmath224 and @xmath225 in ( b1 ) and ( b2 ) . in fig .
[ fig : t2k ] , both the signal ( red solid minus green dash ) and the total ( red solid ) number of events peak in the first bin ( 400 mev @xmath100 600 mev ) , because both the @xmath1 oab fluxes and the oscillation probabilities are largest in the region ; see fig . [
fig : profile ] , ( a3 ) , ( p1 ) , and ( p2 ) for @xmath26 , ( b3 ) , ( p3 ) , and ( p4 ) for @xmath36 . the background levels are higher for the @xmath36 oscillation experiments than the @xmath26 case for both hierarchies .
this is partly because of the higher level of the secondary beam fluxes in the @xmath7 focusing beam @xcite , and partly because the ccqe cross sections are larger for @xmath226 than the @xmath227 ( @xmath228 ) @xcite . since the ratio of the @xmath36 to the @xmath229 event numbers is significantly larger for the inverted hierarchy than that for the normal hierarchy case in fig .
[ fig : t2k ] , one may be tempted to conclude that the neutrino mass hierarchy can be determined by using both @xmath6 and @xmath7 focusing beams at t2k .
this is not the case since the same trend can be expected for @xmath230 , as can be seen clearly from eq .
( [ eq : diff_ae ] ) . , but for tokai - to - oki island ( @xmath70 km ) , where the fiducial volume of 100 kton and the off - axis angle of @xmath3 is assumed for the far detector .
the results are for @xmath182 and @xmath29 , and @xmath231 g/@xmath84 , see eq .
( [ eq : matteroki ] ) .
, title="fig : " ] , but for tokai - to - oki island ( @xmath70 km ) , where the fiducial volume of 100 kton and the off - axis angle of @xmath3 is assumed for the far detector .
the results are for @xmath182 and @xmath29 , and @xmath231 g/@xmath84 , see eq .
( [ eq : matteroki ] ) .
, title="fig : " ] , but for tokai - to - oki island ( @xmath70 km ) , where the fiducial volume of 100 kton and the off - axis angle of @xmath3 is assumed for the far detector .
the results are for @xmath182 and @xmath29 , and @xmath231 g/@xmath84 , see eq .
( [ eq : matteroki ] ) .
, title="fig : " ] , but for tokai - to - oki island ( @xmath70 km ) , where the fiducial volume of 100 kton and the off - axis angle of @xmath3 is assumed for the far detector .
the results are for @xmath182 and @xmath29 , and @xmath231 g/@xmath84 , see eq .
( [ eq : matteroki ] ) .
, title="fig : " ] figure [ fig : t2oki ] shows the typical event numbers for the @xmath209-like ccqe event at oki island ( @xmath70 km ) , where we suppose to place a 100 kton fiducial volume detector .
the results are for @xmath5 pot at j - parc for both @xmath6 and @xmath7 focusing beam at the off - axis angle of @xmath3 , which corresponds to the @xmath1 oab at sk , as shown in fig .
[ fig : relation ] .
we use eqs .
( [ eq : input ] ) and ( [ eq : input_rct ] ) for physics parameters and eq .
( [ eq : matteroki ] ) for the average matter density to generate these event numbers .
the left two panels in fig .
[ fig : t2oki ] , ( a1 ) and ( a2 ) , are for the @xmath6 focusing beam , while the right two panels ( b1 ) and ( b2 ) are for the @xmath7 focusing beam .
the top two panels , ( a1 ) and ( b1 ) , are for the normal mass hierarchy , and the bottom two panels , ( a2 ) and ( b2 ) , are for the inverted mass hierarchy .
the line types are the same as in fig .
[ fig : t2k ] , the red solid lines give the total event numbers , the green dashed lines are the sum of all the background events from secondary beams , the blue short dashed lines give the sum of @xmath36 and @xmath217 for the @xmath6 focusing beam , that of @xmath229 and @xmath221 for the @xmath7 focusing beam . the purple dotted lines shows the contribution from @xmath218 ( @xmath217 ) for @xmath6 ( @xmath7 ) focusing beam . as in the case of sk shown in fig .
[ fig : t2k ] , the background from the secondary beam contributions are higher for the @xmath7 focusing beam than those for the @xmath6 focusing beam . in fig .
[ fig : t2oki ] , we find that the first oscillation peak appears at around 1.0 gev at oki island for all the four cases .
the peaks at @xmath232 gev in the event numbers are obtained by the convolution of the oscillation probability , whose first peak is located around @xmath233 gev for the normal or slightly above 1 gev for the inverted hierarchy , as shown in fig .
[ fig : profile ] ( p1)-(p4 ) , and the @xmath3 oab fluxes that have a peak at around @xmath234 gev in fig .
[ fig : profile ] ( a3 ) and ( b3 ) , as shown by green dashed lines .
the difference between the normal and inverted hierarchy is larger than that of the t2k experiment .
one can observe the second peak in the 400@xmath185600 mev bin for the normal hierarchy case , ( a1 ) and ( b1 ) , but not for the inverted case .
this is because the matter effect to the oscillation phase , the term @xmath126 in eqs .
( [ eq : pme ] ) and ( [ eq : beaprox ] ) , grows with the baseline length l , and shifts the peaks of the oscillation maximum at @xmath235 in the opposite directions ; toward higher ( lower ) energies for the normal ( inverted ) hierarchy .
more accurately speaking , the above phase shift pattern applies for @xmath236 , as can be read off from eq .
( [ eq : beaprox ] ) , where the matter effect diminishes ( enhances ) the shift for the normal ( inverted ) hierarchy .
the pattern reverses for @xmath237 .
likewise , the matter effects on the oscillation amplitudes are also clearly seen : we expect more ( less ) @xmath26 events than @xmath36 events around the first oscillation peak when the hierarchy is normal ( inverted ) as can be seen from eq .
( [ eq : aeaprox ] ) .
this pattern is enhanced when @xmath230 whereas it is diminished when @xmath238 . ,
but for korea ( @xmath12 km ) , where the fiducial volume of 100 kton and the off - axis angle of @xmath13 is assumed for the far detector .
the results are for @xmath182 and @xmath29 , and @xmath239 g/@xmath84 , see eq .
( [ eq : matterkr ] ) .
, title="fig : " ] , but for korea ( @xmath12 km ) , where the fiducial volume of 100 kton and the off - axis angle of @xmath13 is assumed for the far detector .
the results are for @xmath182 and @xmath29 , and @xmath239 g/@xmath84 , see eq .
( [ eq : matterkr ] ) .
, title="fig : " ] , but for korea ( @xmath12 km ) , where the fiducial volume of 100 kton and the off - axis angle of @xmath13 is assumed for the far detector .
the results are for @xmath182 and @xmath29 , and @xmath239 g/@xmath84 , see eq .
( [ eq : matterkr ] ) .
, title="fig : " ] , but for korea ( @xmath12 km ) , where the fiducial volume of 100 kton and the off - axis angle of @xmath13 is assumed for the far detector .
the results are for @xmath182 and @xmath29 , and @xmath239 g/@xmath84 , see eq .
( [ eq : matterkr ] ) .
, title="fig : " ] in figs .
[ fig : t2kk ] , we show the typical event numbers for @xmath209-like ccqe event at a far detector in korea ( @xmath12 km ) with a 100 kton fiducial volume detector . the off - axis angle there is chosen to be @xmath13 , which optimized the t2kk performance with the @xmath1 oab at sk @xcite .
the results are for @xmath240 and @xmath29 and the parameters of eqs .
( [ eq : input ] ) and ( [ eq : input_rct ] ) , with the average matter density of 2.9 g/@xmath84 eq .
( [ eq : matterkr ] ) .
the panels and the lines types are the same as those in figs .
[ fig : t2k ] and [ fig : t2oki ] .
the first peak of the event appears at around @xmath241 gev for the @xmath6 beam for both hierarchies , but that is around 1.8 ( 1.5 ) gev for the @xmath7 beam for the normal ( inverted ) hierarchy .
this is because of the matter effect contribution to the oscillation phase , @xmath126 in eq .
( [ eq : beaprox ] ) , and @xmath162 obtained from eq .
( [ eq : beaprox ] ) by reversing the sign of @xmath83 .
the magnitude of @xmath126 is small for @xmath242 in eq .
( [ eq : beaprox ] ) , whereas that of @xmath162 is enhanced .
as expected , the sign of the difference @xmath243 in eq .
( [ eq : diff_be ] ) reflects the neutrino mass hierarchy . likewise , the difference in the heights of the first oscillation peak is more distinct in fig .
[ fig : t2kk ] for @xmath12 km then that in fig .
[ fig : t2oki ] for @xmath70 km . as expected ,
the oscillation amplitude for the @xmath26 transition is bigger ( smaller ) than that for the @xmath36 transition when the hierarchy is normal ( inverted ) .
although the magnitudes of those enhancement or suppression factor depends on @xmath144 , as can be seen from eqs .
( [ eq : aeaprox ] ) and ( [ eq : diff_ae ] ) , the difference between different baseline lengths among figs .
[ fig : t2k ] , [ fig : t2oki ] , and [ fig : t2kk ] depend solely on the mass hierarchy pattern , as expressed in eq .
( [ eq : ae_diffs ] ) .
we introduce a @xmath244 function @xmath245 in order to compare the physics potential of t2k , t2ko , and t2kk experiments quantitatively on the same footing . the first term of eq .
( [ eq : def_chi1 ] ) gives statistical constraints on the model parameters from the number of the ccqe events in each bin at each detector : @xmath246 where @xmath247 and @xmath248 denotes the number of @xmath210- and @xmath209-like events , respectively , for the @xmath6 focusing beam calculated with eq .
( [ eq : eventnu ] ) in the @xmath193-th bin at each detector , d = sk , oki , kr , whereas @xmath249 and @xmath250 are for the @xmath7 focusing beam .
their square - roots give statistical errors .
the summation is over all bins from 0.4 gev to 5.0 gev for @xmath210-like events at all sites , and for @xmath209-like events from 0.4 gev to 1.2 gev at sk , from 0.4 gev to 2.4 gev at oki , from 0.4 gev to 2.8 gev at a far detector in korea , respectively .
the input event numbers , @xmath251 , are generated with eq .
( [ eq : input ] ) for both hierarchies , and for various value of @xmath47 and @xmath10 .
we use the average matter density along each baseline of eq .
( [ eq : matter ] ) when we calculate the input event numbers .
the event numbers in the fit , @xmath252 and @xmath253 , are obtained by varying all the 6 parameters of the three - neutrino model freely and also by allowing for the systematic errors .
we consider the following systematic errors in this analysis .
we assign @xmath169 uncertainty to the overall matter density along each baseline , @xmath254 although we expect positive correlation among @xmath255 , we allow them to vary independently as a conservative estimate .
we assign @xmath256 uncertainty in the flux normalization of @xmath6 and @xmath7 focusing beam as @xmath257 [ eq : fluxfit ] @xmath131 respectively , where d = sk , oki , kr .
we also ignore possible correlations among the flux errors .
for the ccqe cross sections of neutrino and anti - neutrinos , we assume common @xmath256 error as @xmath258 [ eq : xsecfit ] @xmath20 for neutrino and anti - neutrino events , independently , but we take @xmath259 and @xmath260 because of the @xmath209-@xmath210 universality . here also , we neglect the correlation between @xmath261 and @xmath262 cross section errors .
the systematic error for the fiducial volume of each far detector is also assigned as @xmath263 summing up , we take account of 32 systematic uncertainties in terms of which @xmath264 in eq .
( [ eq : def_chi1 ] ) is expressed as @xmath265\nn\\ & & + \sum_{\beta=\ell,\bar{\ell } } \left(\dfrac{1-f_\beta}{0.03}\right)^2\ , .
\label{eq : def_chi_sys}\end{aligned}\ ] ] there are 32 normalization factors since for each detector `` d '' the fiducial volume @xmath266 , the average matter density @xmath267 , and the 4 fluxes each for @xmath6 @xmath268 and @xmath7 @xmath269 focusing beam are accounted for , in addition to the overall theoretical uncertainties for @xmath226 @xmath270 and @xmath227 @xmath271 ccqe cross sections . the last term of eq .
( [ eq : def_chi1 ] ) , @xmath272 , accounts for the external constraints on the model parameters : @xmath273 the first two terms are from the present constraints from the kamland experiment @xcite listed in eq .
( [ eq : sun_data ] ) . in the last term , we assume that the new reactor experiments @xcite will measure @xmath47 with an uncertainty of 0.01 in the near future . we do not impose the present constraints on @xmath121 and @xmath274 given in eq .
( [ eq : atm_data ] ) , since the experiments studied in this report will measure them more accurately .
in this section , we show physics capability of the t2ko experiment to determine the mass hierarchy and compare it with that of the t2kk @xcite-@xcite and the 2kn experiment . here by 2kn , we examine the option where the additional 100 kton detector is placed in the kamioka site to make the total fiducial volume 122 kton at @xmath0 km , which may be regarded as a small scale version of hyper - kamiokande @xcite . of the t2kk , t2ko , and 2kn
experiment to exclude the wrong mass hierarchy when only the @xmath6 focusing beam of @xmath275 pot is used .
four columns give results for @xmath29 , @xmath18 , @xmath276 , and @xmath277 , respectively , and the horizontal axis in each column gives the off - axis angle at sk .
the solid ( open ) square , circle , and triangle denotes , respectively , the t2kk , t2ko and 2kn results for the normal ( inverted ) hierarchy .
the results are for @xmath182 and the other input model parameters in eq .
( [ eq : input ] ) , as well as the average matter density of eq .
( [ eq : matter ] ) along the three baselines . ] in fig .
[ fig : mass1 ] , we show the minimum @xmath244 of t2kk , t2ko , and 2kn experiment for the @xmath6 focusing beam with @xmath275 pot .
four columns of fig .
[ fig : mass1 ] give results for @xmath29 , @xmath18 , @xmath276 , and @xmath277 , respectively , and the horizontal axis in each column gives the off - axis angle at the sk .
the solid ( open ) square , circle , and triangle denotes , respectively , the sensitivity of the t2kk , t2ko , and 2kn experiments for the normal ( inverted ) hierarchy .
the results are for @xmath182 and the other input model parameters in eq .
( [ eq : input ] ) , as well as the average matter density of eq .
( [ eq : matter ] ) along the three baselines . from fig .
[ fig : mass1 ] , we can tell that the physics potential for the mass hierarchy determination of the t2kk experiment is far better than the other experiments when the combination of @xmath1 oab at sk and @xmath13 oab at a far detector in korea is taken , where the mass hierarchy can be determined by more than 4@xmath8 level for all the 8 cases ( 4 values of @xmath278 and both hierarchies ) .
the sensitivity of the t2kk on the mass hierarchy reduces significantly as the off - axis angle at sk is reduced .
this is because oab with small off - axis angle can not reach korea when off - axis angle at sk is below @xmath279 @xcite ; see fig . [ fig : relation ] . for the t2ko experiment ,
the mass hierarchy can be determined at the 2@xmath8 level or higher .
this capability does not depend strongly on the off - axis angle at sk , because the beam intensity around the first peak ( @xmath232gev ) at oki island does not change much with the oab at sk , see fig .
[ fig : profile ] .
furthermore , it is clearly shown in fig . [
fig : mass1 ] that the 2kn experiment does not have any sensitivity to the neutrino mass hierarchy pattern for any combinations of @xmath10 and the mass hierarchy .
this is essentially because the small differences in the oscillation probabilities between the normal and the inverted hierarchy can easily be compensated by small shifts in the model parameters , such as @xmath121 , @xmath120 , and @xmath10 . , but with both @xmath6 and @xmath7 focusing beams each with @xmath5 pot . ] in fig .
[ fig : mass2 ] , we show the minimum @xmath244 of the t2kk , t2ko , and 2kn experiment to exclude the wrong mass hierarchy when both @xmath6 and @xmath7 focusing beams are used , each with @xmath5 pot to keep the total exposure the same . as for the t2kk experiment , shown by the red squares ,
the improvement is significant especially when @xmath280 and @xmath18 , making @xmath15 greater than 20 for all the 8 combinations of @xmath10 and sgn@xmath281 , not only for the @xmath1 oab but also for the @xmath72 oab at sk @xcite .
likewise for the t2ko experiment , shown by green circles , the improvement is most significant at @xmath282 where the smallest @xmath15 grows from @xmath283 in fig .
[ fig : mass1 ] to @xmath284 in fig .
[ fig : mass2 ] .
this is essentially because the @xmath144 term in the oscillation amplitude shift @xmath125 in eq .
( [ eq : aeaprox ] ) changes sign for the @xmath36 oscillation as shown in eq .
( [ eq : diff_ae ] ) for the difference @xmath285 .
this helps resolving the entanglement between the @xmath144 dependence ad the matter effect in the oscillation amplitudes .
significant improvements are also found for @xmath286 by the same reason .
moderate improvements are found for the @xmath29 case , allowing the t2ko experiment with half - and - half @xmath6 and @xmath7 beams to resolve the mass hierarchy at @xmath66 level for the worst case ( @xmath287 and normal hierarchy ) .
no significant dependence on the oab at sk is found .
the capability of determining the mass hierarchy pattern by the 2kn experiment does not appear even by using both @xmath6 and @xmath7 beams .
the value of @xmath15 stays almost zero , except for just two special combinations ; @xmath286 for the normal hierarchy and @xmath282 for the inverted hierarchy .
these are the two cases , where the difference between the @xmath26 and @xmath36 oscillation amplitudes , @xmath288 in eq .
( [ eq : diff_ae ] ) is largest or smallest , respectively , such that variation of the other model parameters can not account for the difference if the wrong mass hierarchy is assumed . with the same token , @xmath15 exceeds 20 for t2ko or 30 for t2kk ( with @xmath289 oab at sk ) for these two particular combinations .
contour plot for the t2ko experiment to exclude the wrong mass hierarchy in the plane of @xmath47 and @xmath10 .
the left figure is for the normal hierarchy and the right one is for the inverted hierarchy .
the oab combination for both figures is @xmath1 at sk and @xmath3 at oki island with @xmath5 pot for both @xmath6 and @xmath7 focusing beams .
contours for @xmath290 , 9 , 16 , 25 , 36 , 49 are shown .
all the input parameters other than @xmath47 and @xmath10 are shown in eqs .
( [ eq : matter ] ) and ( [ eq : input ] ) .
, title="fig : " ] contour plot for the t2ko experiment to exclude the wrong mass hierarchy in the plane of @xmath47 and @xmath10 .
the left figure is for the normal hierarchy and the right one is for the inverted hierarchy .
the oab combination for both figures is @xmath1 at sk and @xmath3 at oki island with @xmath5 pot for both @xmath6 and @xmath7 focusing beams .
contours for @xmath290 , 9 , 16 , 25 , 36 , 49 are shown .
all the input parameters other than @xmath47 and @xmath10 are shown in eqs .
( [ eq : matter ] ) and ( [ eq : input ] ) .
, title="fig : " ] so far , we have shown results for @xmath240 and 4 representative values of @xmath10 ; @xmath291 , @xmath276 , and @xmath292 . in fig . [
fig : cnt - t2oki ] , we show the contour plot of the @xmath15 for the t2ko experiment to exclude the wrong mass hierarchy in the whole plane of @xmath47 and @xmath10 : the left figure ( a ) is for the normal hierarchy , whereas the right one ( b ) is for the inverted mass hierarchy . since no strong dependence of @xmath15 on the oab angle is found for t2ko potential in fig .
[ fig : cnt - t2oki ] , we choose @xmath1 oab at sk , which gives @xmath3 oab at oki , as shown in fig .
[ fig : relation ] .
the results are for @xmath5 pot each for @xmath6 and @xmath7 focusing beam .
contours in each figure are for @xmath290 , 9 , 16 , 25 , 36 , and 49 .
the input parameters other than @xmath47 and @xmath10 are shown in eqs .
( [ eq : matter ] ) and ( [ eq : input ] ) , exactly the same as those adopted in fig .
[ fig : mass2 ] .
accordingly the @xmath15 values at @xmath182 agree exactly with those presented in fig .
[ fig : mass2 ] for t2ko with @xmath1 oab at sk , for the 4 representative @xmath10 values .
it is clearly seen from fig .
[ fig : cnt - t2oki ] that the mass hierarchy pattern can be distinguished at @xmath66 if @xmath293 for any value of @xmath294 and for both hierarchies .
the most difficult case is found for @xmath295 for the normal hierarchy , while @xmath296 for the inverted hierarchy . on the other hand ,
the discrimination is easiest at @xmath297 for the normal and at @xmath298 for the inverted hierarchy , in accordance with the argument presented above for fig .
[ fig : mass2 ] .
in addition , the contour plots fig .
[ fig : cnt - t2oki ] identify another case at @xmath299 ( @xmath300 ) for the normal ( inverted ) hierarchy , where the difference between the right and the wrong hierarchy is large , giving high @xmath15 .
the spikes of the contours around these @xmath10 values appear as a consequence of the conspiracy between the mass hierarchy dependences in the oscillation amplitude @xmath125 and in the phase shift term @xmath126 .
when @xmath301 , the @xmath26 oscillation amplitude shift @xmath125 cancels between the matter effect term and the @xmath144 term for the normal ( inverted ) hierarchy at around the tokai - to - oki baseline ; see eq .
( [ eq : aeaprox ] ) at @xmath70 km .
the cancellation is not significant at @xmath0 km for t2k , and the two effects add up constructively for @xmath36 oscillation . when the wrong hierarchy is assumed the best fit
is found for @xmath302 , for which there are two solutions , @xmath303 and @xmath304 , which have significantly different phase - shift ; see eq .
( [ eq : beaprox ] ) .
we find that the spike around @xmath305 in fig .
[ fig : cnt - t2oki ] occurs when the @xmath15 solution of the wrong hierarchy model jumps from @xmath303 to @xmath304 . , but for t2kk experiment with the optimum oab combination , @xmath1 oab at sk and @xmath13 oab at @xmath12 km .
@xmath15 values are given along the contours .
, title="fig : " ] , but for t2kk experiment with the optimum oab combination , @xmath1 oab at sk and @xmath13 oab at @xmath12 km .
@xmath15 values are given along the contours .
, title="fig : " ] figure [ fig : cnt - t2kk ] shows the same contour plots as fig .
[ fig : cnt - t2oki ] , but for t2kk with @xmath1 oab at sk and @xmath13 oab at @xmath12 km .
significant increase in the @xmath15 values at t2kk is clearly seen against those in fig .
[ fig : cnt - t2oki ] for t2ko .
now , the wrong mass hierarchy can be excluded at @xmath33 level for @xmath306 ( 0.09 ) if the mass hierarchy is normal ( inverted ) . because the measurement error is dominated by statistics , we find that the @xmath66 sensitivity of t2ko with 100 kton detector can be archived with a 40 kton detector for t2kk
. the cp phase dependence of the t2kk contours is much weaker than that of the t2ko contours , especially for smaller @xmath47 .
this is simply because the matter effect terms at @xmath307 km dominate over the @xmath144 and @xmath117 terms , in the correction terms @xmath125 and @xmath126 ; see eq .
( [ eq : abeapprox ] ) .
contour plot for the t2ko experiment in the plane of @xmath47 and @xmath10 when the mass hierarchy is assumed to be normal ( left ) or inverted ( right ) . allowed regions in the plane of @xmath47 and @xmath10
are shown for the combination of @xmath1 oab at sk and @xmath3 at oki island with @xmath5 pot each for @xmath6 and @xmath7 focusing beams .
the input values of @xmath47 is 0.04 , 0.08 , and 0.12 and @xmath10 is @xmath291 , @xmath18 , @xmath276 , and @xmath277 .
the other input parameters are given in eqs .
( [ eq : matter ] ) and ( [ eq : input ] ) .
the dotted - lines , dashed - lines , and solid - lines show @xmath308 , 4 , and 9 respectively .
the shaded region has `` mirror '' solutions for the wrong mass hierarchy giving @xmath309 .
, title="fig : " ] contour plot for the t2ko experiment in the plane of @xmath47 and @xmath10 when the mass hierarchy is assumed to be normal ( left ) or inverted ( right ) . allowed regions in the plane of @xmath47 and @xmath10
are shown for the combination of @xmath1 oab at sk and @xmath3 at oki island with @xmath5 pot each for @xmath6 and @xmath7 focusing beams .
the input values of @xmath47 is 0.04 , 0.08 , and 0.12 and @xmath10 is @xmath291 , @xmath18 , @xmath276 , and @xmath277 .
the other input parameters are given in eqs .
( [ eq : matter ] ) and ( [ eq : input ] ) .
the dotted - lines , dashed - lines , and solid - lines show @xmath308 , 4 , and 9 respectively .
the shaded region has `` mirror '' solutions for the wrong mass hierarchy giving @xmath309 .
, title="fig : " ] in this section , we investigate the measurement of cp phase @xmath10 in the t2ko experiment , as compared to the t2kk and 2kn options . in all the cases we adopt @xmath1 oab at sk , which makes the sk contribution to the measurements exactly the same , and assume @xmath5 pot each for @xmath6 and @xmath7 focusing beam .
we show in fig .
[ fig : cp - t2ko ] the @xmath15 contour plots for the t2ko experiment in the plane of @xmath47 and @xmath10 , when the mass hierarchy is assumed to be normal ( left ) or inverted ( right ) .
the 12 cases are shown in each figure for @xmath29 , @xmath18 , @xmath276 , @xmath277 , and for @xmath310 , 0.08 , and 0.12 .
the other input parameters are given in eqs .
( [ eq : matter ] ) and ( [ eq : input ] ) .
the allowed regions in the plane of @xmath47 and @xmath10 are inside of the dotted- , dashed- , and solid - contours at @xmath308 , 4 , and 9 , respectively .
the shaded region has `` mirror '' solutions for the wrong mass hierarchy giving @xmath311 , as shown by the red solid contours in fig .
[ fig : cnt - t2oki ] . since the @xmath310 input cases are no longer relevant after the measurements eqs .
( [ eq : dayabay_s13 ] ) and ( [ eq : reno_s13 ] ) by dayabay @xcite and reno@xcite , respectively , the only parameter regions where we should worry about the mirror solution with the wrong hierarchy are around @xmath280 at @xmath312 for the normal hierarchy , and around @xmath313 at @xmath312 for the inverted hierarchy .
since these regions are near the @xmath66 boundary , the mirror solutions may be excluded by extending the experimental period or by enhancing the beam power .
the @xmath310 input cases are kept , since they show the independence of the @xmath10 measurement error on @xmath47 clearly . as explained in ref.@xcite ,
this independence is a consequence of the @xmath314 enhancement of the @xmath144 and @xmath117 dependencies in @xmath125 and @xmath126 , respectively , in eqs .
( [ eq : aeaprox ] ) and ( [ eq : beaprox ] ) , which cancels precisely the statistical error which is proportional to @xmath315 , or the square - root of the @xmath26 and @xmath36 event numbers .
it is clearly seen that @xmath10 can be measured with @xmath11 error for all the 24 cases presented in fig .
[ fig : cp - t2ko ] ( a ) and ( b ) .
this is essentially because the magnitude of the coefficient of @xmath144 in the amplitude shift in eq .
( [ eq : aeaprox ] ) and that of @xmath117 in the phase shift in eq .
( [ eq : beaprox ] ) are approximately equal .
it should be noted that the uncertainty in the @xmath47 is dictated by the external constraint with the error of @xmath316 on the @xmath77 function eq .
( [ eq : def_chi1 ] ) . because of the nearly zero correlation between the errors of @xmath10 and @xmath47 in fig .
[ fig : cp - t2ko ] , further improvements in the precise measurements of @xmath47 will not reduce the errors of @xmath10 significantly . , but for t2kk experiment with @xmath1 oab at sk and @xmath13 oab at @xmath12 km .
, title="fig : " ] , but for t2kk experiment with @xmath1 oab at sk and @xmath13 oab at @xmath12 km .
, title="fig : " ] in order to compare the sensitivity of the @xmath10 measurement between the t2ko and the t2kk experiments , we show in fig .
[ fig : cp - t2kk ] the same contours for the t2kk experiment with @xmath1 oab at sk and @xmath13 oab at @xmath12 km .
it is clearly seen from the 12 sets of contours , each for normal ( left ) and inverted ( right ) hierarchy , that the expected error of @xmath10 is @xmath11 for all the combinations , just as for the t2ko experiment shown in fig .
[ fig : cp - t2ko ] .
this is remarkable since the event number at a far detector in korea at @xmath317 km is significantly smaller than that in oki at @xmath70 km because of the flux which decreases as @xmath203 at long distances .
this decrease the overall flux is compensated by the wide - band structure of the @xmath13 oab as shown by blue dotted lines in fig .
[ fig : profile ] ( a3 ) and ( b3 ) , which enables the far detector in korea to observe not only the first oscillation peak but also the second one as in fig .
[ fig : t2kk ] . around the second peak , @xmath318 , and
the sensitivity to @xmath144 and @xmath117 can be three times higher than the first peak with @xmath319 ; see eqs .
( [ eq : abeapprox ] ) and ( [ eq : diff_ae_and_be ] ) .
in addition , the extended energy range covered by the @xmath13 oab allows the t2kk experiment to measure the phase - shifts @xmath126 and @xmath162 accurately , and hence @xmath117 ; see eqs .
( [ eq : beaprox ] ) and ( [ eq : diff_be ] ) .
indeed , we notice in fig .
[ fig : cp - t2kk ] ( a ) and ( b ) that the error of @xmath10 can be as small as @xmath320 when @xmath29 or @xmath276 . contour plot for the 2kn experiment in the plane of @xmath47 and @xmath10 when the mass hierarchy is assumed to be normal ( @xmath321 ) . allowed regions in the plane of @xmath47 and @xmath10
are shown for experiments with @xmath5 pot each for @xmath6 and @xmath7 focusing beam at @xmath1 off - axis angle .
the input values of @xmath47 are 0.04 , 0.08 , and 0.12 and @xmath10 are @xmath291 ( a ) , @xmath18 ( b ) , @xmath276 ( c ) , and @xmath277 ( d ) .
the other input parameters are listed in eqs .
( [ eq : input ] ) and ( [ eq : matter ] ) .
the red dotted - lines , dashed - lines , and solid - lines show @xmath308 , 4 , and 9 contours , respectively , when the right mass hierarchy is assumed in the fit , whereas the blue contours give @xmath15 measured from the local minimum value ( shown besides the @xmath322 symbol ) at the cross point when the wrong hierarchy is assumed in the fit .
, title="fig : " ] contour plot for the 2kn experiment in the plane of @xmath47 and @xmath10 when the mass hierarchy is assumed to be normal ( @xmath321 ) . allowed regions in the plane of @xmath47 and @xmath10
are shown for experiments with @xmath5 pot each for @xmath6 and @xmath7 focusing beam at @xmath1 off - axis angle .
the input values of @xmath47 are 0.04 , 0.08 , and 0.12 and @xmath10 are @xmath291 ( a ) , @xmath18 ( b ) , @xmath276 ( c ) , and @xmath277 ( d ) .
the other input parameters are listed in eqs .
( [ eq : input ] ) and ( [ eq : matter ] ) .
the red dotted - lines , dashed - lines , and solid - lines show @xmath308 , 4 , and 9 contours , respectively , when the right mass hierarchy is assumed in the fit , whereas the blue contours give @xmath15 measured from the local minimum value ( shown besides the @xmath322 symbol ) at the cross point when the wrong hierarchy is assumed in the fit .
, title="fig : " ] contour plot for the 2kn experiment in the plane of @xmath47 and @xmath10 when the mass hierarchy is assumed to be normal ( @xmath321 ) . allowed regions in the plane of @xmath47 and @xmath10
are shown for experiments with @xmath5 pot each for @xmath6 and @xmath7 focusing beam at @xmath1 off - axis angle .
the input values of @xmath47 are 0.04 , 0.08 , and 0.12 and @xmath10 are @xmath291 ( a ) , @xmath18 ( b ) , @xmath276 ( c ) , and @xmath277 ( d ) .
the other input parameters are listed in eqs .
( [ eq : input ] ) and ( [ eq : matter ] ) .
the red dotted - lines , dashed - lines , and solid - lines show @xmath308 , 4 , and 9 contours , respectively , when the right mass hierarchy is assumed in the fit , whereas the blue contours give @xmath15 measured from the local minimum value ( shown besides the @xmath322 symbol ) at the cross point when the wrong hierarchy is assumed in the fit .
, title="fig : " ] contour plot for the 2kn experiment in the plane of @xmath47 and @xmath10 when the mass hierarchy is assumed to be normal ( @xmath321 ) . allowed regions in the plane of @xmath47 and @xmath10 are shown for experiments with @xmath5 pot each for @xmath6 and @xmath7 focusing beam at @xmath1 off - axis angle .
the input values of @xmath47 are 0.04 , 0.08 , and 0.12 and @xmath10 are @xmath291 ( a ) , @xmath18 ( b ) , @xmath276 ( c ) , and @xmath277 ( d ) .
the other input parameters are listed in eqs .
( [ eq : input ] ) and ( [ eq : matter ] ) .
the red dotted - lines , dashed - lines , and solid - lines show @xmath308 , 4 , and 9 contours , respectively , when the right mass hierarchy is assumed in the fit , whereas the blue contours give @xmath15 measured from the local minimum value ( shown besides the @xmath322 symbol ) at the cross point when the wrong hierarchy is assumed in the fit . , title="fig : " ] , but for the inverted mass hierarchy ( @xmath323 ) .
, title="fig : " ] , but for the inverted mass hierarchy ( @xmath323 ) .
, title="fig : " ] , but for the inverted mass hierarchy ( @xmath323 ) .
, title="fig : " ] , but for the inverted mass hierarchy ( @xmath323 ) .
, title="fig : " ] finally in figs .
[ fig : cp - t2kn ] and [ fig : cp - t2ki ] , we show the allowed regions by the 2kn experiment in the plane of @xmath47 and @xmath10 for the normal ( fig . [
fig : cp - t2kn ] ) and the inverted ( fig .
[ fig : cp - t2ki ] ) mass hierarchy , when both @xmath6 and @xmath7 focusing beam at @xmath1 off - axis angle are used each with @xmath5 pot .
the four @xmath10 cases of @xmath291(a ) , @xmath18(b ) , @xmath276(c ) , @xmath277(d ) are examined for the three @xmath324 , 0.08 , 0.12 , just as in fig .
[ fig : cp - t2ko ] for t2ko and fig .
[ fig : cp - t2kk ] for t2kk .
the other input parameters are also the same , taken from eqs .
( [ eq : matter ] ) and ( [ eq : input ] ) .
the red dotted - lines , dashed - lines , and solid - lines show @xmath308 , 4 , 9 contours , respectively , when the right mass hierarchy is assumed in the fit , whereas the blue contours gives @xmath15 measured from the local minimum at the blue cross point when the wrong hierarchy is assumed in the fit .
the local minimal value of @xmath15 at the blue cross point is given besides the cross mark in the corresponding input @xmath47 column .
as has been explained in section [ sec : mass ] and shown in fig .
[ fig : mass2 ] , the local @xmath15 values are significant only for @xmath325 when the hierarchy is normal ( fig .
[ fig : cp - t2kn ] ( d ) ) , and for @xmath326 when the hierarchy is inverted ( fig .
[ fig : cp - t2ki ] ( b ) ) . in order to show the location of the input parameters clearly
, we show the global minimal point by the red solid blob for each input parameter case .
the global minimum gives @xmath327 in our analysis which ignores fluctuations in the mean number of events in each bin .
we first note that @xmath10 can be constrained uniquely around the above two specific cases ( @xmath328 for @xmath57 , or @xmath329 for @xmath58 ) , since not only the wrong mass hierarchy assumption gives non - negligible local @xmath15 as given in fig .
[ fig : cp - t2kn ] ( d ) and fig .
[ fig : cp - t2ki ] ( b ) , but also the wrong hierarchy assumption favors the right @xmath10 , with slightly larger ( smaller ) fitted @xmath47 that compensate for the matter effect for the normal ( inverted ) hierarchy . the @xmath330 error shown by the red dotted contour is rather large , however , about @xmath331 . in all the other cases ,
the presence of the wrong hierarchy solutions as shown by blue contours significantly reduce the capability of measuring @xmath10 with 2kn .
more importantly , even if we can remove the wrong hierarchy by other experiments such as nova @xcite , the next generation reactor neutrino oscillation experiments @xcite , or by an atmospheric neutrino observation with a huge detector @xcite , the 2kn experiment with only one baseline length can not measure @xmath10 with high accuracy when @xmath332 , or suffers from the @xmath333 solution when @xmath334 or @xmath276 as can be seen from the separate red contour islands on the ( a ) and ( c ) plots in figs .
[ fig : cp - t2kn ] and [ fig : cp - t2ki ] . throughout the analysis of this and the previous sections we fix the oab at sk at @xmath1 in order to make the sk contributions to all the three experiments t2ko , t2kk and 2kn identical .
we find that the performance of 2kn slightly improves if the @xmath335 oab is adopted instead , mainly because these fluxes are slightly wider ( harder ) than the @xmath1 oab as shown by red solid curves in figs .
[ fig : profile ] ( a1)-(a3 ) , ( b1)-(b3 ) .
in this paper , we examine physics potential of a one - beam two - detectors neutrino oscillation experiment with an additional 100 kton water detector in oki island , which is located along the t2k beam line at the baseline length of @xmath70 km . together with super - kamiokande ( sk ) at @xmath0 km
, we can measure neutrino oscillations at two different energies for the same oscillation phase proportional to @xmath336 .
we may call this proposal as t2ko ( tokai - to - kamioka - and - oki ) , whose capability has been compared with t2kk ( tokai - to - kamioka - and - korea ) with @xmath317 km for the far detector in korea , and also with 2kn where the same 100 kton detector is placed at the sk site ( @xmath0 km ) . as shown in fig .
[ fig : t2oki1 ] , since the oki island is located in the east side of the t2k beam center , just like the sk , the off - axis angle at oki island increases as that at sk increases , as shown in fig .
[ fig : relation ] . the off - axis beam ( oab ) with @xmath3 , @xmath337 , and @xmath4 from j - parc can be observed at oki island , when the @xmath1 , @xmath72 , and @xmath2 oab reaches at sk , respectively .
the neutrino energy of the first oscillation maximum for the @xmath338 and @xmath339 is between 1.0 gev and 1.5 gev , which depend on the cp phase and the mass hierarchy pattern , at @xmath340 km . since the @xmath6 ( @xmath7 ) beam between @xmath4 and @xmath3 off - axis angles has significant intensity around these energies , as shown by green dashed lines in the upper six panels in fig .
[ fig : profile ] , we expect that the t2ko experiment can be sensitive to the neutrino mass hierarchy and the cp phase , just like the t2kk experiment @xcite , @xcite-@xcite . for a detector of 100 kton fiducial volume and @xmath5 pot exposure each for both @xmath6 and @xmath7 beams ,
we find that the t2ko experiment can determine the mass hierarchy pattern at @xmath66 level if @xmath341 is larger than 0.09 , by observing the @xmath209-like ccqe ( charged - current quasi elastic ) events ; see fig . [ fig : cnt - t2oki ] . this result does not strongly depend on the off - axis angle of the @xmath6 ( @xmath7 ) beam at oki island , because the neutrino intensity at the first oscillation maximum does not strongly depend on the off - axis angle at oki island .
the t2ko sensitivity to the mass hierarchy is about @xmath14 ( in @xmath15 ) of the t2kk experiment with the optimum oab combination of @xmath1 at sk and @xmath13 at a far detector in korea with the baseline length around 1000 km .
this is because the factor of two higher sensitivity of the t2kk over the t2ko experiment as shown by eqs .
( [ eq : ae_diffs ] ) and ( [ eq : be_diffs ] ) , which should give a factor of 4 in @xmath15 is partially compensated by the smaller average flux by a factor of ( 635 km/1000 km)@xmath342 at a far detector in korea .
the sensitivity of the mass hierarchy pattern of the 2kn experiment , where a 100 kton detector is added at the sk location , is almost zero , except around @xmath297 ( @xmath18 ) for the normal ( inverted ) hierarchy ; see fig .
[ fig : mass2 ] . the cp phase in the mns ( maki - nakagawa - sakata ) lepton flavor mixing matrix @xcite , @xmath10 can be measured with @xmath11 error for all the four cases at @xmath29 , @xmath343 , and @xmath276 , almost independent of the @xmath47 values @xcite as long as the neutrino mass hierarchy is determined .
this is because @xmath144 can be constrained by the difference between the magnitudes of the oscillation probabilities @xmath338 and @xmath339 around the oscillation maximum , whereas @xmath117 can be determined by the oscillation phase around the first oscillation maximum , or the location of the oscillation peak(s ) ; see eqs .
( [ eq : aeaprox ] ) and ( [ eq : diff_ae ] ) .
the sensitivity to the cp phase , @xmath10 , of the t2ko experiment is similar to that of the t2kk experiment , mainly because the smallness of the flux at t2kk is compensated by its capability to measure the second oscillation peak at @xmath344 when the sensitivity to both @xmath117 and @xmath144 is a factor of 3 higher than that around the first peak ; see eqs .
( [ eq : abeapprox ] ) .
the 2kn option , which may be regarded as a first step toward the hyper - kamiokande @xcite , can not generally determine @xmath10 , mainly because it can not resolve mass hierarchy by itself .
only when @xmath345 for the normal hierarchy ( see , fig .
[ fig : cp - t2kn ] ( d ) ) and when @xmath329 for the inverted hierarchy ( see , fig .
[ fig : cp - t2ki ] ( b ) ) , the constraints for both hierarchy assumptions overlap , and the cp phase can be determined uniquely . even if the mass hierarchy is determined by other experiments @xcite ,
@xcite-@xcite the sensitivity to @xmath10 is rather poor at 2kn as shown in figs .
[ fig : cp - t2kn ] and [ fig : cp - t2ki ] .
this is essentially because of the parameter degeneracy unavoidable in experiments with only one baseline length , such as those between @xmath10 and @xmath346 when @xmath347 .
we would like to thank our experimentalist colleagues y. hayato , a.k .
ichikawa , t. kobayashi , and t. nakaya , from whom we learn about the k2k and t2k experiments .
we thank n. isezaki and m. komazawa for teaching us about the geophysics measurements in the sea of japan , or the east sea of korea .
we are also grateful to japanese coast guard for showing the detailed geological information around oki island .
the numerical calculations have been carried out on kekcc at kek .
homestake collaboration , astro .
j. * 496 * 505 ( 1998 ) ; sage collaboration , [ arxiv : astro - ph/9907113 ] ; j. exp .
* 95 * 181 ( 2002 ) [ zh .
. fiz . * 95 * , 211 ( 2002 ) ] [ arxiv : astro - ph/0204245 ] ; gallex collaboration , ; super - kamiokande collaboration , ; erratum _ ibid _ * 81 * 4279 , ( 1998 ) [ arxiv : hep - ex/9805021 ] ; [ arxiv:0803.4312(hep - ex ) ] ; gno collaboration , [ arxiv : hep - ex/0006034 ] ; [ arxiv : hep - ex/0504037 ] ; borexino collaboration , [ arxiv:0708.2251(astro - ph ) ] ; [ arxiv:0805.3843(astro - ph ) ] ; arxiv:0808.2868(astro - ph ) .
sno collaboration , [ arxiv : nucl - ex/0106015 ] ; [ arxiv : nucl - ex/0204008 ] ; [ arxiv : nucl - ex/0309004 ] ; [ arxiv : nucl - ex/0502021 ] ; [ arxiv : nucl - ex/0610020 ] ; [ arxiv:0806.0989 [ nucl - ex ] ] ; [ arxiv:0910.2984 [ nucl - ex ] ] ; arxiv:1109.0763 [ nucl - ex ] .
l. wolfenstein , ; r.r .
lewis , _ ibid . _ * d21 * , 663 ( 1980 ) ; v. barger , s. pakvasa , r.j.n .
phillips , and k. whisnant , _ ibid . _ * d22 * , 2718 ( 1980 ) ; s.p . mikheyev and a.yu .
smirnov , yad .
fiz . * 42 * , 1441 ( 1985 ) [ sov.j.nucl.phys.*42 * , 913 ( 1986 ) ] ; nuovo cimento * c9 * , 17 ( 1986 ) .
fogli , e. lisi , a. marrone , and a. palazzo , prog .
. phys . * 57 * 742 ( 2006 ) [ arxiv : hep - ph/0506083 ] ; m. maltoni , t. schwetz , m.a .
tortola , and j.w.f .
valle , new j. phys .
* 6 * , 122 ( 2004 ) [ arxiv : hep - ph/0405172 ] ; t. schwetz , m. tortola , and j.w.f .
valle , new j. phys . * 10 * , 113011 ( 2008 ) [ arxiv:0808.2016(hep - ph ) ] .
k. hagiwara , talk at fujihara seminar on neutrino mass and seesaw mechanism ( seesaw 1979 - 2004 ) , tsukuba , ibaraki , japan , 23 - 25 feb .
2004 , published in nucl .
* 137 * 84 ( 2004 ) [ arxiv : hep - ph/0410229 ] .
ichikawa , private communication ; the flux data for various off - axis angles are available from the web page ; http://www2.yukawa.kyoto-u.ac.jp/~okamura/t2kk/ , some beam profiles are obtained with interpolation by ourselves .
petcov and t. schwetz , ; r. gandhi , p. ghoshal , s. goswami , p. mehta , s.u .
sankar , and s. shalgar , ; r. gandhi , p. ghoshal , s. goswami , and s.u .
sankar , ; o. mena , i. mocioiu , and s. razzaque , ; a. samanta , ; g. giordano , o. mena , and i. mocioiu , ; e. fernandez - martinez , g. giordano , o. mena , and i. mocioiu , ; v. barger , r. gandhi , p. ghoshal , s. goswami , d. marfatia , s. prakash , s.k .
raut , and s.u .
sankar , ; e.kh .
akhmedov , s. razzaque , and a.yu .
smirnov , arxiv:1205.7071(hep - ph ) . | oki island is located between japan and korea along the tokai - to - kamioka ( t2k ) baseline .
the distance from j - parc to oki island is about 653 km , which is twice that of the t2k experiment ( @xmath0 km ) .
when the off - axis angle of the neutrino beam from j - parc is @xmath1 ( @xmath2 ) at super - kamiokande ( sk ) , the off - axis beam ( oab ) with @xmath3 ( @xmath4 ) reaches at oki island .
we examine physics case of placing a far detector in oki island during the t2k experimental period .
we estimate the matter density profile along the tokai - to - oki baseline by using recent seismological measurements . for a detector of 100 kton
fiducial volume and @xmath5 pot ( protons on target ) exposure for both @xmath6 and @xmath7 beams , we find that the mass hierarchy pattern can be distinguished at 3 @xmath8 level if @xmath9 , by observing the electron - like ccqe ( charged - current quasi elastic ) events . the cp phase in the maki - nakagawa - sakata lepton flavor mixing matrix , @xmath10 , can be constrained with @xmath11 . as a reference , we repeat the same analysis by placing the same detector in korea at @xmath12 km and oab=@xmath13 ( t2kk ) and also by placing it at the sk site ( 2kn ) .
the tokai - to - kamioka - oki ( t2ko ) sensitivity to the mass hierarchy is about @xmath14 ( in @xmath15 ) of t2kk , while the sensitivity to the phase @xmath10 is similar between t2ko and t2kk .
the t2k@xmath16 option has almost no sensitivity to the mass hierarchy , and can not measure the cp phase except when @xmath17 ( @xmath18 ) for the normal ( inverted ) hierarchy .
kek - th-1568 |
the theory of hilbert spaces plays a central role in contemporary mathematics with numerous applications for linear operators , partial differential equations , in nonlinear analysis , approximation theory , optimization theory , numerical analysis , probability theory , statistics and other fields .
the schwarz , triangle , bessel , gram and most recently , grss type inequalities have been frequently used as powerful tools in obtaining bounds or estimating the errors for various approximation formulae occurring in the domains mentioned above .
therefore , any new advancement related to these fundamental facts will have a flow of important consequences in the mathematical fields where these inequalities have been used before .
the main aim of this monograph is to survey some recent results obtained by the author related to reverses of the schwarz , triangle and bessel inequalities .
some grss type inequalities for orthonormal families of vectors in real or complex inner product spaces are presented as well .
generalizations of the boas - bellman , bombieri , selberg , heilbronn and peari inequalities for finite sequences of vectors that are not necessarily orthogonal are also provided .
two extensions of the celebrated ostrowski inequalities for sequences of real numbers and the generalization of wagner s inequality in inner product spaces are pointed out .
finally , some grss type inequalities for @xmath0-tuples of vectors in inner product spaces and their natural applications for the approximation of the discrete fourier and mellin transforms are given as well .
the monograph may be used by researchers in different branches of mathematical and functional analysis where the theory of hilbert spaces is of relevance .
since it is self - contained and all the results are completely proved , the work may be also used by graduate students interested in theory of inequalities and its applications .
every section is followed by the necessary references and in this way each section may be read independently of the others . _
the author _ , december , 2003 .
let @xmath1 be a linear space over the real or complex number field @xmath2 .
the functional @xmath3 is called an _ inner product _ on @xmath1 if it satisfies the conditions 1 .
@xmath4 for any @xmath5 and @xmath6 iff @xmath7 2 .
@xmath8 for any @xmath9 and @xmath10 3 .
@xmath11 for any @xmath12 a first fundamental consequence of the properties ( i)-(iii ) above , is the _ _
schwarz inequality:__@xmath13for any @xmath12 the equality holds in ( [ cbs ] ) if and only if the vectors @xmath14 and @xmath15 are _ linearly dependent , _ i.e. , there exists a nonzero constant @xmath16 so that @xmath17 if we denote @xmath18 then one may state the following properties 1 .
@xmath19 for any @xmath5 and @xmath20 iff @xmath7 2 .
@xmath21 for any @xmath16 and @xmath22 3 .
@xmath23 for any @xmath24 ( the triangle inequality ) ;
i.e. , @xmath25 is a _ norm _ on @xmath26 in this chapter we present some recent reverse inequalities for the schwarz and the triangle inequalities .
more precisely , we point out upper bounds for the nonnegative quantities@xmath27and @xmath28under various assumptions for the vectors @xmath12 if the vectors @xmath24 are not _ orthogonal _ ,
i.e. , @xmath29 then some upper bounds for the supra - unitary quantities@xmath30are provided as well .
let @xmath31 and @xmath32 be two positive @xmath33__tuples
_ _ with@xmath34for each @xmath35 and some constants @xmath36 the following reverses of the cauchy - bunyakovsky - schwarz inequality for positive sequences of real numbers are well known : 1 . _ plya - szeg s inequality _
@xcite@xmath37 2 . _ shisha - mond s inequality _ @xcite@xmath38 ^{2}.\ ] ] 3 . _ ozeki s inequality _
@xcite@xmath39 4 .
_ diaz - metcalf s inequality _
if @xmath41 is a positive sequence , then the following weighted inequalities also hold : 1 . _
cassel s inequality _
@xcite . if the positive real sequences @xmath31 and @xmath42 satisfy the condition @xmath43then@xmath44 2 .
_ greub - reinboldt s inequality _ @xcite .
we have@xmath45provided @xmath31 and @xmath42 satisfy the condition @xmath46 3 .
_ generalised diaz - metcalf s inequality _
@xcite , see also @xcite . if @xmath47 $ ] and @xmath48 @xmath49 and @xmath50 holds , then one has the inequality@xmath51 4 . _ klamkin - mclenaghan s inequality _
if @xmath52 satisfy ( [ 1.5a.1 ] ) , then@xmath53 for other recent results providing discrete reverse inequalities , see the recent monograph online @xcite . in this section , by following @xcite , we point out a new reverse of schwarz s inequality in real or complex inner product spaces .
particular cases for isotonic linear functionals , integrals and sequences are also given .
the following reverse of schwarz s inequality in inner product spaces holds @xcite .
[ t2.1.1]let @xmath54 @xmath55 and @xmath12 if@xmath56or equivalently,@xmath57holds , then one has the inequality@xmath58the constant @xmath59 is sharp in ( [ 2.2.1 ] ) .
the equivalence between @xmath60 and @xmath61 can be easily proved , see for example @xcite .
let us define@xmath62\]]and @xmath63then@xmath64 -\left\vert \left\langle x , y\right\rangle \right\vert ^{2}-\left\vert y\right\vert ^{4}\func{re}\left ( a\overline{a}\right)\]]and@xmath65 -\left\vert x\right\vert ^{2}\left\vert y\right\vert ^{2}-\left\vert y\right\vert ^{4}\func{re}\left ( a\overline{a}\right ) , \]]which gives@xmath66for any @xmath24 and @xmath67 .
this is an interesting identity in itself as well .
if ( [ 2.1.1 ] ) holds , then @xmath68 and thus@xmath69 .
\label{2.4.1}\]]further , if we use the elementary inequality for @xmath70 @xmath71@xmath72 \leq \frac{1}{4}\left\vert u+v\right\vert ^{2},\]]then we have , for @xmath73that@xmath74 \leq \frac{1}{4}\left\vert a - a\right\vert ^{2}\left\vert y\right\vert ^{4}. \label{2.6.1}\]]making use of the inequalities @xmath75 and @xmath76 we deduce @xmath77 .
now , assume that ( [ 2.2.1 ] ) holds with a constant @xmath78 i.e.,@xmath79where @xmath80 satisfy ( [ 2.1.1 ] ) .
consider @xmath81 @xmath82 @xmath83 and @xmath84 @xmath85 with @xmath86 define@xmath87then@xmath88and thus the condition ( [ 2.1.1 ] ) is fulfilled . from ( [ 2.61.1 ] ) we deduce@xmath89and since@xmath90and@xmath91then , by ( [ 2.7.1 ] ) , we obtain@xmath92which gives @xmath93 and the theorem is completely proved .
let @xmath94 be an algebra of real functions defined on @xmath95 and @xmath96 a subclass of @xmath94 satisfying the conditions : 1 .
@xmath97 implies @xmath98 2 .
@xmath99 @xmath100 implies @xmath101 a functional @xmath102 defined on @xmath103 is an _ isotonic linear functional _ on @xmath103 provided that 1 .
@xmath104 for all @xmath105 and @xmath106 2 .
@xmath107 that is , @xmath108 for all @xmath109 implies @xmath110 the functional @xmath102 is _ normalised _ on @xmath111 provided that @xmath112 i.e. , @xmath113 for all @xmath109 implies @xmath114 usual examples of isotonic linear functionals are integrals , sums , etc . now , suppose that @xmath115 @xmath116 is given and satisfies the properties that @xmath117 @xmath118 @xmath119 for all @xmath120 for a given isotonic linear functional @xmath121 with @xmath122 define the mapping @xmath123
by@xmath124 this functional satisfies the following properties : 1 .
@xmath125 for all @xmath126 2 .
@xmath127 for all @xmath128 and @xmath129 ; 3 .
@xmath130 for all @xmath131 the following reverse of schwarz s inequality for positive linear functionals holds @xcite .
[ p3.1.1]let @xmath132 be such that @xmath117 @xmath133 @xmath134 if @xmath135 are real numbers such that@xmath136then for any isotonic linear functional @xmath121 with @xmath137 we have the inequality@xmath138 ^{2}\leq \frac{1}{4}\left ( m - m\right ) ^{2}a^{2}\left ( hg^{2}\right ) .
\label{3.3.1}\]]the constant @xmath59 in ( [ 3.3.1 ] ) is sharp .
we observe that @xmath139 \geq 0.\]]applying theorem [ t2.1.1 ] for @xmath140 we get@xmath141which is clearly equivalent to ( [ 3.3.1 ] ) .
the following corollary holds .
[ c3.2.1]let @xmath142 such that @xmath143 @xmath144 if @xmath135 are real numbers such that ( [ 3.21.1 ] ) holds , then@xmath145the constant @xmath59 is sharp in ( [ 3.4.1 ] ) .
[ r3.3.1]the condition ( [ 3.21.1 ] ) may be replaced with the weaker assumption@xmath146 let @xmath147 be a measure space consisting of a set @xmath148 @xmath149 a @xmath150algebra of subsets of @xmath151 and @xmath152 a countably additive and positive measure on @xmath149 with values in @xmath153 denote by @xmath154 the hilbert space of all @xmath2-valued functions @xmath155 defined on @xmath151 that are @xmath156integrable on @xmath148 i.e. , @xmath157 where @xmath158 is a measurable function on @xmath159 the following proposition contains a reverse of the weighted cauchy - bunyakovsky - schwarz s integral inequality @xcite .
[ p4.1.1]let @xmath54 @xmath55 and @xmath160 if @xmath161 \rho \left ( s\right ) d\mu \left ( s\right ) \geq 0 \label{4.1.1}\]]or equivalently,@xmath162holds , then one has the inequality@xmath163the constant @xmath59 is best possible
. follows by theorem [ t2.1.1 ] applied for the inner product @xmath164,@xmath165 [ r4.2.1]a sufficient condition for ( [ 4.1.1 ] ) to hold is @xmath166 \geq 0,\text { \ for \ } \mu -\text{a.e . }
s\in \omega .\ ] ] in the particular case @xmath167 we have the following reverse of the cauchy - bunyakovsky - schwarz inequality .
[ c4.3.1]let @xmath67 @xmath55 and @xmath168 if@xmath161 d\mu \left ( s\right ) \geq 0 , \label{4.4.1}\]]or equivalently,@xmath169holds , then one has the inequality@xmath170the constant @xmath59 is best possible [ r4.4.1]if @xmath171 then a sufficient condition for either ( [ 4.1.1 ] ) or ( [ 4.4.1 ] ) to hold is @xmath172where , in this case , @xmath173 with @xmath174 for a given sequence @xmath175 of nonnegative real numbers , consider the hilbert space @xmath176 @xmath177 where @xmath178 the following proposition that provides a reverse of the weighted cauchy - bunyakovsky - schwarz inequality for complex numbers holds .
[ p5.1.1]let @xmath67 and @xmath179 if@xmath180 \geq 0 , \label{5.2.1}\]]then one has the inequality@xmath181the constant @xmath59 is sharp
. follows by theorem [ t2.1.1 ] applied for the inner product @xmath182,@xmath183 [ r5.2.1]a sufficient condition for ( [ 5.2.1 ] ) to hold is@xmath184 \geq 0,\text { \ for all } i\in \mathbb{n}.\ ] ] in the particular case @xmath185 @xmath186 , we have the following reverse of the cauchy - bunyakovsky - schwarz inequality . [ c5.3.1]let @xmath67 @xmath55 and @xmath187 if@xmath188 \geq 0 ,
\label{5.5.1}\]]then one has the inequality@xmath189 [ r5.4.1]if @xmath190 , then a sufficient condition for either ( [ 5.2.1 ] ) or ( [ 5.5.1 ] ) to hold is@xmath191with @xmath174 99 j.b .
diaz and f.t .
metcalf , stronger forms of a class of inequalities of g. plya - g .
szeg and l.v .
kantorovich , _ bull .
, _ * 69 * ( 1963 ) , 415 - 418 .
dragomir , some grss type inequalities in inner product spaces , _ j. inequal .
pure & appl .
_ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] . s.s .
dragomir , _ a survey on cauchy - bunyakovsky - schwarz type discrete inequalities _ , rgmia monographs , victoria university , 2002 . `
( on line : http://rgmia.vu.edu.au/monographs/ )
. ` s.s .
dragomir , a counterpart of schwarz s inequality in inner product spaces , _ rgmia res .
_ , * 6*(2003 ) , supplement , article 18 . `
[ on line : http://rgmia.vu.edu.au/v6(e).html ]
. ` w. greub and w. rheinboldt , on a generalisation of an inequality of l.v .
kantorovich , _ proc .
soc . , _ * 10 * ( 1959 ) , 407 - 415 . m.s .
klamkin and r.g .
mclenaghan , an ellipse inequality , _ math . mag .
, _ * 50 * ( 1977 ) , 261 - 263 . d.s .
mitrinovi , j.e .
peari and a.m. fink , _ classical and new inequalities in analysis _
, kluwer academic publishers , dordrecht / boston / london , 1993 .
n. ozeki , on the estimation of the inequality by the maximum , _ j. college arts , chiba univ . , _ * 5*(2 ) ( 1968 ) , 199 - 203 .
g. plya and g. szeg , _ aufgaben und lehrstze aus der analysis , _ vol .
1 , berlin 1925 , pp .
57 and 213 - 214 .
o. shisha and b. mond , bounds on differences of means , _ inequalities i , _ new york - london , 1967 , 293 - 308 .
watson , serial correlation in regression analysis i , _ biometrika _ , * 42 * ( 1955 ) , 327 - 342 .
the following result was proved by j.w.s .
cassels in 1951 ( see appendix 1 of @xcite ) . [ t1.1.2]let @xmath192 @xmath193 be sequences of positive real numbers and @xmath194 a sequence of nonnegative real numbers .
suppose that@xmath195then one has the inequality@xmath196the equality holds in ( [ 1.2.2 ] ) when @xmath197 @xmath198 @xmath199 @xmath200 and @xmath201 if one assumes that @xmath202 and @xmath203 for each @xmath35 then by ( [ 1.2.2 ] ) we may obtain _ greub - reinboldt s inequality _
@xcite@xmath204the following unweighted cassels inequality also holds@xmath205provided @xmath206 and @xmath207 satisfy ( 1.1.2 ) .
this inequality will produce the well known _
plya - szeg inequality _
@xcite , 3b.2:@xmath208provided @xmath202 and @xmath209 for each @xmath210 in @xcite , c.p .
niculescu proved , amongst others , the following generalisation of cassels inequality : [ t1.2.2]let @xmath211 be a vector space endowed with a hermitian product @xmath212 then@xmath213for every @xmath214 and every @xmath215 for which @xmath216 for other reverses of the cauchy - bunyakovsky - schwarz inequality , see the references @xcite @xcite . in this section , by following @xcite , we establish a generalisation of ( [ 1.6.2 ] ) for complex numbers @xmath217 and @xmath151 for which @xmath218 applications for isotonic linear functionals , integrals and sequences are also given .
the following reverse of schwarz s inequality in inner product spaces holds @xcite .
[ t2.1.2]let @xmath67 @xmath55 so that @xmath219 if @xmath24 are such that@xmath220then one has the inequality@xmath221 } { \left [ \func{re}\left ( \overline{a}% a\right ) \right ] ^{\frac{1}{2}}}\leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}% a\right ) \right ] ^{\frac{1}{2}}}\left\vert \left\langle x , y\right\rangle \right\vert .
\label{2.2.2}\]]the constant @xmath222 is sharp in both inequalities .
we have , obviously , that@xmath223 -\left\vert x\right\vert ^{2}-\left [ \func{re}\left ( \overline{a}a\right ) \right ] \left\vert y\right\vert ^{2}\end{aligned}\]]and , thus , by ( [ 2.1.2 ] ) , one has@xmath224 \cdot \left\vert y\right\vert ^{2}\leq \func{re}\left [ a\overline{% \left\langle x , y\right\rangle } + \overline{a}\left\langle x , y\right\rangle % \right ] , \]]which gives@xmath225 ^{\frac{1}{2}}}% \left\vert x\right\vert ^{2}+\left [ \func{re}\left ( \overline{a}a\right ) % \right ] ^{\frac{1}{2}}\left\vert y\right\vert ^{2}\leq \frac{\func{re}\left [ a\overline{\left\langle x , y\right\rangle } + \overline{a}\left\langle x , y\right\rangle \right ] } { \left [ \func{re}\left ( \overline{a}a\right ) % \right ] ^{\frac{1}{2}}}. \label{2.3.2}\]]on the other hand , by the elementary inequality@xmath226valid for @xmath227 and @xmath228 we deduce@xmath229 ^{\frac{1}{2}}}\left\vert x\right\vert ^{2}+\left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{% \frac{1}{2}}\left\vert y\right\vert ^{2}. \label{2.4.2}\]]utilizing ( [ 2.3.2 ] ) and ( [ 2.4.2 ] ) we deduce the first part of ( 2.2.2 ) .
the second part is obvious by the fact that for @xmath230 @xmath231 now , assume that the first inequality in ( [ 2.2.2 ] ) holds with a constant @xmath232 i.e. , @xmath233 } { \left [ \func{re}\left ( \overline{a}a\right ) % \right ] ^{\frac{1}{2 } } } , \label{2.5.2}\]]where @xmath234 and @xmath15 satisfy ( [ 2.2.2 ] ) .
if we choose @xmath235 @xmath236 then obviously ( [ 2.1.2 ] ) holds and from ( [ 2.5.2 ] ) we obtain@xmath237giving @xmath238 the theorem is completely proved .
the following corollary is a natural consequence of the above theorem 1nssd .
[ c2.2.2]let @xmath239 if @xmath24 are such that@xmath240then one has the inequality@xmath241the constant @xmath222 is sharp in ( [ 2.7.2 ] ) .
[ r2.2.1.2]the inequality ( [ 2.7.2 ] ) is equivalent to niculescu s inequality ( [ 1.6.2 ] ) .
the following corollary is also obvious @xcite .
[ c2.3.2]with the assumptions of corollary [ c2.2.2 ] , we have@xmath242and@xmath243 ^{2 } \label{2.9.2 } \\ & \leq \frac{\left ( m - m\right ) ^{2}}{4mm}\left [ \func{re}\left\langle x , y\right\rangle \right ] ^{2}\leq \frac{\left ( m - m\right ) ^{2}}{4mm}% \left\vert \left\langle x , y\right\rangle \right\vert ^{2}. \notag\end{aligned}\]]the constants @xmath222 and @xmath59 are sharp . if we subtract @xmath244 from the first inequality in ( [ 2.7.2 ] ) , we get@xmath245which proves the third inequality in ( [ 2.8.2 ] ) .
the other ones are obvious .
now , if we square the first inequality in ( [ 2.7.2 ] ) and then subtract @xmath246 ^{2},$ ] we get@xmath247 ^{2 } & \leq \left [ \frac{\left ( m+m\right ) ^{2}}{4mm}-1\right ] \left [ \func{re}\left\langle x , y\right\rangle % \right ] ^{2 } \\ & = \frac{\left ( m - m\right ) ^{2}}{4mm}\left [ \func{re}\left\langle x , y\right\rangle \right ] ^{2}\end{aligned}\]]which proves the third inequality in ( [ 2.9.2 ] ) .
the other ones are obvious .
the following proposition holds @xcite .
[ p3.1.2]let @xmath132 be such that @xmath117 @xmath133 @xmath134 if @xmath248 are such that@xmath249then for any isotonic linear functional @xmath121 with @xmath250 we have the inequality@xmath251the constant @xmath59 in ( [ 3.3.2 ] ) is sharp .
we observe that@xmath139 \geq 0.\]]applying corollary [ c2.2.2 ] for @xmath252 we get@xmath253which is clearly equivalent to ( [ 3.3.2 ] ) .
the following additive versions of ( [ 3.3.2 ] ) also hold @xcite .
[ c3.2.2]with the assumption in proposition [ p3.1.2 ] , one has@xmath254 ^{\frac{1% } { 2}}-a\left ( hfg\right ) \\ & \leq \frac{\left ( \sqrt{m}-\sqrt{m}\right ) ^{2}}{2\sqrt{mm}}a\left ( hfg\right)\end{aligned}\]]and @xmath255the constants @xmath222 and @xmath59 are sharp
. [ r3.3.2]the condition ( [ 3.2.2 ] ) may be replaced with the weaker assumption@xmath256 [ r3.4.2]with the assumption ( [ 3.2.2 ] ) or ( [ 3.6.2 ] ) and if @xmath257 with @xmath258 then one has the inequalities@xmath259@xmath260 ^{\frac{1}{2% } } -a\left ( fg\right ) \\ & \leq \frac{\left ( \sqrt{m}-\sqrt{m}\right ) ^{2}}{2\sqrt{mm}}a\left ( fg\right)\end{aligned}\]]and@xmath261 the following proposition contains a reverse of the weighted cauchy - bunyakovsky - schwarz integral inequality . [ p4.1.2]let @xmath54 @xmath55 with @xmath262 and @xmath263 if @xmath161 \rho \left ( s\right ) d\mu \left ( s\right ) \geq 0 , \label{4.1.2}\]]then one has the inequality@xmath264 ^{% \frac{1}{2 } } \label{4.2.2 } \\ & \leq \frac{1}{2}\cdot \frac{\int_{\omega } \rho \left ( s\right ) \func{re}% \left [ a\overline{f\left ( s\right ) } g\left ( s\right ) + \overline{a}f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1}{2 } } } \notag \\ & \leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1% } { 2}}}\left\vert \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right\vert .
\notag\end{aligned}\]]the constant @xmath222 is sharp in ( [ 4.2.2 ] )
. follows by theorem [ t2.1.2 ] applied for the inner product @xmath164,@xmath265 [ r4.2.2]a sufficient condition for ( [ 4.1.2 ] ) to hold is @xmath266 \geq 0,\text { \ for } \mu \text{-a.e . \ } s\in \omega .\ ] ] in the particular case @xmath167 we have the following result .
[ c4.3.2]let @xmath67 @xmath55 with @xmath262 and @xmath267 if@xmath268 d\mu \left ( s\right ) \geq 0 , \label{4.4.2}\]]then one has the inequality@xmath269 ^{\frac{1}{2 } } \\ & \leq \frac{1}{2}\cdot \frac{\int_{\omega } \func{re}\left [ a\overline{% f\left ( s\right ) } g\left ( s\right ) + \overline{a}f\left ( s\right ) \overline{% g\left ( s\right ) } \right ] d\mu \left ( s\right ) } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1}{2 } } } \\ & \leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1% } { 2}}}\left\vert \int_{\omega } f\left ( s\right ) \overline{g\left ( s\right ) } % d\mu \left ( s\right ) \right\vert .\end{aligned}\ ] ] [ r4.4.2]if @xmath171 then a sufficient condition for either ( [ 4.1.2 ] ) or ( [ 4.4.2 ] ) to hold is @xmath270where , in this case , @xmath173 with @xmath174 if @xmath271
are real positive constants , then the following proposition holds .
[ p4.5.2]let @xmath239 if @xmath272 such that@xmath273 d\mu \left ( s\right ) \geq 0,\]]then one has the inequality@xmath274 ^{% \frac{1}{2 } } \\ \leq \frac{1}{2}\cdot \frac{m+m}{\sqrt{mm}}\int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) .\end{gathered}\ ] ] the proof follows by corollary [ c2.2.2 ] applied for the inner product @xmath165 the following additive versions also hold @xcite .
[ c4.6.2]with the assumptions in proposition [ p4.5.2 ] , one has the inequalities@xmath275 ^{\frac{1}{2 } } \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ -\int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } % \right ] d\mu \left ( s\right ) \\ & \leq \frac{\left ( \sqrt{m}-\sqrt{m}\right ) ^{2}}{2\sqrt{mm}}\int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right)\end{aligned}\]]and @xmath276 d\mu \left ( s\right ) \right ) ^{2 } \\ & \leq \frac{\left ( m - m\right ) ^{2}}{4mm}\left ( \int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) \right ) ^{2}.\end{aligned}\ ] ] [ r4.7.2]if @xmath190 , a sufficient condition for ( 4.1.2 ) to hold is @xmath277where @xmath278 for a given sequence @xmath175 of nonnegative real numbers , consider the hilbert space @xmath176 @xmath177 where @xmath178 the following proposition that provides a reverse of the weighted cauchy - bunyakovsky - schwarz inequality for complex numbers holds @xcite .
[ p5.1.2]let @xmath67 with @xmath279 and @xmath280 if@xmath281 \geq 0 , \label{5.2.2}\]]then one has the inequality@xmath282 ^{\frac{% 1}{2 } } & \leq \frac{1}{2}\cdot \frac{\sum_{i=0}^{\infty } w_{i}\func{re}\left [ a\overline{x_{i}}y_{i}+\overline{a}x_{i}\overline{y_{i}}\right ] } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1}{2 } } } \label{5.3.2 } \\
& \leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1% } { 2}}}\left\vert \sum_{i=0}^{\infty } w_{i}x_{i}\overline{y_{i}}\right\vert . \notag\end{aligned}\]]the
constant @xmath222 is sharp in ( [ 5.3.2 ] )
. follows by theorem [ t2.1.2 ] applied for the inner product @xmath283,@xmath183 [ r5.2.2]a sufficient condition for ( [ 5.2.2 ] ) to hold is@xmath184 \geq 0\text { \ for all } i\in \mathbb{n}. \label{5.4.2}\ ] ] in the particular case @xmath167 we have the following result .
[ c5.3.2]let @xmath67 with @xmath279 and @xmath187 if@xmath188 \geq 0,\]]then one has the inequality@xmath284 ^{\frac{1}{2}% } & \leq \frac{1}{2}\cdot \frac{\sum_{i=0}^{\infty } \func{re}\left [ a% \overline{x_{i}}y_{i}+\overline{a}x_{i}\overline{y_{i}}\right ] } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1}{2 } } } \\ & \leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1% } { 2}}}\left\vert \sum_{i=0}^{\infty } x_{i}\overline{y_{i}}\right\vert .\end{aligned}\ ] ] [ r5.4.2]if @xmath190 , then a sufficient condition for either ( [ 5.2.2 ] ) or ( [ 5.4.2 ] ) to hold is@xmath285where , in this case , @xmath173 with @xmath174 for @xmath286 @xmath287 , then the following proposition also holds .
[ p5.5.2]let @xmath239 if @xmath288 such that @xmath289 \geq 0 , \label{5.7.2}\]]then one has the inequality@xmath290 ^{\frac{% 1}{2}}\leq \frac{1}{2}\cdot \frac{m+m}{\sqrt{mm}}\sum_{i=0}^{\infty } w_{i}% \func{re}\left ( x_{i}\overline{y_{i}}\right ) .\ ] ] the proof follows by corollary [ c2.2.2 ] applied for the inner product @xmath291 the following additive version also holds @xcite .
[ c5.6.2]with the assumptions in proposition [ p5.5.2 ] , one has the inequalities@xmath292 ^{\frac{% 1}{2}}-\sum_{i=0}^{\infty } w_{i}\func{re}\left ( x_{i}\overline{y_{i}}\right ) \\ & \leq \frac{\left ( \sqrt{m}-\sqrt{m}\right ) ^{2}}{2\sqrt{mm}}% \sum_{i=0}^{\infty } w_{i}\func{re}\left ( x_{i}\overline{y_{i}}\right)\end{aligned}\]]and@xmath293 ^{2 } \\ & \leq \frac{\left ( m - m\right ) ^{2}}{4mm}\left [ \sum_{i=0}^{\infty } w_{i}% \func{re}\left ( x_{i}\overline{y_{i}}\right ) \right ] ^{2}.\end{aligned}\ ] ] [ r5.7.2]if @xmath190 , a sufficient condition for ( 5.7.2 ) to hold is@xmath294where @xmath278 j.b .
diaz and f.t .
metcalf , stronger forms of a class of inequalities of g. plya - g .
szeg and l.v .
kantorovich , _ bull .
_ * 69 * ( 1963 ) , 415 - 418 .
dragomir , some grss type inequalities in inner product spaces , _ j. inequal .
pure & appl .
_ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] . s.s .
dragomir , a generalisation of the cassels and greub - reinboldt inequalities in inner product spaces , preprint , _ mathematics arxiv _ , math.ca/0307130 , [ ` on line : http://front.math.ucdavis.edu/math.ca/0306352 ` ] .
w. greub and w. rheinboldt , on a generalisation of an inequality of l.v .
kantorovich , _ proc .
soc . , _ * 10 * ( 1959 ) , 407 - 415 . m.s .
klamkin and r.g .
mclenaghan , an ellipse inequality , _ math . mag .
, _ * 50 * ( 1977 ) , 261 - 263 . d.s .
mitrinovi , j.e .
peari and a.m. fink , _ classical and new inequalities in analysis _
, kluwer academic publishers , dordrecht / boston / london , 1993 . c.p .
niculescu , converses of the cauchy - schwartz inequality in the c*-framework , _ rgmia res .
_ , * 4*(2001 ) , article 3 . `
[ on line : http://rgmia.vu.edu.au/v4n1.html ] .
` n. ozeki , on the estimation of the inequality by the maximum , _
j. college arts , chiba univ .
, _ * 5*(2 ) ( 1968 ) , 199 - 203 .
g. plya and g. szeg , _ aufgaben und lehrstze aus der analysis , _ vol .
1 , berlin 1925 , pp .
57 and 213 - 214 .
o. shisha and b. mond , bounds on differences of means , _ inequalities i , _ new york - london , 1967 , 293 - 308 .
watson , serial correlation in regression analysis i , _ biometrika _ , * 42 * ( 1955 ) , 327 - 342 .
it has been proven in @xcite , that@xmath295provided , either @xmath296or equivalently,@xmath297holds , where @xmath298 @xmath299 the constant @xmath59 in ( [ 1.7.3 ] ) is best possible .
if we choose @xmath300 @xmath301 @xmath302 @xmath303 @xmath304 , then by ( 1.8.3 ) and ( [ 1.9.3 ] ) we have,@xmath305or equivalently,@xmath306implying the following reverse of schwarz s inequality:@xmath307the constant @xmath59 in ( [ 1.12.3 ] ) is sharp . note
that , this inequality is an improvement of ( [ 2.2.1 ] ) , but it may not be very convenient for applications . in @xcite , it has also been proven that @xmath308provided either ( [ 1.8.3 ] ) or ( [ 1.9.3 ] ) holds true .
if we make the same choice for @xmath309 and @xmath310 as above , then we deduce the inequality@xmath311provided either ( [ 1.10.3 ] ) or ( [ 1.11.3 ] ) holds true .
the constant @xmath59 is best possible in ( [ 1.12.3b ] ) .
the following result holds @xcite .
[ t2.1.3]let @xmath312 be an inner product space over the real or complex number field @xmath2 @xmath313 and @xmath314 @xmath315 are such that@xmath316 1 .
if @xmath317 then we have the inequalities@xmath318 ^{2}\leq r^{2}\left\vert x\right\vert ^{2}. \label{2.2.3}\]]the constant @xmath319 in front of @xmath320 is best possible in the sense that it can not be replaced by a smaller one .
if @xmath321 then@xmath322the constant @xmath323 is best possible in both inequalities .
3 . if @xmath324 then@xmath325here the constant @xmath323 is also best possible . since @xmath326 then obviously @xmath327 which is equivalent to @xmath328 1 . if @xmath317 then we may divide ( [ 2.5.3 ] ) by @xmath329 getting @xmath330using the elementary inequality@xmath331we may state that@xmath332making use of ( [ 2.6.3 ] ) and ( [ 2.7.3 ] ) , we deduce@xmath333which is an interesting inequality in itself as well .
+ taking the square in ( [ 2.8.3 ] ) and re - arranging the terms , we deduce the third inequality in ( [ 2.2.3 ] ) .
the others are obvious . + to prove the sharpness of the constant , assume , under the hypothesis of the theorem , that , there exists a constant @xmath334 such that@xmath335 ^{2}\leq cr^{2}\left\vert x\right\vert ^{2 } , \label{2.9.3}\]]provided @xmath336 and @xmath337 + let @xmath338 @xmath339 @xmath340 with @xmath341 and @xmath342 put @xmath343 then obviously @xmath344 @xmath345 and @xmath346 , @xmath347 and thus @xmath348 ^{2}=\varepsilon .$ ] using ( [ 2.9.3 ] ) , we may write that@xmath349giving @xmath350letting @xmath351 we get from ( [ 2.10.3 ] ) that @xmath352 and the sharpness of the constant is proved .
the inequality ( [ 2.3.3 ] ) is obvious by ( [ 2.5.3 ] ) since @xmath353 the best constant follows in a similar way to the above .
the inequality ( [ 2.4.3 ] ) is obvious .
the best constant may be proved in a similar way to the above .
we omit the details .
the following reverse of schwarz s inequality holds @xcite .
[ t2.2.3]let @xmath312 be an inner product space over @xmath2 and @xmath354 @xmath355 such that either@xmath356or equivalently,@xmath357holds . 1 .
if @xmath358 then we have the inequalities@xmath359 \right\ } ^{2}}{\func{re% } \left ( \gamma \overline{\gamma } \right ) } \label{2.13.3 } \\ & \leq \frac{1}{4}\cdot \frac{\left\vert \gamma + \gamma \right\vert ^{2}}{% \func{re}\left ( \gamma \overline{\gamma } \right ) } \left\vert \left\langle x , y\right\rangle \right\vert ^{2}. \notag\end{aligned}\]]the constant @xmath59 is best possible in both inequalities .
if @xmath360 then @xmath361 \leq \left\vert \gamma + \gamma \right\vert \left\vert \left\langle x , y\right\rangle \right\vert .\ ] ] 3 .
if @xmath362 then @xmath363 \\ & \leq -\func{re}\left ( \gamma \overline{\gamma } \right ) \left\vert y\right\vert ^{2}+\left\vert \gamma + \gamma \right\vert \left\vert \left\langle x , y\right\rangle \right\vert .\end{aligned}\ ] ] the proof of the equivalence between the inequalities ( [ 2.11.3 ] ) and ( [ 2.12.3 ] ) follows by the fact that in an inner product space @xmath364 for @xmath365 is equivalent with @xmath366 ( see for example @xcite ) .
consider , for @xmath367 @xmath368 and @xmath369 then@xmath370 1 .
if @xmath358 then the hypothesis of ( i ) in theorem [ t2.1.3 ] is satisfied , and by the second inequality in ( [ 2.2.3 ] ) we have@xmath371 \right\ } ^{2}\leq \frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}\left\vert x\right\vert ^{2}\left\vert y\right\vert ^{2}\]]from where we derive@xmath372 \right\ } ^{2},\]]giving the first inequality in ( [ 2.13.3 ] ) .
+ the second inequality is obvious . + to prove the sharpness of the constant @xmath373 assume that the first inequality in ( [ 2.13.3 ] ) holds with a constant @xmath232
i.e. , @xmath374 \right\ } ^{2}}{\func{re}\left ( \gamma \overline{\gamma } \right ) } , \label{2.16.3}\]]provided @xmath375 and either ( [ 2.11.3 ] ) or ( [ 2.12.3 ] ) holds .
+ assume that @xmath376 and let @xmath377 then ( [ 2.11.3 ] ) holds and by ( [ 2.16.3 ] ) we deduce@xmath378giving@xmath379let @xmath380 and choose in ( [ 2.17.3 ] ) , @xmath381 @xmath382 to get @xmath383 for any @xmath384 letting @xmath385 we deduce @xmath386 and the sharpness of the constant is proved .
+ \(ii ) and ( iii ) are obvious and we omit the details . we observe that the second bound in ( [ 2.13.3 ] ) for @xmath387 is better than the second bound provided by ( [ 2.2.2 ] ) . the following corollary provides a reverse inequality for the additive version of schwarz s inequality @xcite .
[ c2.3.3]with the assumptions of theorem [ t2.2.3 ] and if @xmath388 then we have the inequality:@xmath389the constant @xmath59 is best possible in ( [ 2.18.3 ] ) .
the proof is obvious from ( [ 2.13.3 ] ) on subtracting in both sides the same quantity @xmath390 the sharpness of the constant may be proven in a similar manner to the one incorporated in the proof of ( i ) , theorem [ t2.2.3 ] .
we omit the details . for other recent results in connection to schwarz s inequality , see adr.3 , @xcite and @xcite .
the following reverse of the triangle inequality holds @xcite .
[ p2.4.3]let @xmath312 be an inner product space over the real or complex number field @xmath2 @xmath391 and @xmath392 @xmath315 are such that @xmath393then we have the inequality@xmath394 using the inequality ( [ 2.8.3 ] ) , we may write that@xmath395which gives@xmath396since@xmath397then by ( [ 2.21.3 ] ) , we have@xmath398giving the desired inequality ( [ 2.20.3 ] ) .
the following proposition providing a simpler reverse for the triangle inequality also holds @xcite .
[ p2.5.3]let @xmath312 be an inner product space over @xmath2 and @xmath354 @xmath399 such that either@xmath240or equivalently,@xmath400holds .
then we have the inequality@xmath401{mm}}\sqrt{\func{re}% \left\langle x , y\right\rangle } .
\label{2.24.3}\ ] ] choosing in ( [ 2.8.3 ] ) , @xmath402 @xmath403 we get@xmath404giving @xmath405following the same argument as in the proof of proposition [ p2.4.3 ] , we deduce the desired inequality ( [ 2.24.3 ] ) . for some results related to triangle inequality in inner product spaces ,
see @xcite , @xcite , @xcite and @xcite .
let @xmath147 be a measurable space consisting of a set @xmath148 @xmath149 a @xmath150algebra of parts and @xmath406 a countably additive and positive measure on @xmath149 with values in @xmath407 let @xmath408 be a @xmath409measurable function on @xmath151 with @xmath410 denote by @xmath154 the hilbert space of all real or complex valued functions defined on @xmath151 and @xmath411integrable on @xmath148 i.e.,@xmath412it is obvious that the following inner product@xmath413generates the norm @xmath414 of @xmath415 and all the above results may be stated for integrals .
it is important to observe that , if @xmath416 \geq 0,% \text { \ for } \mu -\text{a.e . }
s\in \omega , \]]then , obviously,@xmath417 \label{7.4.3 } \\ & = \int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) \geq 0 .
\notag\end{aligned}\]]the reverse is evidently not true in general .
moreover , if the space is real , i.e. , @xmath418 , then a sufficient condition for ( [ 7.4.3 ] ) to hold is:@xmath419 we now provide , by the use of certain results obtained above , some integral inequalities that may be used in practical applications .
[ p7.1.3]let @xmath420 and @xmath315 with the properties that@xmath421then we have the inequalities@xmath422 ^{2 } \notag \\ & \leq r^{2}\int_{\omega } \rho \left ( s\right ) \left\vert g\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) .
\notag\end{aligned}\]]the constant @xmath423 in front of @xmath320 is best possible .
the proof follows by theorem [ t2.1.3 ] and we omit the details 3nssd .
[ p7.2.3]let @xmath420 and @xmath424 such that @xmath425 and@xmath426 \geq 0,\text { \ for } \mu -\text{a.e . }
s\in \omega .\]]then we have the inequalities@xmath427 \right\ } ^{2}}{\func{re}\left ( \gamma \overline{\gamma } \right ) } \notag \\ &
\leq \frac{1}{4}\cdot \frac{\left\vert \gamma + \gamma \right\vert ^{2}}{% \func{re}\left ( \gamma \overline{\gamma } \right ) } \left\vert \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right\vert ^{2}. \notag\end{aligned}\]]the constant @xmath59 is best possible in both inequalities .
the proof follows by theorem [ t2.2.3 ] and we omit the details .
[ c7.3.3]with the assumptions of proposition [ p7.2.3 ] , we have the inequality@xmath428the constant @xmath59 is best possible .
if the space is real and we assume , for @xmath429 that@xmath430then , by ( [ 7.9.3 ] ) and ( [ 7.10.3 ] ) , we deduce the inequalities@xmath431 ^{2}d\mu \left ( s\right ) \int_{\omega } \rho \left ( s\right ) \left [ g\left ( s\right ) % \right ] ^{2}d\mu \left ( s\right ) \label{7.12.3 } \\ \leq \frac{1}{4}\cdot \frac{\left ( m+m\right ) ^{2}}{mm}\left [ \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) g\left ( s\right ) d\mu \left ( s\right ) % \right ] ^{2}\end{gathered}\]]and @xmath432 ^{2}d\mu \left ( s\right ) \int_{\omega } \rho \left ( s\right ) \left [ g\left ( s\right ) \right ] ^{2}d\mu \left ( s\right ) \label{7.13.3 } \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ -\left [ \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) g\left ( s\right ) d\mu \left ( s\right ) \right ] ^{2 } \notag \\ & \leq \frac{1}{4}\cdot \frac{\left ( m - m\right ) ^{2}}{mm}\left [ \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) g\left ( s\right ) d\mu \left ( s\right ) % \right ] ^{2}. \notag\end{aligned}\]]the inequality ( [ 7.12.3 ] ) is known in the literature as cassel s inequality . a. de rossi , a strengthened cauchy - schwarz inequality for biorthogonal wavelets , _ math .
appl . , _ * 2 * ( 1999 ) , no .
2 , 263282 .
j. b. diaz and f. t. metcalf , a complementary triangle inequality in hilbert and banach spaces , _ proc .
soc_. , * 17 * ( 1966 ) , 8897 . s.s .
dragomir , a counterpart of schwarz s inequality in inner product spaces , _ rgmia res .
, _ * 6*(2003 ) , _ supplement _ , article 18 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , a generalisation of the cassels and grueb - reinboldt inequalities in inner product spaces , preprint , _ _ mathematics ar__@xmath433_iv _ , math.ca/0307130 , ` [ on line http://front.math.ucdavis.edu/ ] . `
dragomir , some companions of the grss inequality in inner product spaces , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 8 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , on bessel and grss inequalities for orthornormal families in inner product spaces , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 12 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , reverses of schwarz , triangle and bessel inequalities in inner product spaces , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 19 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , some grss type inequalities in inner product spaces , _ j. inequal . pure & appl . math . _ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] .
s. s. dragomir and b. mond , on the superadditivity and monotonicity of schwarz s inequality in inner product spaces , _ makedon .
oddel . mat .-
prilozi , _ * 15 * ( 1994 ) , no .
2 , 522 ( 1996 ) .
h. gunawan , on n - inner products , n - norms , and the cauchy - schwarz inequality , _ sci .
jpn . , _ * 55 * ( 2002 ) , no .
1 , 5360 . s.m .
khaleelulla , on diaz - metcalf s complementary triangle inequality , _ kyungpook math .
j. , _ * 15 * ( 1975 ) , 911 .
p.m. milii , on a complementary inequality of the triangle inequality ( french ) , _
vesnik , _ _ * * 41 * * ( 1989 ) , no .
2 , 8388 .
d. k. rao , a triangle inequality for angles in a hilbert space , _ rev .
colombiana mat .
, _ * 10 * ( 1976 ) , no .
3 , 9597 . the following result holds @xcite .
[ t2.1.4]let @xmath312 be an inner product space over the real or complex number field @xmath2 , @xmath434 and @xmath435 if@xmath436then we have the inequalities:@xmath437the constant @xmath222 is best possible in ( [ 2.2.4 ] ) in the sense that it can not be replaced by a smaller constant .
the condition ( [ 2.1.4 ] ) is clearly equivalent to@xmath438using the elementary inequality@xmath439and ( [ 2.3.4 ] ) , we deduce@xmath440giving the last inequality in ( [ 2.2.4 ] ) .
the other inequalities are obvious . to prove the sharpness of the constant @xmath441 assume that@xmath442for any @xmath434 and @xmath315 satisfying ( [ 2.1.4 ] ) .
assume that @xmath443 @xmath341 and @xmath444 if @xmath445 @xmath446 and if we define @xmath447 then @xmath448 showing that the condition ( [ 2.1.4 ] ) is fulfilled . on the other hand,@xmath449utilising ( [ 2.3a.4 ] )
, we conclude that@xmath450multiplying ( [ 2.4b.4 ] ) by @xmath451 and then dividing by @xmath452 we get@xmath453letting @xmath454 in ( [ 2.5.4 ] ) , we deduce @xmath455 and the theorem is proved .
the following result also holds @xcite .
[ t2.2.4]let @xmath312 be an inner product space over @xmath2 and @xmath354 @xmath355 @xmath456 so that either@xmath457or equivalently,@xmath458holds
. then we have the inequalities@xmath459 \right\vert \notag \\ & \leq \left\vert x\right\vert \left\vert y\right\vert -\func{re}\left
[ \frac{\bar{\gamma}+\bar{\gamma}}{\left\vert \gamma + \gamma \right\vert } % \left\langle x , y\right\rangle \right ] \notag \\ & \leq \frac{1}{4}\cdot \frac{\left\vert \gamma -\gamma \right\vert ^{2}}{% \left\vert \gamma + \gamma \right\vert } \left\vert y\right\vert ^{2}. \notag\end{aligned}\]]the constant @xmath59 in the last inequality is best possible .
consider for @xmath460 @xmath461 and @xmath462 thus from ( [ 2.2.4 ] ) , we get@xmath463 \right\vert \\ & \leq \left\vert x\right\vert \left\vert \frac{\gamma + \gamma } { 2}% \right\vert \left\vert y\right\vert -\func{re}\left
[ \frac{\bar{\gamma}+\bar{% \gamma}}{\left\vert
\gamma + \gamma \right\vert } \left\langle x , y\right\rangle \right ] \\ & \leq \frac{1}{8}\cdot \left\vert \gamma -\gamma \right\vert ^{2}\left\vert y\right\vert ^{2}.\end{aligned}\]]dividing by @xmath464 we deduce the desired inequality ( [ 2.8.4 ] ) . to prove the sharpness of the constant @xmath373 assume that there exists a @xmath334 such that:@xmath465 \leq c\cdot \frac{\left\vert \gamma -\gamma \right\vert ^{2}}{\left\vert \gamma + \gamma \right\vert } \left\vert y\right\vert ^{2 } , \label{2.9.4}\]]provided either ( [ 2.6.4 ] ) or ( [ 2.7.4 ] ) holds .
consider the real inner product space @xmath466 with @xmath467 @xmath468 @xmath469 let @xmath470 and @xmath471 with @xmath472 then , by ( [ 2.9.4 ] ) , we deduce@xmath473if @xmath474 @xmath475 then @xmath476showing that the condition ( [ 2.6.4 ] ) is valid . replacing @xmath477 and @xmath478 in ( [ 2.10.4 ] ) , we deduce@xmath479if in ( [ 2.11.4 ] ) we choose @xmath480 @xmath481 with @xmath339 then we have @xmath482giving@xmath483finally , multiplying ( [ 2.12.4 ] ) with @xmath484 and thus dividing by @xmath485 we deduce@xmath486letting @xmath454 in ( [ 2.13.4 ] ) we get @xmath487 and the sharpness of the constant is proved .
the following reverse of the triangle inequality in inner product spaces holds @xcite .
[ p2.3.4]let @xmath312 be an inner product space over the real or complex number field @xmath2 , @xmath434 and @xmath435 if @xmath488 then we have the inequality@xmath489 since@xmath490then by theorem [ t2.1.4 ] we deduce@xmath491from where we obtain@xmath492giving the desired result ( [ 2.14.4 ] ) .
we may state the following result @xcite .
[ p2.4.4]let @xmath312 be an inner product space over @xmath2 and @xmath354 @xmath399 such that either@xmath240or equivalently,@xmath493holds .
then we have the inequality@xmath494 by theorem [ t2.2.4 ] for @xmath495 @xmath496 we have the inequality@xmath497then we may state that@xmath498from where we get@xmath499giving the desired inequality ( [ 2.19.4 ] ) .
we provide now , by the use of certain results obtained above , some integral inequalities that may be used in practical applications .
[ p7.1.4]let @xmath420 and @xmath315 with the property that@xmath500then we have the inequalities@xmath275 ^{\frac{1}{2 } } \label{7.7.4 } \\ & \qquad \qquad -\left\vert \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right\vert \notag \\ & \leq \left [ \int_{\omega } \rho \left ( s\right ) \left\vert f\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) \int_{\omega } \rho \left ( s\right ) \left\vert g\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) \right ] ^{% \frac{1}{2 } } \notag \\ & \qquad \qquad -\left\vert \int_{\omega } \rho \left ( s\right ) \func{re}% \left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) \right\vert \notag \\ & \leq \left [ \int_{\omega } \rho \left ( s\right ) \left\vert f\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) \int_{\omega } \rho \left ( s\right ) \left\vert g\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) \right ] ^{% \frac{1}{2 } } \notag \\ & \qquad \qquad -\int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) \notag \\ & \leq \frac{1}{2}r^{2}. \notag\end{aligned}\]]the constant @xmath222 is best possible in ( [ 7.7.4 ] ) .
the proof follows by theorem [ t2.1.4 ] , and we omit the details .
[ p7.2.4]let @xmath420 and @xmath424 so that @xmath501 and@xmath426 \geq 0,\text { \ \ for \ } \mu -\text{a.e .
\ } s\in \omega .\]]then we have the inequalities@xmath275 ^{\frac{1}{2 } } \label{7.9.4 } \\ & \qquad \qquad -\left\vert \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right\vert \notag \\ & \leq \left [ \int_{\omega } \rho \left ( s\right ) \left\vert f\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) \int_{\omega } \rho \left ( s\right ) \left\vert g\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) \right ] ^{% \frac{1}{2 } } \notag \\ & \qquad \qquad -\left\vert \func{re}\left [ \frac{\bar{\gamma}+\bar{\gamma}}{% \left\vert \gamma + \gamma \right\vert } \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right ] \right\vert \notag\end{aligned}\]]@xmath502 ^{% \frac{1}{2 } } \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ -\func{re}\left [ \frac{\bar{\gamma}+\bar{\gamma}}{\left\vert \gamma + \gamma \right\vert } % \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right ] \\ & \leq \frac{1}{4}\cdot \frac{\left\vert \gamma -\gamma \right\vert ^{2}}{% \left\vert \gamma + \gamma \right\vert } \int_{\omega } \rho \left ( s\right ) \left\vert g\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) .\end{aligned}\]]the constant @xmath59 is best possible .
if the space is real and we assume , for @xmath429 that@xmath503then , by ( [ 7.9.4 ] ) , we deduce the inequality:@xmath504 ^{\frac{1}{2 } } \\ & & \qquad \qquad -\left\vert \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right\vert \\ & \leq & \frac{1}{4}\cdot \frac{\left ( m - m\right ) ^{2}}{m+m}\int_{\omega } \rho \left ( s\right ) \left\vert g\left ( s\right ) \right\vert ^{2}d\mu \left ( s\right ) .\end{aligned}\]]the constant @xmath59 is best possible .
the following reverse of the triangle inequality for integrals holds .
[ p.7.3.4]assume that the functions @xmath505 satisfy ( [ 7.10.4 ] ) .
then we have the inequality@xmath506 the proof follows by proposition [ p2.4.4 ] .
dragomir , a counterpart of schwarz s inequality in inner product spaces , _ rgmia res .
, _ * 6*(2003 ) , _ supplement _ , article 18 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , a generalisation of the cassels and grueb - reinboldt inequalities in inner product spaces , preprint , _ _ mathematics ar__@xmath433_iv _ , math.ca/0307130 , ` [ on line http://front.math.ucdavis.edu/ ] . `
dragomir , some companions of the grss inequality in inner product spaces , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 8 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , reverses of schwarz , triangle and bessel inequalities in inner product spaces , _ preprint _ [ ` on line ` _ _ : _ _ ` http://www.mathpreprints.com/ ` ` math / preprint / sever/20030828.2/1/ ] ` s.s .
dragomir , new reverses of schwarz , triangle and bessel inequalities in inner product spaces , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 20 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] . `
over the last five years , the development of grss type inequalities has experienced a surge , having been stimulated by their applications in different branches of applied mathematics including : in perturbed quadrature rules ( see for example @xcite , @xcite ) and in the approximation of integral transforms ( see @xcite , @xcite ) and the references therein . for two lebesgue integrable functions @xmath507
\rightarrow \mathbb{r}$ ] , consider the ebyev functional : @xmath508 in 1934 , g. grss @xcite showed that @xmath509provided @xmath510 are real numbers with the property @xmath511 .
\label{1.35.0}\ ] ] the constant @xmath59 is best possible in ( [ 1.25.0 ] ) in the sense that it can not be replaced by a smaller one .
another less well known inequality for @xmath512 was derived in 1882 by ebyev @xcite under the assumption that @xmath513 , @xmath514 exist and are continuous in @xmath515 $ ] and is given by@xmath516where @xmath517 } \left\vert f^{\prime } \left ( t\right ) \right\vert .$ ] the constant @xmath518 can not be improved in the general case .
ebyev s inequality ( [ 1.45.0 ] ) also holds if @xmath519 \rightarrow \mathbb{r}$ ] are assumed to be absolutely continuous and @xmath520 .$ ] in 1970 , a.m. ostrowski @xcite proved , amongst others , the following result that is in a sense a combination of the ebyev and grss results @xmath521provided @xmath155 is lebesgue integrable on @xmath515 $ ] and satisfying ( [ 1.35.0 ] ) with @xmath522 \rightarrow \mathbb{r}$ ] being absolutely continuous and @xmath523 .$ ] here the constant @xmath524 is also sharp .
finally , let us recall a result by lupa ( see for example 1b5.0 ) , which states that : @xmath525provided @xmath526 are absolutely continuous and @xmath527 $ ] .
the constant @xmath528 is the best possible here also . for other grss type integral inequalities ,
see the books @xcite , @xcite , and the papers @xcite @xcite , where further references are given . in @xcite
, p. cerone has obtained the following identity that involves a stieltjes integral ( lemma 2.1 , p. 3 ) : [ l2.1b5.0]let @xmath507 \rightarrow \mathbb{r}$ ] , where @xmath155 is of bounded variation and @xmath529 is continuous on @xmath515 , $ ] then the @xmath512 satisfies the identity , @xmath530where @xmath531with @xmath532 using this representation and the properties of stieltjes integrals he obtained the following result in bounding the functional @xmath533 ( theorem 2.5 , p. 4 ) : [ t2.2b5.0]with the assumptions in lemma [ l2.1b5.0 ] , we have : @xmath534 } \left\vert \psi \left ( t\right ) \right\vert \bigvee_{a}^{b}\left ( f\right ) , & \\ & \\ l\int_{a}^{b}\left\vert \psi \left ( t\right ) \right\vert dt , & \text{for } l-% \text{lipschitzian ; } \\ & \\ \int_{a}^{b}\left\vert \psi \left ( t\right ) \right\vert df\left ( t\right ) , & \text{for } f\text { monotonic nondecreasing,}%
\end{array}% \right.\end{gathered}\]]where @xmath535 denotes the total variation of @xmath155 on @xmath515 .$ ] now , if we use the function @xmath536 , @xmath537then by ( [ a.1b5.0 ] ) we may obtain the identity : @xmath538 in @xcite various upper bounds for @xmath539 have been given , from which we would like to mention only the following ones [ t2.5b5.0]let @xmath540 \rightarrow \mathbb{r}$ ] be a function of bounded variation and @xmath522 \rightarrow \mathbb{r } $ ] an absolutely continuous function so that @xmath310 is bounded on @xmath541 then one has the inequality : @xmath542where @xmath310 is as given by ( [ 2.0b5.0 ] ) and @xmath543 the case of lipschitzian functions @xmath540 \rightarrow \mathbb{r% } $ ] is embodied in the following theorem as well @xcite .
[ t2.6b5.0]let @xmath540 \rightarrow \mathbb{r}$ ] be an @xmath544lipschitzian function on @xmath515 $ ] and @xmath522 \rightarrow \mathbb{r}$ ] an absolutely continuous function on @xmath545 .$ ] then @xmath546 ; \\ & \\ l\left ( b - a\right ) ^{\frac{1}{q}}\left [ b\left ( q+1,q+1\right ) \right ] ^{% \frac{1}{q}}\left\vert \varphi \right\vert _ { p } , & p>1,\;\frac{1}{p}+\frac{1% } { q}=1 \\ & \text{if \hspace{0.05in}}\varphi \in l_{p}\left [ a , b\right ] ;
\\ \frac{l}{4}\left\vert \varphi \right\vert _ { 1 } , & \text{if \hspace{0.05in}}% \varphi \in l_{1}\left [ a , b\right ] \text{,}% \end{array}% \right.\end{aligned}\]]where @xmath547 are the usual lebesgue @xmath548norms on @xmath515 $ ] and @xmath549 is euler s beta function . finally , the following result containing stieltjes integral holds cdb5.0 : [ t2.7b5.0]let @xmath540 \rightarrow \mathbb{r}$ ] be a monotonic nondecreasing function on @xmath515 .$ ] if @xmath529 is continuous , then one has the inequality : @xmath550 ^{q}df\left ( t\right ) \right ) ^{\frac{1% } { q}}\left ( \dint_{a}^{b}\left\vert \varphi \left ( t\right ) \right\vert ^{p}df\left ( t\right ) \right ) ^{\frac{1}{p}},\ ; \\
\hfill p>1,\;\frac{1}{p}+\frac{1}{q}=1 ; \\ \\
\dfrac{1}{\left ( b - a\right ) ^{2}}\sup\limits_{t\in \left [ a , b\right ] } \left\vert \varphi \left ( t\right ) \right\vert \dint_{a}^{b}\left ( t - a\right ) \left ( b - t\right ) df\left ( t\right ) .%
\end{array}% \right.\end{aligned}\ ] ] in @xcite , the authors have considered the following functional @xmath551 \cdot \frac{1}{b - a}% \int_{a}^{b}f\left ( t\right ) dt,\]]provided that the involved integrals exist . in the same paper ,
the following result in estimating the above functional has been obtained .
[ ta.1b5.0]let @xmath552 \rightarrow \mathbb{r}$ ] be such that @xmath553 is lipschitzian on @xmath515 , $ ] i.e. , @xmath554 \;\;\left ( l>0\right)\]]and @xmath155 is riemann integrable on @xmath515 .$ ] if @xmath555 are such that @xmath556 , \]]then we have the inequality @xmath557the constant @xmath222 is sharp in the sense that it can not be replaced by a smaller constant . in @xcite , the following result complementing the above one was obtained .
[ ta.2b5.0]let @xmath552 \rightarrow \mathbb{r}$ ] be such that @xmath558 \rightarrow \mathbb{r}$ ] is of bounded variation in @xmath515 $ ] and @xmath540 \rightarrow \mathbb{r}$ ] is @xmath559lipschitzian @xmath560 then we have the inequality @xmath561the constant @xmath222 is sharp in the above sense .
the main aim of this section is to survey some recent inequalities of the grss type holding in the general setting of inner product spaces .
natural applications for lebesgue integrals in measure spaces are presented as well .
p. cerone , on an identity for the chebychev functional and some ramifications , _ j. ineq .
, _ * 3*(1 ) ( 2002 ) , article 4 [ ` on line ` ` http://jipam.vu.edu.au/v3n1/034_01.html ` ] .
p. cerone and s.s .
dragomir , new upper and lower bounds for the cebysev functional , _ j. ineq .
pure and appl .
_ , * 3*(5 ) ( 2002 ) , article 77 . `
[ on line : http://jipam.vu.edu.au/v3n5/048_02.html ]
. ` p. cerone and s.s .
dragomir , a refinement of the gruss inequality and applications , _ rgmia res .
rep . coll .
_ , * 5*(2)(2002 ) .
article 14 ` [ on line : http://rgmia.vu.edu.au/v5n2.html ] ` .
chebyshev , sue les expressions approximatives des intgrales dfinies par les autres prises entre les mme limites , _ proc .
charkov , _ * 2 * ( 1982 ) , 93 - 98 .
cheng and j. sun , note on the perturbed trapezoid inequality , _
j. inequal .
pure & appl .
math . _ * 3*(2002 ) .
2 , article 29 ` [ on line : ` ` http://jipam.vu.edu.au/v3n2/046_01.html ] . `
dragomir , a generalization of grss inequality in inner product spaces and applications , _ j. math .
_ , * 237 * ( 1999 ) , 74 - 82 .
dragomir , a grss type integral inequality for mappings of @xmath562-hlder s type and applications for trapezoid formula , _ tamkang journal of mathematics _ , * 31*(1 ) ( 2000 ) , 43 - 47 .
dragomir , some integral inequalities of grss type , _ indian j. pure appl . math . , _ * 31*(4 ) ( 2000 ) , 397 - 415 . ` ` s.s .
dragomir , some integral inequalities of grss type , _ indian j. of pure and appl . math .
, _ * 31*(4 ) ( 2000 ) , 397415 .
dragomir , sharp bounds of ebyev functional for stieltjes integrals and applications , _ rgmia res .
, _ * 5 * ( 2002 ) , supplement , article 26 .
[ ` on line ` : ` http://rgmia.vu.edu.au/v5(e).html ` ] .
dragomir , new estimates of the ebyev functional for stieltjes integrals and applications , _ rgmia res .
, _ * 5 * ( 2002 ) , supplement , article 27 [ ` on line ` : ` http://rgmia.vu.edu.au/v5(e).html ` ] .
dragomir and i. fedotov , an inequalityof grss type for riemann - stieltjes integral and applications for special means , _ tamkang j. math .
, _ * 29*(4 ) ( 1998 ) , 287 - 292 .
dragomir and i. fedotov , a grss type inequality for mappings of bounded variation and applications to numerical analysis , _ non .
, _ * 6*(3 ) ( 2001 ) , 425 - 437 .
dragomir and a. kalam , an approximation of the fourier sine transform via grss type inequalities and applications for electrical circuits , _
j. ksiam _ , * 63*(1 ) ( 2002 ) , 33 - 45 .
a.m. fink , a treatise on grss inequality , th.m .
rassias and h.m .
srivastava ( ed . ) , kluwer academic publishers , ( 1999 ) , 93114 .
g. grss , ber das maximum des absoluten betrages von @xmath563 _ math .
z. , _ * 39 * ( 1934 ) , 215 - 226 .
g. hanna , s.s .
dragomir and j. roumeliotis , an approximation for the finite - fourier transform of two independent variables , _ proc .
4th int . conf .
on modelling and simulation _ , victoria university , melbourne , 2002 , 375 - 380 d.s .
mitrinovi , j.e . peari and a.m. fink ,
_ classical and new inequalities in analysis _ , kluwer academic publishers , dordrecht , 1993 .
a.m. ostrowski , on an integral inequality , _ aequat .
, _ * 4 * ( 1970 ) , 358 - 373 .
j. peari , f. prochan and y. tong , _ convex functions , partial orderings and statistical applications , _ academic press , san diego , 1992 .
watson , serial correlation in regression analysis , i , _ biometrika _ , * 42*(1955 ) , 327 - 341 . in @xcite , the author has proved the following grss type inequality in real or complex inner product spaces .
[ 1.1.5]let @xmath564 be an inner product space over @xmath565 and @xmath566 if @xmath567 are real or complex numbers and @xmath568 are vectors in @xmath1 such that the conditions @xmath569hold , then we have the inequality @xmath570the constant @xmath59 is best possible in the sense that it can not be replaced by a smaller constant .
some particular cases of interest for integrable functions with real or complex values and the corresponding discrete versions are listed below .
[ 1.2.5]let @xmath507 \rightarrow \mathbb{k}\left ( \mathbb{% k = r}\text { , } \mathbb{c}\right ) $ ] be lebesgue integrable and such that @xmath571 \geq 0,\;\;\;\func{re}% \left [ \left ( \gamma -g\left ( x\right ) \right ) \left ( \overline{g\left ( x\right ) } -\overline{\gamma } \right ) \right ] \geq 0\]]for a.e .
@xmath572 , $ ] where @xmath573 are real or complex numbers and @xmath574 denotes the complex conjugate of @xmath575 then we have the inequality @xmath576the constant @xmath59 is best possible .
the discrete case is embodied in [ 1.3.5]let @xmath577 and @xmath578 are real or complex numbers such that @xmath579 \geq 0,\text { \ } \func{re}\left [ \left ( \gamma -y_{i}\right ) \left ( \overline{y_{i}}-\overline{\gamma } \right ) \right ] \geq 0\]]for each @xmath210 then we have the inequality @xmath580the constant @xmath59 is best possible . for other applications of theorem [ 1.1.5 ] ,
see the recent paper ssdig.5 . in the present section , by following @xcite ,
we show that the condition @xmath581 may be replaced by an equivalent but simpler assumption and a new proof of theorem [ 1.1.5 ] is produced .
a refinement of the grss type inequality @xmath582 some companions and applications for integrals are pointed out as well .
the following lemma holds @xcite .
[ l2.1.5 ] let @xmath583 be vectors in the inner product space @xmath584 over @xmath585 with @xmath586 then @xmath587if and only if @xmath588 define @xmath589a simple calculation shows that @xmath590 -\func{re}\left\langle a , a\right\rangle -\left\vert x\right\vert ^{2}\]]and thus , obviously , @xmath591 iff @xmath68 showing the required equivalence .
the following corollary is obvious [ c2.2.5 ] let @xmath592 with @xmath593 and @xmath594 with @xmath595 then @xmath596iff @xmath597 [ r2.3.5 ] if @xmath598 , then @xmath599 \geq 0\]]if and only if @xmath600where @xmath601 .
if @xmath602 , and @xmath603 then @xmath604 if and only if @xmath605 the following lemma also holds @xcite .
[ l2.2.5 ] let @xmath592 with @xmath299 then one has the following representation @xmath606 observe , for any @xmath607 , that @xmath608 \\
& = \left\vert x\right\vert ^{2}-\left\vert \left\langle x , e\right\rangle \right\vert ^{2}.\end{aligned}\]]using schwarz s inequality , we have @xmath609 ^{2 } & = \left\vert \left\langle x-\lambda e , x-\left\langle x , e\right\rangle e\right\rangle \right\vert ^{2 } \\ & \leq \left\vert x-\lambda e\right\vert ^{2}\left\vert x-\left\langle x , e\right\rangle e\right\vert ^{2 } \\ & = \left\vert x-\lambda e\right\vert ^{2}\left [ \left\vert x\right\vert ^{2}-\left\vert \left\langle x , e\right\rangle \right\vert ^{2}\right ] , \end{aligned}\]]giving the bound @xmath610taking the infimum in @xmath611 over @xmath612 , we deduce @xmath613since , for @xmath614 we get @xmath615 then the representation @xmath616 is proved .
we are now able to provide a different proof for the grss type inequality in inner product spaces ( mentioned in the introduction ) , than the one from the paper @xcite .
[ t2.1.5]let @xmath564 be an inner product space over @xmath617 and @xmath566 if @xmath618 are real or complex numbers and @xmath568 are vectors in @xmath1 such that the conditions @xmath581 hold , or equivalently , the following assumptions @xmath619are valid , then one has the inequality @xmath620the constant @xmath59 is best possible
. it can be easily shown ( see for example the proof of theorem 1 from ssd.5 ) that @xmath621 ^{\frac{1}{2}}\left [ \left\vert y\right\vert ^{2}-\left\vert \left\langle y , e\right\rangle \right\vert ^{2}\right ] ^{\frac{1}{2 } } , \label{a.5.5}\]]for any @xmath24 and @xmath566 using lemma l2.2.5 and the conditions @xmath622 we obviously have that @xmath623 ^{\frac{1}{2}}=\inf_{\lambda \in \mathbb{k}% } \left\vert x-\lambda e\right\vert \leq \left\vert x-\frac{\varphi + \phi } { 2}% \cdot e\right\vert \leq \frac{1}{2}\left\vert \phi -\varphi \right\vert\]]and @xmath624 ^{\frac{1}{2}}=\inf_{\lambda \in \mathbb{k}% } \left\vert y-\lambda e\right\vert \leq \left\vert y-\frac{\gamma + \gamma } { 2% } \cdot e\right\vert \leq \frac{1}{2}\left\vert \gamma -\gamma \right\vert\]]and by @xmath625 the desired inequality @xmath626 is obtained .
the fact that @xmath59 is the best possible constant , has been shown in @xcite and we omit the details .
the following result improving @xmath581 holds 05nssd .
[ t3.1.5]let @xmath564 be an inner product space over @xmath617 and @xmath566 if @xmath618 are real or complex numbers and @xmath568 are vectors in @xmath1 such that the conditions @xmath581 , or equivalently , @xmath627 hold , then we have the inequality @xmath628 ^{\frac{1}{2}}\left [ \func{re}\left\langle \gamma e - y , y-\gamma e\right\rangle \right ] ^{\frac{1}{2 } } \notag \\
& \leq \left ( \frac{1}{4}\left\vert \phi -\varphi \right\vert \cdot \left\vert \gamma -\gamma \right\vert \right ) .
\notag\end{aligned}\]]the constant @xmath59 is best possible . as in @xcite , we have @xmath629 \left [ \left\vert y\right\vert ^{2}-\left\vert \left\langle y , e\right\rangle \right\vert ^{2}\right ] , \label{3.2.5}\]]@xmath630 -\func{re}\left\langle \phi e - x , x-\varphi e\right\rangle\end{gathered}\]]and @xmath631 -% \func{re}\left\langle \gamma e - x , x-\gamma e\right\rangle .\end{gathered}\]]using the elementary inequality @xmath632we may state that @xmath633 \leq \frac{1}{4}\left\vert \phi -\varphi \right\vert ^{2 } \label{3.5.5}\]]and @xmath634 \leq \frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}. \label{3.6.5}\]]consequently , by @xmath635 we may state that @xmath636 ^{% \frac{1}{2}}\right ) ^{2}\right ]
\\ \times \left [ \frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}-\left ( % \left [ \func{re}\left\langle \gamma e - y , y-\gamma e\right\rangle \right ] ^{% \frac{1}{2}}\right ) ^{2}\right ] .\end{gathered}\]]finally , using the elementary inequality for positive real numbers @xmath637we have @xmath638 ^{\frac{1}{2}% } \right ) ^{2}\right ] \\ \times \left [ \frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}-\left ( % \left [ \func{re}\left\langle \gamma e - y , y-\gamma e\right\rangle \right ] ^{% \frac{1}{2}}\right ) ^{2}\right ] \\ \leq \left ( \frac{1}{4}\left\vert \phi -\varphi \right\vert \cdot \left\vert \gamma -\gamma \right\vert -\left [ \func{re}\left\langle \phi e - x , x-\varphi e\right\rangle \right ] ^{\frac{1}{2}}\left [ \func{re}\left\langle \gamma e - y , y-\gamma e\right\rangle \right ] ^{\frac{1}{2}}\right ) ^{2},\end{gathered}\]]giving the desired inequality @xmath639 the following companion of the grss inequality in inner product spaces holds @xcite . [ t4.1.5 ]
let @xmath640 be an inner product space over @xmath641 and @xmath566 if @xmath642 and @xmath24 are such that @xmath643or equivalently , @xmath644then we have the inequality @xmath645
\leq \frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}. \label{4.2.a.5}\]]the constant @xmath59 is best possible in the sense that it can not be replaced by a smaller constant .
start with the obvious inequality @xmath646since @xmath647then using @xmath648 we may write @xmath649 & = \func{re}\left [ \left\langle x-\left\langle x , e\right\rangle e , y-\left\langle y , e\right\rangle e\right\rangle \right ] \label{4.4.5 } \\ & \leq \frac{1}{4}\left\vert x-\left\langle x , e\right\rangle e+y-\left\langle y , e\right\rangle e\right\vert ^{2 } \notag \\ & = \left\vert \frac{x+y}{2}-\left\langle \frac{x+y}{2},e\right\rangle \cdot e\right\vert ^{2 } \notag \\ & = \left\vert \frac{x+y}{2}\right\vert ^{2}-\left\vert \left\langle \frac{x+y% } { 2},e\right\rangle \right\vert ^{2}. \notag\end{aligned}\]]if we apply grss inequality in inner product spaces for , say , @xmath650 we get @xmath651making use of @xmath652 and @xmath653 we deduce @xmath654 the fact that @xmath59 is the best possible constant in @xmath655 follows by the fact that if in @xmath656 we choose @xmath657 then it becomes @xmath658 implying that @xmath659 for which , by grss inequality in inner product spaces , we know that the constant @xmath660 is best possible . the following corollary might be of interest if one wanted to evaluate the absolute value of @xmath645 .\ ] ] [ c4.2.5]let @xmath564 be an inner product space over @xmath641 and @xmath566 if @xmath642 and @xmath24 are such that @xmath661or equivalently , @xmath662holds , then we have the inequality @xmath663 \right\vert \leq \frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}. \label{4.8.5}\]]if the inner product space
@xmath1 is real , then @xmath664 @xmath665or equivalently , @xmath666implies @xmath667 in both inequalities @xmath668 and @xmath669 the constant @xmath59 is best possible .
we only remark that , if @xmath670holds , then by theorem [ t4.1.5 ] , we get @xmath671 \leq \frac{1}{4}% \left\vert \gamma -\gamma \right\vert ^{2},\]]showing that @xmath645 \geq -\frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}. \label{4.12.5}\]]making use of @xmath672 and @xmath673 we deduce the desired result @xmath674 finally , we may state and prove the following dual result as well 05nssd .
[ p4.1.5]let @xmath564 be an inner product space over @xmath641 and @xmath566 if @xmath675 and @xmath24 are such that @xmath633 \leq 0 , \label{4.13.5}\]]then we have the inequalities @xmath676 ^{\frac{1}{2 } } \label{4.14.5 } \\ & \leq \frac{\sqrt{2}}{2}\left [ \left\vert x-\phi e\right\vert ^{2}+\left\vert x-\varphi e\right\vert ^{2}\right ] ^{\frac{1}{2}}. \notag\end{aligned}\ ] ] we know that the following identity holds true @xmath677see @xmath678@xmath679 + \func{re}\left\langle x-\phi e , x-\varphi e\right\rangle .\end{gathered}\]]using the assumption @xmath680 and the fact that @xmath681by @xmath682 we deduce the first inequality in @xmath683 the second inequality in @xmath684 follows by the fact that for any @xmath685 one has @xmath686the proposition is thus proved .
the following proposition holds @xcite .
[ p5.1.5 ] if @xmath687 and @xmath688 , are such that @xmath689 and @xmath690 d\mu \left ( s\right ) & \geq 0 , \label{5.1.5 } \\
\int_{\omega } \func{re}\left [ \left ( \gamma h\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{g\left ( s\right ) } -\overline{\gamma } % \overline{h\left ( s\right ) } \right ) \right ] d\mu \left ( s\right ) & \geq 0 \notag\end{aligned}\]]or , equivalently @xmath691hold , then we have the following refinement of the grss integral inequality @xmath692
d\mu \left ( s\right ) \right . \\ \times \left .
\int_{\omega } \func{re}\left [ \left ( \gamma h\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{g\left ( s\right ) } -\overline{% \gamma } \overline{h\left ( s\right ) } \right ) \right ] d\mu \left ( s\right ) % \right ] ^{\frac{1}{2}}.\end{gathered}\]]the constant @xmath59 is best possible .
the proof follows by theorem [ t3.1.5 ] on choosing @xmath693 with the inner product @xmath694we omit the details .
[ r.1.5 ] it is obvious that a sufficient condition for @xmath695 to hold is @xmath696 \geq 0,\]]and @xmath697 \geq 0,\]]for @xmath698a.e.@xmath699 or equivalently , @xmath700for @xmath698a.e.@xmath701 the following result may be stated as well @xcite .
[ c5.1.5]if @xmath702 , @xmath703 and @xmath704 are such that : @xmath705 & \geq 0 , \\ \func{re}\left [ \left ( t - g\left ( s\right ) \right ) \left ( \overline{g\left ( s\right ) } -\bar{t}\right ) \right ] & \geq 0,\text { \hspace{0.05in}for a.e .
} % s\in \omega , \end{aligned}\]]or , equivalently @xmath706then we have the inequality @xmath707
d\mu \left ( s\right ) \right . \\ \times \left .
\int_{\omega } \func{re}\left [ \left ( t - g\left ( s\right ) \right ) \left ( \overline{g\left ( s\right ) } -\bar{t}\right ) \right ] d\mu \left ( s\right ) \right ] ^{\frac{1}{2}}.\end{gathered}\ ] ] using theorem [ t4.1.5 ] we may state the following result as well 05nssd .
[ p5.3.5]if @xmath687 and @xmath708 are such that @xmath709 and @xmath710 \right .
\label{5.7.5 } \\
\times \left . \left
[ \frac{\overline{f\left ( s\right ) } + \overline{g\left ( s\right ) } } { 2}-\bar{\gamma}\bar{h}\left ( s\right ) \right ] \right\ } d\mu \left ( s\right ) \geq 0\end{gathered}\]]or equivalently , @xmath711holds , then we have the inequality @xmath712 d\mu \left ( s\right ) \\ & \qquad \qquad \qquad -\func{re}\left [ \int_{\omega } f\left ( s\right ) \overline{h\left ( s\right ) } d\mu \left ( s\right ) \cdot \int_{\omega } h\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right ] \\ & \leq \frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}.\end{aligned}\]]if ( [ 5.7.5 ] ) and ( [ 5.8.5 ] ) hold with @xmath713 instead of @xmath714 , then @xmath715 it is obvious that a sufficient condition for ( [ 5.7.5 ] ) to hold is @xmath716 \cdot \left [ \frac{\overline{f\left ( s\right ) } % + \overline{g\left ( s\right ) } } { 2}-\bar{\gamma}\bar{h}\left ( s\right ) \right ] \right\ } \geq 0,\]]for a.e .
@xmath717 or equivalently , @xmath718 finally , the following corollary holds .
[ c5.4.5]if @xmath719 , @xmath720 and @xmath704 are such that @xmath721 \geq 0,\text { \hspace{0.05in}for a.e .
} % s\in \omega \label{5.13.5}\]]or , equivalently @xmath722then we have the inequality @xmath723
d\mu \left ( s\right ) \\ & \qquad \qquad \qquad -\func{re}\left
[ \frac{1}{\mu \left ( \omega \right ) } % \int_{\omega } f\left ( s\right ) d\mu \left ( s\right ) \cdot \frac{1}{\mu \left ( \omega \right ) } \int_{\omega } \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right ] \\ &
\leq \frac{1}{4}\left\vert z - z\right\vert ^{2}.\end{aligned}\]]if ( [ 5.13.5 ] ) and ( [ 5.14.5 ] ) hold with @xmath713 instead of @xmath714 @xmath724 then @xmath725 it is obvious that if one chooses the discrete measure above , then all the inequalities in this section may be written for sequences of real or complex numbers .
we omit the details .
dragomir , a generalization of grss inequality in inner product spaces and applications , _ j. math .
_ , * 237*(1999 ) , 74 - 82 .
dragomir and i. gomm , some integral and discrete versions of the grss inequality for real and complex functions and sequences , _ rgmia res .
_ , * 5*(2003 ) , no .
3 , article 9 ` [ on line : ` ` http://rgmia.vu.edu.au/v5n3.html ] . ` s.s .
dragomir , some grss type inequalities in inner product spaces , _ j. inequal .
pure & appl . math .
_ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] the following grss type inequality in inner product spaces holds 6nssd .
[ t2.1.6]let @xmath726 with @xmath727 and the scalars @xmath728 @xmath729 such that @xmath730 and @xmath731 if @xmath732or equivalently , @xmath733holds , then we have the inequality @xmath734where @xmath735 is defined by @xmath736 } { \func{re}\left ( \bar{a}a\right ) } \right ] ^{% \frac{1}{2}}.\]]the constant @xmath59 is best possible in the sense that it can not be replaced by a smaller constant .
start with the inequality @xmath737now , assume that @xmath738 and @xmath739 with @xmath740 and @xmath741 this last inequality is equivalent to @xmath742 .\]]dividing this inequality by @xmath743 ^{\frac{1}{2}}>0,$ ] we deduce @xmath744 ^{\frac{1}{2}}}% \left\vert u\right\vert ^{2}+\left [ \func{re}\left ( \bar{c}c\right )
\right ] ^{\frac{1}{2}}\left\vert v\right\vert ^{2}\leq \frac{\func{re}\left [ c% \overline{\left\langle u , v\right\rangle } + \bar{c}\left\langle u , v\right\rangle \right ] } { \left [ \func{re}\left ( \bar{c}c\right ) \right ] ^{%
\frac{1}{2}}}. \label{2.7.6}\]]on the other hand , by the elementary inequality @xmath745we deduce @xmath746 ^{\frac{1}{2}}}\left\vert u\right\vert ^{2}+\left [ \func{re}\left ( \bar{c}c\right ) \right ] ^{\frac{1}{2}}\left\vert v\right\vert ^{2}. \label{2.8.6}\]]making use of ( [ 2.7.6 ] ) and ( [ 2.8.6 ] ) and the fact that for any @xmath747 , @xmath748 we get @xmath749 } { 2\left [ \func{re}\left ( \bar{c}c\right ) \right ] ^{% \frac{1}{2}}}\leq \frac{\left\vert c\right\vert + \left\vert c\right\vert } { 2% \left [ \func{re}\left ( \bar{c}c\right ) \right ] ^{\frac{1}{2}}}\left\vert \left\langle u , v\right\rangle \right\vert .\]]consequently @xmath750 } -1\right ] \left\vert \left\langle u , v\right\rangle \right\vert ^{2 } \label{2.9.6 } \\ & = \frac{1}{4}\frac{\left ( \left\vert c\right\vert -\left\vert c\right\vert \right ) ^{2}+4\left [ \left\vert \bar{c}c\right\vert -\func{re}\left ( \bar{c}% c\right ) \right ] } { \func{re}\left ( \bar{c}c\right ) } \left\vert \left\langle u , v\right\rangle \right\vert ^{2 } \notag \\ & = \frac{1}{4}m^{2}\left ( c , c\right ) \left\vert \left\langle u , v\right\rangle \right\vert ^{2}. \notag\end{aligned}\]]now , if we write ( [ 2.9.6 ] ) for the choices @xmath751 @xmath752 and @xmath753 @xmath752 respectively and use ( [ 2.5.6 ] ) , we deduce the desired result ( [ 2.3.6 ] ) . the sharpness of the constant will be proved in the case where @xmath1 is a real inner product space .
the following corollary which provides a simpler grss type inequality for real constants ( and in particular , for real inner product spaces ) holds @xcite .
[ c2.2.6]with the assumptions of theorem [ t2.1.6 ] and if @xmath754 are such that @xmath755 @xmath756 and @xmath757then we have the inequality @xmath758the constant @xmath59 is best possible .
the prove the sharpness of the constant @xmath59 assume that the inequality ( [ 2.11.6 ] ) holds in real inner product spaces with @xmath759 and for a constant @xmath760 i.e. , @xmath761provided that @xmath762 or equivalently , @xmath763 we choose @xmath764 @xmath765 @xmath766 then we have @xmath767and by ( [ 2.11.a.6 ] ) we get @xmath768now , if we let @xmath769 @xmath770 @xmath771 then obviously @xmath772which shows that the condition ( [ 2.10.6 ] ) is fulfilled , and by ( 2.11.b.6 ) we get @xmath773for any @xmath174 this implies @xmath774for any @xmath174 finally , let @xmath775 , @xmath776 @xmath777 then from ( 2.11.c.6 ) we get @xmath778 for any @xmath779 which produces @xmath780 [ r2.3.6]if @xmath781 are assumed not to be zero , then the inequality ( 2.3.6 ) is equivalent to @xmath782while the inequality ( [ 2.11.6 ] ) is equivalent to @xmath783the constant @xmath59 is best possible in both inequalities .
the following result holds @xcite .
[ t3.1.6]let @xmath312 be an inner product space over @xmath2 @xmath784 if @xmath424 , @xmath785 with @xmath786 and @xmath787 are such that @xmath788or equivalently , @xmath789then we have the inequality @xmath645
\leq \frac{1}{16}\cdot \frac{1}{% \lambda \left ( 1-\lambda \right ) } \left\vert \gamma -\gamma \right\vert ^{2}. \label{3.3.6}\]]the constant @xmath790 is the best possible constant in ( [ 3.3.6 ] ) in the sense that it can not be replaced by a smaller one . we know that for any @xmath791 one has @xmath792then for any @xmath793 and @xmath787 one has @xmath794since @xmath795using ( [ 3.4.6 ] ) , we have @xmath796 \label{3.5.6 } \\ & = \func{re}\left [ \left\langle x-\left\langle x , e\right\rangle e , y-\left\langle y , e\right\rangle e\right\rangle \right ] \notag \\ & \leq \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left\vert \lambda \left ( x-\left\langle x , e\right\rangle e\right ) + \left ( 1-\lambda \right ) \left ( y-\left\langle y , e\right\rangle e\right ) \right\vert ^{2 } \notag \\ & = \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left\vert \lambda x+\left ( 1-\lambda \right ) y-\left\langle \lambda x+\left ( 1-\lambda \right ) y , e\right\rangle e\right\vert ^{2}. \notag\end{aligned}\]]since , for @xmath797 with @xmath727 one has the equality @xmath798then by ( [ 3.5.6 ] ) we deduce the inequality @xmath799 \label{3.7.6 } \\ \leq \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left [ \left\vert \lambda x+\left ( 1-\lambda \right ) y\right\vert ^{2}-\left\vert \left\langle \lambda x+\left ( 1-\lambda \right ) y , e\right\rangle \right\vert ^{2}\right ] .\end{gathered}\]]now , if we apply grss inequality @xmath800provided @xmath801for @xmath802 we have @xmath803utilising ( [ 3.7.6 ] ) and ( [ 3.8.6 ] ) we deduce the desired inequality ( [ 3.3.6 ] ) . to prove the sharpness of the constant @xmath790 , assume that ( 3.3.6 ) holds with a constant @xmath78 provided that ( [ 3.1.6 ] ) is valid , i.e. , @xmath645 \leq c\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } \left\vert \gamma -\gamma \right\vert ^{2}. \label{3.9.6}\]]if in ( [ 3.9.6 ] ) we choose @xmath657 given that ( [ 3.1.6 ] ) holds with @xmath804 and @xmath805 then @xmath806provided @xmath807since we know , in grss inequality , that the constant @xmath59 is best possible , then by ( [ 3.10.6 ] ) , one has @xmath808giving , for @xmath809 @xmath810 the theorem is completely proved . the following corollary is a natural consequence of the above result 6nssd .
[ c3.2.6]assume that @xmath811 and @xmath812 are as in theorem [ t3.1.6 ] .
if @xmath813or equivalently , @xmath814then we have the inequality @xmath663 \right\vert
\leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } \left\vert \gamma -\gamma \right\vert ^{2}. \label{3.13.6}\]]the constant @xmath790 is best possible in ( [ 3.13.6 ] ) .
using theorem [ t3.1.6 ] for @xmath815 instead of @xmath15 , we have that @xmath816which implies that @xmath671 \leq \frac{1}{16}% \cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } \left\vert \gamma -\gamma \right\vert ^{2}\]]giving @xmath645 \geq -\frac{1}{16}\cdot \frac{1}{% \lambda \left ( 1-\lambda \right ) } \left\vert \gamma -\gamma \right\vert ^{2}. \label{3.14.6}\]]consequently , by ( [ 3.3.6 ] ) and ( [ 3.14.6 ] ) we deduce the desired inequality ( [ 3.13.6 ] ) .
[ r1.6]if @xmath817 with @xmath818 and , for @xmath819@xmath820then @xmath821if ( [ 3.15.6 ] ) holds with @xmath713instead of @xmath714 , then @xmath822 [ r2.6]if @xmath823 in ( [ 3.1.6 ] ) , then we obtain the result from @xcite , i.e. , @xmath824or equivalently , @xmath825implies @xmath645 \leq \frac{1}{4}\left\vert \gamma -\gamma \right\vert ^{2}. \label{3.19.6}\]]the constant @xmath59 is best possible in ( [ 3.19.6 ] ) . for @xmath809 corollary [ c3.2.6 ] and
remark [ r1.6 ] will produce the corresponding results obtained in @xcite .
we omit the details .
the following proposition holds @xcite .
[ p5.1.6 ] if @xmath687 and @xmath688 , are such that @xmath826 @xmath827 and @xmath690 d\mu \left ( s\right ) & \geq 0 , \label{5.1.6 } \\
\int_{\omega } \func{re}\left [ \left ( \gamma h\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{g\left ( s\right ) } -\overline{\gamma } % \overline{h\left ( s\right ) } \right ) \right ] d\mu \left ( s\right ) & \geq 0 , \notag\end{aligned}\]]or equivalently , @xmath691then we have the following grss type integral inequality @xmath828where@xmath829 } { \func{re}\left ( \phi \overline{\varphi } \right ) } \right ] ^{% \frac{1}{2}}.\]]the constant @xmath59 is best possible .
the proof follows by theorem [ t3.1.6 ] on choosing @xmath693 with the inner product @xmath694we omit the details .
[ r.1.6 ] it is obvious that a sufficient condition for @xmath830 to hold is @xmath696 \geq 0,\]]and @xmath697 \geq 0,\]]for @xmath698a.e.@xmath699 or equivalently , @xmath831
and@xmath832for @xmath698a.e.@xmath701 the following result may be stated as well .
[ c5.1.6]if @xmath702 , @xmath703 and @xmath704 are such that : @xmath705 & \geq 0 , \\ \func{re}\left [ \left ( t - g\left ( s\right ) \right ) \left ( \overline{g\left ( s\right ) } -\bar{t}\right ) \right ] & \geq 0,\text { \hspace{0.05in}for a.e . }
% s\in \omega\end{aligned}\]]or equivalently , @xmath833then we have the inequality @xmath834 [ r.3.6 ] the case of real functions incorporates the following interesting inequality @xmath835provided @xmath836@xmath837for @xmath698a.e .
@xmath717 where @xmath838 are real numbers and the integrals at the denominator are not zero . here
the constant @xmath59 is best possible in the sense mentioned above .
using theorem [ t3.1.6 ] we may state the following result as well 6nssd .
[ p5.3.6]if @xmath687 and @xmath708 are such that @xmath709 and@xmath839 \right .
\label{5.7.6 } \\
\times \left . \left [ \lambda \overline{f\left ( s\right ) } + \left ( 1-\lambda \right ) \overline{g\left ( s\right ) } -\bar{\gamma}\bar{h}\left ( s\right ) % \right ] \right\ } d\mu \left ( s\right ) \geq 0,\end{gathered}\]]or equivalently ,
@xmath840then we have the inequality @xmath712 d\mu \left ( s\right ) \\ & \qquad \qquad \qquad -\func{re}\left [ \int_{\omega } f\left ( s\right ) \overline{h\left ( s\right ) } d\mu \left ( s\right ) \cdot \int_{\omega } h\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right ] \\ & \leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } % \left\vert \gamma -\gamma \right\vert ^{2}.\end{aligned}\]]the constant @xmath790 is best possible .
if ( [ 5.7.6 ] ) and ( [ 5.8.6 ] ) hold with @xmath713 instead of @xmath714 ( see corollary [ c3.2.6 ] ) , then @xmath841 it is obvious that a sufficient condition for ( [ 5.7.6 ] ) to hold is @xmath842 \cdot % \left [ \lambda \overline{f\left ( s\right ) } + \left ( 1-\lambda \right ) \overline{g\left ( s\right ) } -\bar{\gamma}\bar{h}\left ( s\right ) \right ] \right\ } \geq 0\]]for a.e .
@xmath717 or equivalently , @xmath843for a.e .
@xmath844 finally , the following corollary holds .
[ c5.4.6]if @xmath719 , @xmath720 and @xmath704 are such that @xmath845 \geq 0 \label{5.13.6}\]]for a.e .
@xmath717 or , equivalently @xmath846for a.e .
@xmath717 then we have the inequality @xmath723
d\mu \left ( s\right ) \\ & \qquad \qquad \qquad -\func{re}\left [ \frac{1}{\mu \left ( \omega \right ) } % \int_{\omega } f\left ( s\right ) d\mu \left ( s\right ) \cdot \frac{1}{\mu \left ( \omega \right ) } \int_{\omega } \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right ] \\ & \leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } % \left\vert z - z\right\vert ^{2}.\end{aligned}\]]if ( [ 5.13.6 ] ) and ( [ 5.14.6 ] ) hold with @xmath713 instead of @xmath714 @xmath724 then @xmath847 [ r.4.6 ] it is obvious that if one chooses the discrete measure above , then all the inequalities in this section may be written for sequences of real or complex numbers .
we omit the details .
dragomir , a generalization of grss inequality in inner product spaces and applications , _ j. math .
_ , * 237*(1999 ) , 74 - 82 .
dragomir , some grss type inequalities in inner product spaces , _ j. inequal .
pure & appl .
_ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] . s.s .
dragomir , some companions of the grss inequality in inner product spaces , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 8 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir and i. gomm , some integral and discrete versions of the grss inequality for real and complex functions and sequences , _ rgmia res .
_ , * 5*(2003 ) , no .
3 , article 9 ` [ on line ` ` http://rgmia.vu.edu.au/v5n3.html ] . ` we may state the following result @xcite . [ t4.1.7a]let @xmath848 be an inner product space over the real or complex number field @xmath849 @xmath850 and @xmath851 with @xmath299 if @xmath852 and @xmath853then we have the inequality@xmath854the inequality ( [ 4.2.7a ] ) is sharp in the sense that the constant @xmath423 in front of @xmath855 can not be replaced by a smaller constant .
start with the inequality@xmath856using theorem [ t2.1.3 ] for @xmath857 we may state that@xmath858utilizing ( [ 4.3.7a ] ) and ( [ 4.4.7a ] ) , we deduce@xmath859which is clearly equivalent to the desired inequality ( [ 4.2.7a ] ) .
the sharpness of the constant follows by the fact that for @xmath657 @xmath860 we get from ( [ 4.2.7a ] ) that@xmath861provided @xmath593 and @xmath862 the inequality ( [ 4.6.7a ] ) is sharp , as shown in theorem t2.1.3 , and the proof is completed .
another companion of the grss inequality may be stated as well 7nssd .
[ t4.2.7a]let @xmath848 be an inner product space over @xmath2 and @xmath726 with @xmath863 suppose also that @xmath728 @xmath864 such that @xmath865 @xmath866 if either@xmath867or equivalently,@xmath868holds , then we have the inequality@xmath869the constant @xmath59 is best possible . we know ,
that@xmath870if we use corollary [ c2.3.3 ] , then we may state that@xmath871and@xmath872utilizing ( [ 4.10.7a ] ) ( [ 4.12.7a ] ) , we deduce@xmath873which is clearly equivalent to the desired inequality ( [ 4.9.7a ] ) .
the sharpness of the constant follows from corollary [ c2.3.3 ] , and we omit the details . with the assumptions of theorem [ t4.2.7a ] and
if @xmath874 ( that is actually the interesting case ) , then one has the inequality@xmath875the constant @xmath59 is best possible .
we may state the following result @xcite .
[ t4.1.7b]let @xmath848 be an inner product space over the real or complex number field @xmath849 and @xmath726 with @xmath299 if @xmath876 and @xmath853then we have the inequalities@xmath877the constant @xmath222 is best possible in the sense that it can not be replaced by a smaller constant .
start with the inequality@xmath878using theorem [ t2.1.4 ] for @xmath857 we have@xmath879and , in a similar way@xmath880utilising ( [ 4.3.7b ] ) ( [ 4.5.7b ] ) , we may state that@xmath881giving the desired inequality ( [ 4.2.7b ] ) . to prove the sharpness of the constant @xmath222 ,
let us assume that @xmath804 in ( [ 4.2.7b ] ) , to get@xmath882provided @xmath883 if @xmath884 then dividing ( [ 4.7.7b ] ) with @xmath885 we get@xmath886provided @xmath887 @xmath888 however , ( [ 4.8.7b ] ) is in fact ( [ 2.2.4 ] ) for @xmath857 for which we have shown that @xmath222 is the best possible constant .
the following result also holds @xcite .
[ t4.2.7b]with the assumptions of theorem [ t4.1.7b ] , we have the inequality@xmath889 note that , from theorem [ t2.2.4 ] , we have@xmath890provided @xmath891 taking the square of ( [ 4.10.7b ] ) and re - arranging the terms , we obtain:@xmath892provided @xmath891 using the assumption of the theorem , we then have@xmath893and @xmath894utilising ( [ 4.3.7b ] ) , ( [ 4.12.7b ] ) and ( [ 4.13.7b ] ) , we deduce the desired inequality ( [ 4.9.7b ] ) . the following result may be stated as well @xcite . [ t4.3.7b]let @xmath848 be an inner product space over @xmath2 and @xmath726 with @xmath863 suppose also that @xmath728 @xmath895 such that @xmath896 if either@xmath897or equivalently,@xmath898holds , then we have the inequality@xmath899the constant @xmath59 is best possible in ( [ 4.16.7b ] ) . from theorem [ t2.2.4
] , we may state that@xmath900and@xmath901making use of ( [ 4.3.7b ] ) and ( [ 4.17.7b ] ) , ( [ 4.18.7b ] ) , we deduce the first inequality in ( [ 4.16.7b ] ) .
the best constant follows by the use of theorem [ t2.2.4 ] , and we omit the details . finally , we may state the following theorem as well @xcite .
[ t4.4.7b]with the assumptions of theorem [ t4.3.7b ] , we have the inequality@xmath902 using theorem [ t2.2.4 ] , we may state that@xmath903this inequality implies that@xmath904giving@xmath905 .
\label{4.20.7b}\]]similarly , we have@xmath906 .
\label{4.21.7b}\]]by making use of ( [ 4.3.7b ] ) and ( [ 4.20.7b ] ) , ( [ 4.21.7b ] ) , we deduce the desired inequality ( [ 4.19.7b ] ) . for some recent results on grss type inequalities in inner product spaces , see @xcite , @xcite and @xcite .
the following grss type integral inequality for real or complex - valued functions also holds @xcite .
[ p.7.3.7a]let @xmath907 with @xmath908 and @xmath728 such that @xmath909 and@xmath910 & \geq & 0 , \\ \func{re}\left [ \left ( bh\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{g\left ( s\right ) } -\overline{b}\overline{h\left ( s\right ) } \right ) % \right ] & \geq & 0\text{,}\end{aligned}\]]for @xmath698a.e .
@xmath844 then we have the inequalities@xmath911the constant @xmath59 is best possible .
the proof follows by theorem [ t4.2.7a ] . by making use of theorem [ t4.3.7b ] , we may also state [ p7.3.7b]let @xmath907 be such that @xmath912 suppose also that @xmath913 with @xmath914 and@xmath910 & \geq & 0\text { , } \\ \func{re}\left [ \left ( bh\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{g\left ( s\right ) } -\overline{b}\overline{h\left ( s\right ) } \right ) % \right ] & \geq & 0,\text { \ \ } \end{aligned}\]]for @xmath698a.e .
@xmath844 then we have the inequality@xmath915the constant @xmath59 is best possible .
dragomir , a generalization of grss inequality in inner product spaces and applications , _ j. math .
appl . , _ * 237 * ( 1999 ) , no .
1 , 7482 . s.s .
dragomir , a note on bessel s inequality , _ austral .
_ * 28 * ( 2001 ) , no . 5 , 246248 .
dragomir , some grss type inequalities in inner product spaces , _ j. inequal .
pure & appl .
_ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] . s.s .
dragomir , some companions of the grss inequality in inner product spaces , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 8 , ` [ on line http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , reverses of schwarz , triangle and bessel inequalities in inner product spaces , _ preprint _ ` [ on line : ` _ _ _ _ ` http://www.mathpreprints.com/math/ ` ` preprint / sever/20030828.2/1/ ] . `
dragomir , new reverses of schwarz , triangle and bessel inequalities in inner product spaces , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 20 , ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . `
p. f. renaud , a matrix formulation of grss inequality , _ linear algebra appl .
_ * 335 * ( 2001 ) , 95100 .
let @xmath564 be an inner product space over @xmath2 @xmath916 and @xmath917 a finite or infinite family of _ orthonormal vectors _ in @xmath1 , i.e.,@xmath918it is well known that , the following inequality due to bessel , holds @xmath919 for any @xmath920 where the meaning of the sum is:@xmath921if @xmath564 is an infinite dimensional hilbert space and @xmath922 an orthonormal family in @xmath923 then we also have@xmath924for any @xmath925 here the meaning of the series is the usual one . in this chapter
we establish reverses of the bessel inequality and some grss type inequalities for orthonormal families , namely , upper bounds for the expressions@xmath926and@xmath927under various assumptions for the vectors @xmath568 and the orthonormal family @xmath928 in @xcite , the author has proved the following grss type inequality in real or complex inner product spaces .
[ t1.8]let @xmath564 be an inner product space over @xmath2 @xmath929 and @xmath930 @xmath299 if @xmath931 are real or complex numbers and @xmath568 are vectors in @xmath1 such that the conditions@xmath932hold , then we have the inequality@xmath933the constant @xmath59 is best possible in the sense that it can not be replaced by a smaller constant . in @xcite ,
the following refinement of ( [ 1.2.8 ] ) has been pointed out .
[ t2.8]let @xmath923 @xmath2 and @xmath934 be as in theorem [ t1.8 ] . if @xmath935 satisfy ( [ 1.1.8 ] ) or equivalently @xmath936then@xmath937 ^{\frac{1}{2}}\left [ \func{re}\left\langle \gamma e - y , y-\gamma e\right\rangle \right ] ^{\frac{1}{2}}.\end{gathered}\ ] ] in @xcite , n. ujevi has generalised theorem [ t1.8 ] for the case of real inner product spaces as follows .
[ t3.8]let @xmath564 be an inner product space over the real number field @xmath938 , and @xmath939 an orthornormal family in @xmath26 if @xmath940 , @xmath941 satisfy the condition@xmath942then one has the inequality:@xmath943 ^{\frac{1}{2}% } .
\label{1.6.8}\]]the constant @xmath59 is best possible in the sense that it can not be replaced by a smaller constant .
we note that the key point in his proof is the following identity:@xmath944holding for @xmath920 @xmath945 , @xmath946 and @xmath947 an orthornormal family of vectors in the real inner product space @xmath26 in this section , by following @xcite , we point out a reverse of bessel s inequality in both real and complex inner product spaces .
this result will then be employed to provide a refinement of the grss type inequality @xmath948 for real or complex inner products .
related results as well as integral inequalities for general measure spaces are also given .
the following lemma holds @xcite .
[ l2.1.8]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath950 and @xmath951 @xmath952 real or complex numbers .
the following statements are equivalent for @xmath953 1 .
@xmath954 2 .
@xmath955 it is easy to see that for @xmath956 the following are equivalent ( see ( * ? ? ?
* lemma 1 ) ) 1 .
@xmath957 and 2 .
@xmath958 now , for @xmath959 @xmath960 we have@xmath961giving , for @xmath962 the desired equivalence .
the following reverse of bessel s inequality holds @xcite .
[ t2.2.8]let @xmath963 @xmath964 @xmath965 @xmath966 and @xmath5 such that either ( i ) or ( ii ) of lemma l2.1.8 holds .
then we have the inequality:@xmath967the constant @xmath59 is best in both inequalities .
.\]]observe that@xmath970 + \sum_{i\in h}\func{re}\left [ \overline{\phi _ { i}}\left\langle x , e_{i}\right\rangle \right ] -\sum_{i\in h}\func{re}\left [ \phi _ { i}\overline{\phi _ { i}}\right ] -\sum_{i\in h}\left\vert \left\langle x , e_{i}\right\rangle \right\vert ^{2}\]]and@xmath971 \\ & = \sum_{i\in h}\func{re}\left [ \phi _ { i}\overline{\left\langle x , e_{i}\right\rangle } \right ] + \sum_{i\in h}\func{re}\left [ \overline{\phi _ { i}}\left\langle x , e_{i}\right\rangle \right ] -\left\vert x\right\vert ^{2}-\sum_{i\in h}\func{re}\left [ \phi _ { i}\overline{\phi _ { i}}\right ] .\end{aligned}\]]consequently , subtracting @xmath972 from @xmath973 we deduce the following equality that is interesting in its turn@xmath974 \label{2.5.8 } \\ -\func{re}\left [ \left\langle \sum_{i\in h}\phi _ { i}e_{i}-x , x-\sum_{i\in h}\phi _ { i}e_{i}\right\rangle \right ] .\end{gathered}\]]using the following elementary inequality for complex numbers@xmath975 \leq \frac{1}{4}\left\vert a+b\right\vert ^{2},\ \ \ \ a , b\in \mathbb{k},\]]for the choices @xmath976 @xmath977 @xmath952 we deduce@xmath978 \leq \frac{1}{4}% \sum_{i\in h}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}. \label{2.6.8}\]]making use of ( [ 2.5.8 ] ) , ( [ 2.6.8 ] ) and the assumption ( i ) , we deduce ( [ 2.4.8 ] ) .
the sharpness of the constant @xmath59 was proved for a single element @xmath979 @xmath593 in @xcite , or for the real case in @xcite .
we can give here a simple proof as follows .
assume that there is a @xmath334 such that@xmath980provided @xmath981 @xmath14 and @xmath949 satisfy ( i ) or ( ii ) .
we choose @xmath982 @xmath983 @xmath984 @xmath985 @xmath986 to get from ( [ 2.7.8 ] ) that@xmath987provided@xmath988if we choose @xmath989 @xmath990 then ( [ 2.9.9 ] ) is fulfilled and by ( [ 2.8.9a ] ) we get @xmath991 giving @xmath992 the following result holds @xcite . [
t3.1.8]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath950 and @xmath965 @xmath993 and @xmath12 if either@xmath994or equivalently,@xmath995hold , then we have the inequalities@xmath996 ^{\frac{1}{2}}% \left [ \func{re}\left\langle \sum_{i\in f}\gamma _ { i}e_{i}-y , y-\sum_{i\in f}\gamma _ { i}e_{i}\right\rangle \right ] ^{\frac{1}{2 } } \notag \\ & \leq \frac{1}{4}\left ( \sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}\cdot \left ( \sum_{i\in f}\left\vert \gamma _ { i}-\gamma _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}. \notag\end{aligned}\]]the constant @xmath59 is best possible . using schwartz s inequality in the inner product space @xmath584 one has@xmath997and since a simple calculation shows that @xmath998and @xmath999for any @xmath354 then by ( [ 3.4.8 ] ) and by the reverse of bessel s inequality in theorem [ t2.2.8 ] , we have@xmath1000 \notag \\ & \quad \times \left [ \frac{1}{4}\sum_{i\in f}\left\vert \gamma _ { i}-\gamma _ { i}\right\vert ^{2}-\func{re}\left\langle \sum_{i\in f}\gamma _ { i}e_{i}-y , y-\sum_{i\in f}\gamma _ { i}e_{i}\right\rangle \right ] \notag \\ & \leq \left [ \frac{1}{4}\left ( \sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}\cdot \left ( \sum_{i\in f}\left\vert \gamma _ { i}-\gamma _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}% } \right .
\notag \\ & \quad -\left . \left
[ \func{re}\left\langle \sum_{i\in f}\phi _ { i}e_{i}-x , x-\sum_{i\in f}\phi _ { i}e_{i}\right\rangle \right ] ^{\frac{1}{2}}% \left [ \func{re}\left\langle \sum_{i\in f}\gamma _ { i}e_{i}-y , y-\sum_{i\in f}\gamma _ { i}e_{i}\right\rangle \right ] ^{\frac{1}{2}}\right ] \notag\end{aligned}\]]where , for the last inequality , we have made use of the inequality@xmath637holding for any @xmath1001 taking the square root in ( [ 3.5.8 ] ) and observing that the quantity in the last square bracket is nonnegative ( see for example ( [ 2.4.8 ] ) ) , we deduce the desired result ( [ 3.3.8 ] ) .
the best constant has been proved in @xcite for one element and we omit the details .
the following companion of the grss inequality also holds @xcite .
[ t4.1.8]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath950 and @xmath1002 and @xmath24 such that@xmath1003or equivalently,@xmath1004holds , then we have the inequality@xmath1005 \leq \frac{1}{% 4}\sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}. \label{4.3.8}\]]the constant @xmath59 is best possible .
start with the well known inequality @xmath1006since@xmath1007for any @xmath354 then , by ( [ 4.4.8 ] ) , we get@xmath1008 \label{4.5.8 } \\ & = \func{re}\left [ \left\langle x-\sum_{i\in f}\left\langle x , e_{i}\right\rangle e_{i},y-\sum_{i\in f}\left\langle y , e_{i}\right\rangle e_{i}\right\rangle \right ] \notag \\ & \leq \frac{1}{4}\left\vert x-\sum_{i\in f}\left\langle x , e_{i}\right\rangle e_{i}+y-\sum_{i\in
f}\left\langle y , e_{i}\right\rangle e_{i}\right\vert ^{2 } \notag \\ & = \left\vert \frac{x+y}{2}-\sum_{i\in f}\left\langle \frac{x+y}{2}% , e_{i}\right\rangle e_{i}\right\vert ^{2 } \notag \\ & = \left\vert \frac{x+y}{2}\right\vert ^{2}-\sum_{i\in f}\left\vert \left\langle \frac{x+y}{2},e_{i}\right\rangle \right\vert ^{2}. \notag\end{aligned}\]]if we apply the reverse of bessel s inequality in theorem [ t2.2.8 ] for @xmath1009 we may state that@xmath1010now , by making use of ( [ 4.5.8 ] ) and ( [ 4.6.8 ] ) , we deduce ( [ 4.3.8 ] ) .
the fact that @xmath59 is the best constant in ( [ 4.3.8 ] ) follows by the fact that if in ( [ 4.1.8 ] ) we choose @xmath657 then it becomes ( i ) of lemma [ l2.1.8 ] , implying ( [ 2.4.8 ] ) , for which , we have shown that @xmath660 was the best constant .
the following corollary may be of interest if we wish to evaluate the absolute value of@xmath1005 .\ ] ] [ c4.2.8]with the assumptions of theorem [ t4.1.8 ] and if @xmath1011or equivalently,@xmath1012holds , then we have the inequality@xmath1013 \right\vert \leq \frac{1}{4}\sum_{i\in f}\left\vert \phi _
{ i}-\phi _ { i}\right\vert ^{2}. \label{4.9.8}\ ] ] we only remark that , if@xmath1014holds , then by theorem [ t4.1.8 ] for @xmath815 instead of @xmath1015 we have@xmath1016 \leq \frac{1}{% 4}\sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2},\]]showing that@xmath1005 \geq -\frac{1% } { 4}\sum_{i\in f}\left\vert \phi
_ { i}-\phi _ { i}\right\vert ^{2}. \label{4.10.8}\]]making use of ( [ 4.3.8 ] ) and ( [ 4.10.8 ] ) , we deduce the desired inequality ( [ 4.9.8 ] ) .
[ r4.3.8]if @xmath1 is a real inner product space and @xmath1017 with the property that@xmath1018or equivalently,@xmath1019then we have the grss type inequality@xmath1020 let @xmath147 be a measure space consisting of a set @xmath148 @xmath149 a @xmath150algebra of parts and @xmath406 a countably additive and positive measure on @xmath149 with values in @xmath1021 let @xmath408 be a @xmath698measurable function on @xmath159 denote by @xmath154 the hilbert space of all real or complex valued functions defined on @xmath151 and @xmath411integrable on @xmath148 i.e.,@xmath412 consider the family @xmath1022 of functions in @xmath1023 with the properties that@xmath1024where @xmath1025 is @xmath1026 if @xmath1027 and @xmath1028 if @xmath1029 @xmath1030 is an orthornormal family in @xmath1031 the following proposition holds @xcite .
[ p5.1.8]let @xmath1022 be an orthornormal family of functions in @xmath1032 @xmath949 a finite subset of @xmath1033 @xmath1034 @xmath1035 and @xmath1036 such that either@xmath1037 d\mu \left ( s\right ) \geq 0\end{gathered}\]]or equivalently,@xmath1038then we have the inequality@xmath1039 d\mu \left ( s\right ) \notag \\ & \leq \frac{1}{4}\sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}. \notag\end{aligned}\]]the constant @xmath59 is best possible in both inequalities .
the proof follows by theorem [ t2.2.8 ] applied for the hilbert space @xmath1023 and the orthornormal family @xmath1040 the following grss type inequality also holds @xcite .
[ p5.2.8]let @xmath1022 and @xmath949 be as in proposition [ p5.1.8 ] . if @xmath1041 @xmath1042 and @xmath505 so that either@xmath1043 d\mu \left ( s\right ) & \geq 0 , \\
\int_{\omega } \rho \left ( s\right ) \func{re}\left [ \left ( \sum_{i\in f}\gamma _ { i}f_{i}\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{% g}\left ( s\right ) -\sum_{i\in f}\overline{\gamma _ { i}}\text { } \overline{f_{i}% } \left ( s\right ) \right ) \right ] d\mu \left ( s\right ) & \geq 0,\end{aligned}\]]or equivalently,@xmath1044hold , then we have the inequalities@xmath1045@xmath1046 d\mu \left ( s\right ) \right ] ^{\frac{1}{2 } } \\ & \quad \times \left [ \int_{\omega } \rho \left ( s\right ) \func{re}\left [ \left ( \sum_{i\in f}\gamma _ { i}f_{i}\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{g}\left ( s\right ) -\sum_{i\in f}\overline{\gamma _ { i}}\overline{f_{i}}\left ( s\right ) \right ) \right ] d\mu \left ( s\right ) % \right ] ^{\frac{1}{2 } } \\ & \leq \frac{1}{4}\left ( \sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}\left ( \sum_{i\in f}\left\vert \gamma _ { i}-\gamma _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}.\end{aligned}\]]the constant @xmath59 is the best possible .
the proof follows by theorem [ t3.1.8 ] and we omit the details .
[ r5.3.8]similar results may be stated if we apply the other inequalities obtained above .
we omit the details . in the case of real spaces ,
the following corollaries provide much simpler sufficient conditions for the reverse of bessel s inequality ( [ 5.5.8 ] ) or for the grss type inequality ( [ 5.8.8 ] ) to hold .
[ c5.4.8]let @xmath1022 be an orthornormal family of functions in the real hilbert space @xmath1047 @xmath949 a finite part of @xmath1033 @xmath1048 @xmath1049 and @xmath1050 such that@xmath1051then we have the inequalities@xmath1052 ^{2 } \\ &
\leq \frac{1}{4}\sum_{i\in f}\left ( m_{i}-m_{i}\right ) ^{2 } \\ & \ \ \ \ \ \ -\int_{\omega } \rho \left ( s\right ) \left ( \sum_{i\in f}m_{i}f_{i}\left ( s\right ) -f\left ( s\right )
\right ) \left ( f\left ( s\right ) -\sum_{i\in f}m_{i}f_{i}\left ( s\right ) \right ) d\mu \left ( s\right ) \\ & \leq \frac{1}{4}\sum_{i\in f}\left ( m_{i}-m_{i}\right ) ^{2}.\end{aligned}\]]the constant @xmath59 is best possible .
[ c5.5.8]let @xmath1022 and @xmath949 be as in corollary [ c5.4.8 ] . if @xmath1053 @xmath1049 and @xmath1054 are such that@xmath1055and@xmath1056then we have the inequalities@xmath1057@xmath1058 ^{\frac{1}{2 } } \\
& \quad \times \left [ \int_{\omega } \rho \left ( s\right ) \left ( \sum_{i\in f}n_{i}f_{i}\left ( s\right ) -g\left ( s\right ) \right ) \left ( g\left ( s\right ) -\sum_{i\in f}n_{i}f_{i}\left ( s\right ) \right ) d\mu \left ( s\right ) \right ] ^{\frac{1}{2 } } \\ & \leq \frac{1}{4}\left ( \sum_{i\in f}\left ( m_{i}-m_{i}\right ) ^{2}\right ) ^{\frac{1}{2}}\left ( \sum_{i\in f}\left ( n_{i}-n_{i}\right ) ^{2}\right ) ^{% \frac{1}{2}}.\end{aligned}\ ] ] s.s .
dragomir , a generalisation of grss inequality in inner product spaces and applications , _ j. math .
, _ * 237 * ( 1999 ) , 74 - 82 .
dragomir , some grss type inequalities in inner product spaces , _ j. inequal .
pure & appl .
_ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] . s.s .
dragomir , a counterpart of bessel s inequality in inner product spaces and some grss type related results , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 10 , ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . `
n. ujevi , a generalisation of grss inequality in prehilbertian spaces , _ math .
& appl . , _ to appear .
the following lemma holds @xcite .
[ l2.1.9]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath1033 @xmath1059 @xmath966 , @xmath315 and @xmath925 if@xmath1060then we have the inequality@xmath1061 consider@xmath1062and@xmath1063 \\ & = \sum_{i\in f}\left\vert \lambda _ { i}\right\vert ^{2}+\sum_{i\in f}\left\vert \left\langle x , e_{i}\right\rangle \right\vert ^{2}-\sum_{i\in f}% \overline{\lambda _ { i}}\left\langle x , e_{i}\right\rangle -\sum_{i\in f}\lambda _ { i}\overline{\left\langle x , e_{i}\right\rangle } .\end{aligned}\]]if we subtract @xmath972 from @xmath1064 we deduce the following identity that is interesting in its own right@xmath1065from which we easily deduce ( [ 2.2.9 ] ) .
the following reverse of bessel s inequality holds @xcite .
[ t2.2.9]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath1033 @xmath1066 @xmath1067 @xmath1068 real or complex numbers . for @xmath920 if either 1 .
@xmath1069 + or equivalently , 2 .
@xmath1070 holds , then the following reverse of bessel s inequality@xmath1071is valid .
the constant @xmath59 is best possible in both inequalities .
if we apply lemma [ l2.1.9 ] for @xmath1072 and @xmath1073we deduce the first inequality in ( [ 2.4.9 ] ) .
let us prove that @xmath59 is best possible in the second inequality in ( [ 2.4.9 ] ) .
assume that there is a @xmath334 such that@xmath1074provided that @xmath1066 @xmath1075 and @xmath949 satisfy ( i ) and ( ii ) .
we choose @xmath982 @xmath1076 @xmath1077 @xmath1078 @xmath985 @xmath1079 to get from ( [ 2.7.9 ] ) that@xmath1080provided@xmath1081from ( [ 2.8.9 ] ) we get@xmath1082provided ( [ 2.9.9a ] ) holds .
if we choose @xmath989 @xmath990 then ( [ 2.9.9a ] ) is fulfilled and by ( [ 2.10.9 ] ) we get @xmath991 giving @xmath992 [ r2.3.9]if @xmath982 @xmath1083 @xmath1084 and for @xmath1085 and @xmath5 one has either@xmath1086or equivalently,@xmath1087then@xmath1088the constant @xmath59 is best possible in both inequalities .
[ r2.4.9]it is important to compare the bounds provided by theorem t2.2.8 and theorem [ t2.2.9 ] . for this purpose , consider@xmath1089and@xmath1090where @xmath1 is a real inner product , @xmath930 @xmath727 @xmath1091 @xmath1092 with @xmath1093or equivalently , @xmath1094 if we choose @xmath1095 @xmath1096 then we have@xmath1097provided @xmath1098 consider @xmath764 @xmath1099 and @xmath1100 then @xmath1101 and we must compare@xmath1102with@xmath1103provided @xmath1104 if we choose @xmath1105 then @xmath1106 and @xmath1107 showing that @xmath1108 if we choose @xmath1109 then @xmath1110 @xmath1111 showing that @xmath1112 we may state the following proposition .
[ p2.5.9]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath1033 @xmath1066 @xmath1113 @xmath1042 .
if @xmath5 either satisfies ( i ) , or equivalently , ( ii ) of theorem [ t2.2.9 ] , then the upper bounds@xmath1114for the bessel s difference @xmath1115 can not be compared in general .
the following result holds @xcite .
[ t3.1.9]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath950 , @xmath981 @xmath993 and @xmath12 if either@xmath994or equivalently,@xmath995hold , then we have the inequalities@xmath1116the constant @xmath59 is best possible . using schwartz s inequality in the inner product space @xmath584 one has@xmath1117and since a simple calculation shows that @xmath998and @xmath999for any @xmath354 then by ( [ 3.4.9 ] ) and by the reverse of bessel s inequality in theorem [ t2.2.9 ] , we have@xmath1118 \notag \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \times \left [ \frac{1}{4}\sum_{i\in f}\left\vert \gamma _ { i}-\gamma _ { i}\right\vert ^{2}-\sum_{i\in f}\left\vert \frac{\gamma _ { i}+\gamma _ { i}}{2}-\left\langle y , e_{i}\right\rangle \right\vert ^{2}\right ] \notag \\ & : = k .
\notag\end{aligned}\]]using aczl s inequality for real numbers , i.e. , we recall that@xmath1119provided that @xmath1120 @xmath1121 we may state that@xmath1122 ^{2}.\end{gathered}\]]using ( [ 3.5.9 ] ) and ( [ 3.7.9 ] ) we conclude that@xmath1123 ^{2}.\end{gathered}\]]taking the square root in ( [ 3.8.9 ] ) and taking into account that the quantity in the last square bracket is nonnegative ( this follows by ( 2.4.9 ) and by the cauchy - bunyakovsky - schwarz inequality ) , we deduce the second inequality in ( [ 3.3.9 ] ) . the fact that @xmath59 is the best possible constant follows by theorem [ t2.2.9 ] and we omit the details .
the following corollary may be stated @xcite .
[ c3.2.9]let @xmath930 @xmath727 @xmath1124 and @xmath24 are such that either@xmath1125or equivalently,@xmath936hold
. then we have the following refinement of grss inequality@xmath1126the constant @xmath59 is best possible in both inequalities .
the following companion of the grss inequality also holds @xcite .
[ t4.1.9]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath950 and @xmath1002 , @xmath24 and @xmath1127 such that either@xmath1128or equivalently,@xmath1129holds .
then we have the inequality@xmath1008 \label{4.3.9 } \\ & \leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } % \sum_{i\in f}\left\vert \phi
_ { i}-\phi _ { i}\right\vert ^{2 } \notag \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ -\frac{1}{4}\frac{1}{\lambda \left ( 1-\lambda \right ) } \sum_{i\in f}\left\vert \frac{\phi _ { i}+\phi _ { i}}{2}-\left\langle \lambda x+\left ( 1-\lambda \right ) y , e_{i}\right\rangle \right\vert ^{2 } \notag \\ & \leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } % \sum_{i\in f}\left\vert \phi _ { i}-\phi _
{ i}\right\vert ^{2}. \notag\end{aligned}\]]the constant @xmath790 is the best possible constant in ( [ 4.3.9 ] ) in the sense that it can not be replaced by a smaller constant .
we know that for any@xmath1130 one has@xmath792then for any @xmath793 and @xmath787 one has@xmath1131since@xmath1007for any @xmath354 then , by ( [ 4.4.9 ] ) , we get@xmath1008 \label{4.5.9 } \\ & = \func{re}\left [ \left\langle x-\sum_{i\in f}\left\langle x , e_{i}\right\rangle e_{i},y-\sum_{i\in f}\left\langle y , e_{i}\right\rangle e_{i}\right\rangle \right ] \notag \\ & \leq \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left\vert \lambda \left ( x-\sum_{i\in f}\left\langle x , e_{i}\right\rangle e_{i}\right ) + \left ( 1-\lambda \right ) \left ( y-\sum_{i\in f}\left\langle y , e_{i}\right\rangle e_{i}\right ) \right\vert ^{2 } \notag \\ & = \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left\vert \lambda x+\left ( 1-\lambda \right ) y-\sum_{i\in f}\left\langle \lambda x+\left ( 1-\lambda \right ) y , e_{i}\right\rangle e_{i}\right\vert ^{2 } \notag \\ & = \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left [ \left\vert \lambda x+\left ( 1-\lambda \right ) y\right\vert ^{2}-\sum_{i\in f}\left\vert \left\langle \lambda x+\left ( 1-\lambda \right ) y , e_{i}\right\rangle \right\vert ^{2}\right ] .
\notag\end{aligned}\]]if we apply the reverse of bessel s inequality in theorem [ t2.2.9 ] for @xmath1132 we may state that@xmath1133now , by making use of ( [ 4.5.9 ] ) and ( [ 4.6.9 ] ) , we deduce ( [ 4.3.9 ] ) .
the fact that @xmath790 is the best possible constant in ( [ 4.3.9 ] ) follows by the fact that if in ( [ 4.1.9 ] ) we choose @xmath657 then it becomes ( i ) of theorem [ t2.2.9 ] , implying for @xmath823 ( [ 2.4.9 ] ) , for which , we have shown that @xmath59 was the best constant . [
r1.9]in practical applications we may use only the inequality between the first and the last terms in ( [ 4.3.9 ] ) . [ r2.9]if in theorem [ t4.1.9 ] , we choose @xmath809 then we get@xmath1008 \\ & \leq \frac{1}{4}\sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}-\sum_{i\in f}\left\vert \frac{\phi
_ { i}+\phi _ { i}}{2}-\left\langle \frac{x+y}{2},e_{i}\right\rangle \right\vert ^{2 } \\ & \leq \frac{1}{4}\sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2},\end{aligned}\]]provided@xmath1134or equivalently,@xmath1135 [ c4.2.9]with the assumptions of theorem [ t4.1.9 ] and if@xmath1136or ,
equivalently@xmath1137then we have the inequality@xmath1013 \right\vert \leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } \sum_{i\in f}\left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}. \label{4.10.9}\]]the constant @xmath790 is best possible in ( [ 4.10.9 ] ) .
[ r4.3.9]if @xmath1 is a real inner product space and @xmath1017 with the property@xmath1138or equivalently,@xmath1139 ^{\frac{1}{2}},\]]then we have the grss type inequality@xmath1140 the following proposition holds @xcite .
[ p5.1.9]let @xmath1022 be an orthornormal family of functions in @xmath1032 @xmath949 a finite subset of @xmath1033 @xmath1034 @xmath1035 and @xmath1036 so that either@xmath1141 d\mu \left ( s\right ) \geq 0\]]or equivalently,@xmath1142then we have the inequality@xmath1143the constant @xmath59 is best possible in both inequalities .
the proof follows by theorem [ t2.2.9 ] applied for the hilbert space @xmath1023 and the orthornormal family @xmath1040 the following grss type inequality also holds @xcite .
[ p5.2.9]let @xmath1022 and @xmath949 be as in proposition [ p5.1.9 ] . if @xmath1041 @xmath1042 and @xmath505 so that either@xmath1043 d\mu \left ( s\right ) & \geq 0 , \\ \int_{\omega } \rho \left ( s\right ) \func{re}\left [ \left ( \sum_{i\in f}\gamma _ { i}f_{i}\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{% g}\left ( s\right ) -\sum_{i\in f}\overline{\gamma
_ { i}}\text { } \overline{f_{i}% } \left ( s\right ) \right ) \right ] d\mu \left ( s\right ) & \geq 0,\end{aligned}\]]or equivalently,@xmath1044hold , then we have the inequalities@xmath1144the constant @xmath59 is the best possible .
the proof follows by theorem [ t3.1.9 ] and we omit the details . [ r5.3.9]similar results may be stated if one applies the inequalities in the above subsections .
we omit the details . in the case of real spaces ,
the following corollaries provide much simpler sufficient conditions for the reverse of bessel s inequality ( [ 5.5.9 ] ) or for the grss type inequality ( [ 5.8.9 ] ) to hold .
[ c5.4.9]let @xmath1022 be an orthornormal family of functions in the real hilbert space @xmath1047 @xmath949 a finite part of @xmath1033 @xmath1048 @xmath1049 and @xmath1050 so that@xmath1051then we have the inequalities@xmath1052 ^{2 } \\
& \leq \frac{1}{4}\sum_{i\in f}\left ( m_{i}-m_{i}\right ) ^{2}-\sum_{i\in f}% \left [ \frac{m_{i}+m_{i}}{2}-\int_{\omega } \rho \left ( s\right ) f\left ( s\right ) f_{i}\left ( s\right ) d\mu \left ( s\right ) \right ] ^{2 } \\ & \leq \frac{1}{4}\sum_{i\in f}\left ( m_{i}-m_{i}\right ) ^{2}.\end{aligned}\]]the constant @xmath59 is best possible .
[ c5.5.9]let @xmath1022 and @xmath949 be as in corollary [ c5.4.9 ] . if @xmath1053 @xmath1049 and @xmath1054 are such that@xmath1055and@xmath1145hold , then we have the inequalities@xmath1057@xmath1146 s.s .
dragomir , a generalisation of grss inequality in inner product spaces and applications , _ j. math .
appl . , _ * 237 * ( 1999 ) , 74 - 82
dragomir , some grss type inequalities in inner product spaces , _ j. inequal .
pure & appl .
_ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] . s.s .
dragomir , a counterpart of bessel s inequality in inner product spaces and some grss type related results , _ rgmia res .
_ , * 6*(2003 ) , supplement , article 10 . `
[ on line : http://rgmia.vu.edu.au/v6(e).html ] .
dragomir , on bessel and grss inequalities for orthornormal families in inner product spaces , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 12 , ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . `
n. ujevi , a generalisation of grss inequality in prehilbertian spaces , _ math .
ineq . & appl . , _ to appear .
the following reverse of bessel s inequality holds @xcite .
[ t2.1.10]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath1033 and @xmath951 @xmath952 real or complex numbers such that @xmath1147 if @xmath5 is such that either 1 .
@xmath1069or equivalently , 2 .
@xmath1070 holds , then one has the inequality @xmath1148the constant @xmath59 is best possible in the sense that it can not be replaced by a smaller constant .
observe that@xmath1149 -\left\vert x\right\vert ^{2}-\sum_{i\in f}\func{re}\left ( \phi _ { i}% \overline{\phi _ { i}}\right ) , \end{gathered}\]]giving , from ( i ) , that@xmath1150 . \label{2.4.10}\ ] ] on the other hand , by the elementary inequality@xmath1151we deduce@xmath1152 ^{% \frac{1}{2}}}+\left [ \sum_{i\in f}\func{re}\left ( \phi _ { i}\overline{\phi _ { i}}\right ) \right ] ^{\frac{1}{2}}. \label{2.5.10}\]]dividing ( [ 2.4.10 ] ) by @xmath1153 ^{\frac{1}{2}}>0 $ ] and using ( [ 2.5.10 ] ) , we obtain@xmath1154 } { \left [ \sum_{i\in f}\func{re}% \left ( \phi _ { i}\overline{\phi _ { i}}\right ) \right ] ^{\frac{1}{2 } } } , \label{2.6.10}\]]which is also an interesting inequality in itself . using the cauchy - bunyakovsky - schwarz inequality for real numbers , we get@xmath1155 & \leq \sum_{i\in f}\left\vert \phi _ { i}\overline{\left\langle x , e_{i}\right\rangle } + \overline{\phi _ { i}}\left\langle x , e_{i}\right\rangle \right\vert \label{2.7.10 } \\ & \leq \sum_{i\in f}\left ( \left\vert \phi _ { i}\right\vert + \left\vert \phi _ { i}\right\vert \right ) \left\vert \left\langle x , e_{i}\right\rangle \right\vert \notag \\ & \leq \left [ \sum_{i\in f}\left ( \left\vert \phi _ { i}\right\vert + \left\vert \phi _ { i}\right\vert \right ) ^{2}\right ] ^{\frac{1}{2}}\left [ \sum_{i\in f}\left\vert \left\langle x , e_{i}\right\rangle \right\vert ^{2}% \right ] ^{\frac{1}{2}}. \notag\end{aligned}\]]making use of ( [ 2.6.10 ] ) and ( [ 2.7.10 ] ) , we deduce the desired result ( [ 2.1.10 ] ) . to prove the sharpness of the constant @xmath373 let us assume that ( [ 2.1.10 ] ) holds with a constant @xmath232 i.e. , @xmath1156provided @xmath1157 @xmath1158 satisfies ( i ) .
choose @xmath982 @xmath1159 @xmath727 @xmath1160 @xmath1161 with @xmath1162 then , by ( [ 2.8.10 ] ) , we get@xmath1163provided@xmath1164if @xmath1165 then obviously ( [ 2.10.10 ] ) holds , and by ( [ 2.9.10 ] ) we get@xmath1166giving @xmath1167 for @xmath239 now , if in this inequality we choose @xmath1168 @xmath1169 @xmath1170 then we get @xmath383 for @xmath339 from where we deduce @xmath1171 [ r2.2.10]by the use of ( [ 2.6.10 ] ) , the second inequality in ( 2.7.10 ) and the hlder inequality , we may state the following reverses of bessel s inequality as well:@xmath1172 ^{\frac{1}{2}}% } \\ \times \left\ { \begin{array}{l } \max\limits_{i\in f}\left\ { \left\vert \phi _ { i}\right\vert + \left\vert \phi _ { i}\right\vert \right\ } \sum\limits_{i\in f}\left\vert \left\langle x , e_{i}\right\rangle \right\vert ; \\
\\ \left [ \sum\limits_{i\in f}\left ( \left\vert \phi _ { i}\right\vert + \left\vert \phi _ { i}\right\vert \right ) ^{p}\right ] ^{\frac{1}{p}}\left ( \sum\limits_{i\in f}\left\vert \left\langle x , e_{i}\right\rangle \right\vert ^{q}\right ) ^{\frac{1}{q}},\text { } \\ \hfill \text{for\ } p>1,\ \frac{1}{p}+\frac{1}{q}=1 ; \\ \\ \max_{i\in f}\left\vert \left\langle x , e_{i}\right\rangle \right\vert \sum\limits_{i\in f}\left [ \left\vert \phi _ { i}\right\vert + \left\vert \phi _ { i}\right\vert \right ] .% \end{array}% \right.\end{gathered}\ ] ] the following corollary holds @xcite .
[ c2.3.10]with the assumption of theorem [ t2.1.10 ] and if either ( i ) or ( ii ) holds , then@xmath1173where@xmath1174 \right\ } } { \sum_{i\in f}\func{re}\left ( \phi _ { i}\overline{\phi _ { i}}\right ) } \right ] ^{\frac{1}{2}}.\]]the constant @xmath59 is best possible
. the inequality ( [ 2.12.10 ] ) follows by ( [ 2.1.10 ] ) on subtracting the same quantity @xmath1175 from both sides . to prove the sharpness of the constant @xmath373 assume that ( 2.12.10 ) holds with @xmath232 i.e. , @xmath1176provided the condition ( i ) holds .
choose @xmath982 @xmath1159 @xmath727 @xmath1177 @xmath1178 @xmath1179 in ( [ 2.13.10 ] )
to get@xmath1180provided@xmath1181if @xmath764 @xmath1182 @xmath1183 then we have@xmath1184and by ( [ 2.14.10 ] ) we get@xmath1185now , if we let @xmath1186 @xmath1187 @xmath1188 then obviously@xmath1189which shows that ( [ 2.14.1.10 ] ) is fulfilled , and thus by ( [ 2.15.10 ] ) we obtain@xmath1190for any @xmath1191 this implies@xmath1192for any @xmath1191 finally , let @xmath1193 @xmath1194 , @xmath1195 .
then from ( [ 2.16.10 ] ) we get @xmath1196 for any @xmath380 which produces @xmath992 [ r2.4.10]if @xmath917 is an orthornormal family in the real inner product space @xmath1197 and @xmath1048 , @xmath966 ( @xmath949 is a finite part of @xmath950 ) and @xmath5 are such that @xmath1198 for @xmath1199 with @xmath1200 and@xmath1201then we have the inequality@xmath1202 ^{2}\leq \frac{1}{4}\cdot \frac{\sum_{i\in f}\left ( m_{i}-m_{i}\right ) ^{2}}{\sum_{i\in f}m_{i}m_{i}}\cdot \sum_{i\in f}% \left [ \left\langle x , e_{i}\right\rangle \right ] ^{2}.\]]the constant @xmath59 is best possible . the following reverse of the schwarz s inequality in inner product spaces holds .
[ c2.5.10]let @xmath24 and @xmath1203 @xmath71 with the property that @xmath1204 if either@xmath1205or equivalently,@xmath1206holds , then we have the inequalities@xmath1207 } { \sqrt{\delta \overline{\delta } } } \label{2.20.10 } \\ & \leq \frac{1}{2}\cdot \frac{\left\vert \delta \right\vert + \left\vert \delta \right\vert } { \sqrt{\delta \overline{\delta } } } \left\vert \left\langle x , y\right\rangle \right\vert , \notag\end{aligned}\]]@xmath1208@xmath1209and @xmath1210the constants @xmath222 and @xmath59 are best possible .
the inequality ( [ 2.20.10 ] ) follows from ( [ 2.6.10 ] ) on choosing @xmath1211 @xmath1212
@xmath1213 @xmath1214 @xmath1215 the inequality ( 2.21.10 ) is equivalent with ( [ 2.20.10 ] ) .
the inequality ( [ 2.22.10 ] ) follows from ( [ 2.1.10 ] ) for @xmath1216 and the same choices as above .
finally , ( [ 2.23.10 ] ) is obviously equivalent with ( 2.22.10 ) .
the following result holds @xcite .
[ t3.1.10]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath950 , @xmath981 @xmath993 and @xmath12 if either@xmath994or equivalently,@xmath995hold , then we have the inequality@xmath1217where @xmath1218 is defined in corollary [ c2.3.10 ]
. the constant @xmath59 is best possible . by the reverse of bessel s inequality in corollary [ c2.3.10 ] , we have@xmath1219taking the square root in ( [ 3.5.10 ] ) , we deduce ( [ 3.3.10 ] ) .
the fact that @xmath59 is the best possible constant follows by corollary [ c2.3.10 ] and we omit the details .
the following corollary for real inner product spaces holds @xcite .
[ c3.2.10]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath950 , @xmath1220 @xmath1221 and @xmath24 such that @xmath1222 @xmath1223 and@xmath1224then we have the inequality@xmath1225the constant @xmath790 is best possible . in the case where the family @xmath917 reduces to a single vector
, we may deduce from theorem [ t3.1.10 ] the following particular case first obtained in @xcite .
[ c3.3.10]let @xmath930 @xmath727 @xmath1124 with @xmath1226 @xmath375 and @xmath1227 such that either @xmath1228or equivalently,@xmath1229hold , then@xmath1230where@xmath1231 } { \func{re}% \left ( \phi \overline{\phi } \right ) } \right ] ^{\frac{1}{2}}.\]]the constant @xmath59 is best possible .
[ r3.4.10]if @xmath1 is real , @xmath930 @xmath593 and @xmath754 are such that @xmath755 @xmath756 and@xmath1232then@xmath1233the constant @xmath59 is best possible .
if @xmath1234 @xmath1235 then the following equivalent form of ( [ 3.12.10 ] ) also holds@xmath783 the following companion of the grss inequality also holds @xcite .
[ t4.1.10]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath950 , @xmath1236 , @xmath24 and @xmath1237 such that either@xmath1238or equivalently,@xmath1129holds .
then we have the inequality@xmath1239 \label{4.3.10 } \\ \leq \frac{1}{16}\cdot \frac{1}{\lambda \left ( 1-\lambda \right ) } \sum_{i\in f}m^{2}\left ( \mathbf{\phi } , \mathbf{\phi } , f\right ) \sum_{i\in f}\left\vert \left\langle \lambda x+\left ( 1-\lambda \right ) y , e_{i}\right\rangle \right\vert ^{2}.\end{gathered}\]]the constant @xmath790 is the best possible constant in ( [ 4.3.10 ] ) in the sense that it can not be replaced by a smaller constant . using
the known inequality@xmath1240we may state that for any @xmath793 and @xmath787 @xmath1241since@xmath1007for any @xmath354 then , by ( [ 4.4.10 ] ) , we get@xmath1008 \label{4.5.10 } \\ & = \func{re}\left [ \left\langle x-\sum_{i\in f}\left\langle x , e_{i}\right\rangle e_{i},y-\sum_{i\in f}\left\langle y , e_{i}\right\rangle e_{i}\right\rangle \right ] \notag \\ & \leq \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left\vert \lambda \left ( x-\sum_{i\in f}\left\langle x , e_{i}\right\rangle e_{i}\right ) + \left ( 1-\lambda \right ) \left ( y-\sum_{i\in f}\left\langle y , e_{i}\right\rangle e_{i}\right ) \right\vert ^{2 } \notag \\ & = \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left\vert \lambda x+\left ( 1-\lambda \right ) y-\sum_{i\in f}\left\langle \lambda x+\left ( 1-\lambda \right ) y , e_{i}\right\rangle e_{i}\right\vert ^{2 } \notag \\ & = \frac{1}{4\lambda \left ( 1-\lambda \right ) } \left [ \left\vert \lambda x+\left ( 1-\lambda \right ) y\right\vert ^{2}-\sum_{i\in f}\left\vert \left\langle \lambda x+\left ( 1-\lambda \right ) y , e_{i}\right\rangle \right\vert ^{2}\right ] .
\notag\end{aligned}\]]if we apply the reverse of bessel s inequality from corollary [ c2.3.10 ] for @xmath1242 we may state that@xmath1243now , by making use of ( [ 4.5.10 ] ) and ( [ 4.6.10 ] ) , we deduce ( 4.3.10 ) . the fact that @xmath790 is the best possible constant in ( [ 4.3.10 ] ) follows by the fact that if in ( [ 4.1.10 ] ) we choose @xmath657 then it becomes ( i ) of theorem [ t2.1.10 ] , implying for @xmath823 the inequality ( [ 2.12.10 ] ) , for which , we have shown that @xmath59 is the best constant .
[ r4.2.10]if in theorem [ t4.1.10 ] , we choose @xmath809 then we get@xmath1005 \leq \frac{1}{% 4}m^{2}\left ( \mathbf{\phi } , \mathbf{\phi } , f\right ) \sum_{i\in
f}\left\vert \left\langle \frac{x+y}{2},e_{i}\right\rangle \right\vert ^{2},\]]provided@xmath1134or equivalently,@xmath1135 the following proposition holds @xcite .
[ p5.1.10]let @xmath1022 be an orthornormal family of functions in @xmath1032 @xmath949 a finite subset of @xmath1033 @xmath1034 @xmath1035 such that @xmath1244 and @xmath1036 so that either@xmath1141 d\mu \left ( s\right ) \geq 0,\]]or equivalently,@xmath1038then we have the inequality@xmath1245 ^{\frac{1}{2}}}\]]@xmath1246 ^{\frac{1}{p}}\left ( \dsum\limits_{i\in f}\left\vert \dint_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{f_{i}}\left ( s\right ) d\mu \left ( s\right ) \right\vert ^{q}\right ) ^{\frac{1}{q}},\text { } \\ \hfill \text{\ for \ } p>1,\ \frac{1}{p}+\frac{1}{q}=1 \\ \\ \max\limits_{i\in f}\left\vert \dint_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{f_{i}}\left ( s\right ) d\mu \left ( s\right ) \right\vert \dsum\limits_{i\in f}\left [ \left\vert \phi _ { i}\right\vert + \left\vert \phi _ { i}\right\vert \right ] .%
\end{array}% \right.\]]in particular , we have@xmath1247the constant @xmath59 is best possible .
the proof is obvious by theorem [ t2.1.10 ] and remark [ r2.2.10 ] .
we omit the details .
the following proposition also holds .
[ p5.2.10]assume that @xmath1248 and @xmath949 satisfy the assumptions of proposition [ p5.1.10 ] .
then we have the following reverse of bessel s inequality:@xmath1249where , as above,@xmath1250 \right\ } } { \func{re}\left ( \phi _ { i}\overline{\phi _ { i}}% \right ) } \right ] ^{\frac{1}{2}}.
\label{5.8.10}\]]the constant @xmath59 is the best possible .
the following grss type inequality also holds .
[ p5.3.10]let @xmath1022 and @xmath949 be as in proposition [ p5.1.10 ] . if @xmath1251 @xmath1042 and @xmath1252 such that either@xmath1043 d\mu \left ( s\right ) & \geq 0 , \\ \int_{\omega } \rho \left ( s\right ) \func{re}\left [ \left ( \sum_{i\in f}\gamma _ { i}f_{i}\left ( s\right ) -g\left ( s\right )
\right ) \left ( \overline{% g}\left ( s\right ) -\sum_{i\in f}\overline{\gamma _ { i}}\text { } \overline{f_{i}% } \left ( s\right ) \right ) \right ] d\mu \left ( s\right ) & \geq 0,\end{aligned}\]]or equivalently,@xmath1253hold , then we have the inequality@xmath1254where @xmath1218 is as defined in ( [ 5.8.10 ] ) .
the constant @xmath59 is the best possible .
the proof follows by theorem [ t3.1.10 ] and we omit the details . in the case of real spaces ,
the following corollaries provide much simpler sufficient conditions for the reverse of bessel s inequality ( [ 5.7.10 ] ) or for the grss type inequality ( [ 5.11.10 ] ) to hold .
[ c5.4.10]let @xmath1022 be an orthornormal family of functions in the real hilbert space @xmath1047 @xmath949 a finite part of @xmath1033 @xmath1198 @xmath1255 with @xmath1256 and @xmath1257 so that@xmath1051then we have the inequality@xmath1052 ^{2 } \\ &
\leq \frac{1}{4}\cdot \frac{\sum_{i\in f}\left ( m_{i}-m_{i}\right ) ^{2}}{% \sum_{i\in f}m_{i}m_{i}}\cdot \sum_{i\in f}\left [ \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) f_{i}\left ( s\right ) d\mu \left ( s\right ) \right ] ^{2}.\end{aligned}\]]the constant @xmath59 is best possible .
[ c5.5.10]let @xmath1022 and @xmath949 be as above . if @xmath1258 @xmath1042 with @xmath1259 and @xmath1260 such that@xmath1055and@xmath1145then we have the inequality@xmath1261 ^{\frac{1}{2}}.\end{aligned}\ ] ] s.s .
dragomir , a counterpart of bessel s inequality in inner product spaces and some grss type related results , _ rgmia res . rep .
_ , * 6*(2003 ) , supplement , article 10 . `
[ on line : http://rgmia.vu.edu.au/v6(e).html ] .
dragomir , on bessel and grss inequalities for orthornormal families in inner product spaces , _ rgmia res .
_ , * 6*(2003 ) , supplement , ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , some new results related to bessel and grss inequalities for orthornormal families in inner product spaces , _ rgmia res . rep .
_ * 6*(2003 ) , _ supplement _ , article 13 , ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . ` s.s .
dragomir , some grss type inequalities in inner product spaces , _ j. inequal .
pure & appl .
_ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] . let @xmath312 be a real or complex infinite dimensional hilbert space and @xmath1262 an orthornormal family in @xmath1 , i.e. , we recall that @xmath1263 if @xmath1264 , @xmath1027 and @xmath1265 for @xmath186 .
it is well known that , if @xmath920 then the sum @xmath1266 is convergent and the following inequality , called _ _ bessel s inequality__@xmath1267holds . if @xmath1268 where @xmath1269 or @xmath190 , is the hilbert space of all complex or real sequences that are @xmath323-summable and @xmath1270 then the sum @xmath1271 is convergent in @xmath1 and if @xmath1272 then @xmath1273 we may state the following result @xcite .
[ t5.1.11a]let @xmath1274 be an infinite dimensional hilbert space over the real or complex number field @xmath2 , @xmath1275 an orthornormal family in @xmath923 @xmath1276 and @xmath315 with the property that@xmath1277if @xmath5 is such that@xmath1278then we have the inequality@xmath1279 \right ) ^{2}}{\sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}-r^{2 } } \label{5.4.11a } \\
& \leq \frac{\left\vert \sum_{i=1}^{\infty } \overline{\lambda _ { i}}% \left\langle x , e_{i}\right\rangle \right\vert ^{2}}{\sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}-r^{2 } } \notag \\ & \leq \frac{\sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}}{% \sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}-r^{2}}% \sum_{i=1}^{\infty } \left\vert \left\langle x , e_{i}\right\rangle \right\vert ^{2 } \notag\end{aligned}\]]and@xmath1280 applying the third inequality in ( [ 2.2.3 ] ) for @xmath1281 we have@xmath1282 ^{2}\leq r^{2}\left\vert x\right\vert ^{2 } \label{5.6.11a}\]]and since@xmath1283 , \end{aligned}\]]then , by ( [ 5.6.11a ] ) , we deduce@xmath1284 ^{2}\leq r^{2}\left\vert x\right\vert ^{2},\]]giving the first inequality in ( [ 5.4.11a ] ) . the second inequality is obvious by the modulus property .
the last inequality follows by the cauchy - bunyakovsky - schwarz inequality@xmath1285the inequality ( [ 5.5.11a ] ) follows by the last inequality in ( 5.4.11a ) on subtracting from both sides the quantity @xmath1286 the following result provides a generalization for the reverse of bessel s inequality obtained in @xcite .
[ t5.2.11a]let @xmath1274 and @xmath1275 be as in theorem t5.1.11a .
suppose that @xmath1287 @xmath1288 are sequences of real or complex numbers such that@xmath1289if @xmath5 is such that either@xmath1290or equivalently,@xmath1291holds , then we have the inequalities@xmath1292 \right ) ^{2}}{% \sum_{i=1}^{\infty } \func{re}\left ( \gamma _ { i}\overline{\gamma _ { i}}\right ) } \label{5.10.11a } \\ & \leq \frac{1}{4}\cdot \frac{\left\vert \sum_{i=1}^{\infty } \left ( \overline{\gamma _ { i}}+\overline{\gamma _ { i}}\right ) \left\langle x , e_{i}\right\rangle \right\vert ^{2}}{\sum_{i=1}^{\infty } \func{re}\left ( \gamma _ { i}\overline{\gamma _ { i}}\right ) } \notag \\ & \leq \frac{1}{4}\cdot \frac{\sum_{i=1}^{\infty } \left\vert \gamma _ { i}+\gamma _ { i}\right\vert ^{2}}{\sum_{i=1}^{\infty } \func{re}\left ( \gamma _ { i}\overline{\gamma _ { i}}\right ) } \sum_{i=1}^{\infty } \left\vert \left\langle x , e_{i}\right\rangle \right\vert ^{2}.
\notag\end{aligned}\]]the constant @xmath59 is best possible in all inequalities in ( 5.10.11a ) .
we also have the inequalities:@xmath1293here the constant @xmath59 is also best possible . since @xmath1294 , @xmath1295
then also @xmath1296 showing that the series@xmath1297are convergent . also , the series @xmath1298are convergent in the hilbert space @xmath26 now , we observe that the inequalities ( [ 5.10.11a ] ) and ( [ 5.11.11a ] ) follow from theorem [ t5.1.11a ] on choosing @xmath1299 @xmath186 and @xmath1300 the fact that @xmath59 is the best constant in both ( [ 5.10.11a ] ) and ( [ 5.11.11a ] ) follows from theorem [ t2.2.3 ] and corollary c2.3.3 , and we omit the details . for some recent results related to the bessel inequality , see @xcite , @xcite , @xcite , and @xcite . the following result related to the grss inequality in inner product spaces holds @xcite .
[ t6.1.11a]let @xmath1274 be an infinite dimensional hilbert space over the real or complex number field @xmath2 , and @xmath1275 an orthornormal family in @xmath26 assume that @xmath1301 and @xmath876 with the properties that@xmath1302if @xmath24 are such that@xmath1303then we have the inequalities@xmath1304 applying schwarz s inequality for the vectors @xmath1305 @xmath1306 we have@xmath1307since@xmath1308and@xmath1309then by ( [ 5.5.11a ] ) applied for @xmath14 and @xmath1015 and from ( [ 6.4.11a ] ) , we deduce the first part of ( [ 6.3.11a ] ) .
the second part follows by bessel s inequality .
the following grss type inequality may be stated as well .
[ t6.2.11a]let @xmath1274 be an infinite dimensional hilbert space and @xmath1310 an orthornormal family in @xmath26 suppose that @xmath1311 @xmath1312 @xmath1313 @xmath1314 are sequences of real and complex numbers such that@xmath1315if @xmath24 are such that either@xmath1316or equivalently,@xmath1317holds , then we have the inequality@xmath1318 ^{\frac{1}{2}}\left [ \sum_{i=1}^{\infty } \func{re}\left ( \phi _ { i}\overline{\phi _ { i}}\right ) % \right ] ^{\frac{1}{2}}}\left\vert x\right\vert \left\vert y\right\vert\end{aligned}\]]the constant @xmath59 is best possible in the first inequality
. follows by ( [ 5.11.11a ] ) and ( [ 6.4.11a ] ) .
the best constant follows from theorem [ t4.2.7a ] , and we omit the details .
we may state the following result @xcite .
[ t5.1.11b]let @xmath1274 be an infinite dimensional hilbert space over the real or complex number field @xmath2 , @xmath1275 is an orthornormal family in @xmath923 @xmath1319 @xmath1320 and @xmath435 if @xmath5 is such that@xmath1278then we have the inequality@xmath1321the constant @xmath222 is best possible in ( [ 5.3.11b ] ) in the sense that it can not be replaced by a smaller constant .
let @xmath1322 then by theorem t2.1.4 , we have @xmath1323giving@xmath1324since@xmath1325using the cauchy - bunyakovsky - schwarz inequality , we may state that@xmath1326and thus , by ( [ 5.4.11b ] ) and ( [ 5.5.11b ] ) , we may state that@xmath1327from where we get the desired inequality in ( [ 5.3.11b ] ) .
the best constant , follows by theorem [ t2.1.4 ] on choosing @xmath1328 with @xmath1329 and we omit the details . under the assumptions of theorem [ t5.1.11b ] , and if we multiply by @xmath1330 then we deduce , from ( [ 5.3.11b ] ) , that@xmath1331where for the last inequality , we have used bessel s inequality@xmath1332 the following result also holds @xcite .
[ t5.2.11b]assume that @xmath1197 and @xmath1275 are as in theorem [ t5.1.11b ] . if @xmath1333 @xmath1334 with @xmath1335 , and @xmath5 with the property that , either@xmath1336or equivalently,@xmath1337holds , then we have the inequality@xmath1338the constant @xmath59 is best possible in the sense that it can not be replaced by a smaller constant .
since @xmath1339 @xmath1295 then we have that @xmath1340 showing that the series@xmath1341are convergent .
in addition , the series @xmath1342 , @xmath1343 and @xmath1344 are also convergent in the hilbert space @xmath26 now , we observe that the inequality ( [ 5.9.11b ] ) follows from theorem t5.1.11b on choosing @xmath1345 @xmath1346 and @xmath1347 the fact that @xmath59 is the best possible constant in ( [ 5.9.11b ] ) follows from theorem [ t2.2.4 ] , and we omit the details . with the assumptions of theorem [ t5.2.11b ]
, we have @xmath1348 \notag \\ & \leq \frac{1}{2}\cdot \frac{\sum_{i=1}^{\infty } \left\vert \gamma _ { i}-\gamma _ { i}\right\vert ^{2}}{\left ( \sum_{i=1}^{\infty } \left\vert \gamma _ { i}+\gamma _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}}\left\vert x\right\vert .
\notag\end{aligned}\ ] ] the following result holds @xcite .
[ t6.1.11b]let @xmath1274 be an infinite dimensional hilbert space over the real or complex number field @xmath2 and @xmath1275 an orthornormal family in @xmath26 if @xmath1349 @xmath1350 , @xmath1351 @xmath876 and @xmath24 are such that@xmath1352then we have the inequality@xmath1353 ^{\frac{1}{2}}\left [ \left\vert y\right\vert + \left ( \sum_{i=1}^{\infty } \left\vert \left\langle y , e_{i}\right\rangle \right\vert ^{2}\right ) ^{\frac{1}{2}}\right ] ^{\frac{1% } { 2}}}{\left ( \sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}\right ) ^{\frac{1}{4}}\left ( \sum_{i=1}^{\infty } \left\vert \mu _ { i}\right\vert ^{2}\right ) ^{\frac{1}{4 } } } \\ & \leq r_{1}r_{2}\frac{\left\vert x\right\vert ^{\frac{1}{2}}\left\vert y\right\vert ^{\frac{1}{2}}}{\left ( \sum_{i=1}^{\infty } \left\vert \lambda _ { i}\right\vert ^{2}\right ) ^{\frac{1}{4}}\left ( \sum_{i=1}^{\infty } \left\vert \mu _ { i}\right\vert ^{2}\right ) ^{\frac{1}{4}}}.\end{aligned}\ ] ] it follows by ( [ 5.6.11b ] ) applied for @xmath14 and @xmath1354 we omit the details . finally we may state the following theorem @xcite .
[ t6.2.11b]assume that @xmath1197 and @xmath1275 are as in theorem [ t6.1.11b ] .
if @xmath1333 @xmath1355 @xmath1356 @xmath1357 with @xmath1358 , @xmath1359 , and @xmath24 are such that , either@xmath1316or equivalently,@xmath1317holds , then we have the inequality@xmath1360 ^{\frac{1}{2}}\left [ \left\vert y\right\vert + \left ( \sum_{i=1}^{\infty } \left\vert \left\langle y , e_{i}\right\rangle \right\vert ^{2}\right ) ^{\frac{1}{2}}\right ] ^{\frac{1% } { 2}}}{\left ( \sum_{i=1}^{\infty } \left\vert \phi _ { i}+\phi _ { i}\right\vert ^{2}\right ) ^{\frac{1}{4}}\left ( \sum_{i=1}^{\infty } \left\vert \gamma _ { i}+\gamma _ { i}\right\vert ^{2}\right ) ^{\frac{1}{4 } } } \\ & \leq \frac{1}{2}\cdot \frac{\left ( \sum_{i=1}^{\infty } \left\vert \phi _ { i}-\phi _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}\left ( \sum_{i=1}^{\infty } \left\vert \gamma _ { i}-\gamma _ { i}\right\vert ^{2}\right ) ^{\frac{1}{2}}}{\left ( \sum_{i=1}^{\infty } \left\vert \phi _ { i}+\phi _ { i}\right\vert ^{2}\right ) ^{\frac{1}{4}}\left ( \sum_{i=1}^{\infty } \left\vert \gamma _ { i}+\gamma _ { i}\right\vert ^{2}\right ) ^{\frac{1}{4}}}\left\vert x\right\vert ^{\frac{1}{2}}\left\vert y\right\vert ^{\frac{1}{2}}.\end{aligned}\ ] ] the proof follow by ( [ 5.10.11b ] ) applied for @xmath14 and @xmath1354 we omit the details . x.h .
cao , bessel sequences in a hilbert space .
_ gongcheng shuxue xuebao _ 1**7 * * ( 2000 ) , no .
2 , 9298 . s.s .
dragomir , a note on bessel s inequality , _ austral .
_ * 28 * ( 2001 ) , no . 5 , 246248 .
dragomir , on bessel and grss inequalities for orthornormal families in inner product spaces , _ rgmia res .
, _ * 6*(2003 ) , _ supplement _ , article 12 , ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , a counterpart of bessel s inequality in inner product spaces and some grss type related results , _ rgmia res .
_ * 6*(2003 ) , _ supplement _ , article 10 , ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir , some new results related to bessel and grss inequalities for orthornormal families in inner product spaces , _ rgmia res . rep .
_ * 6*(2003 ) , _ supplement _ , article 13 , ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . ` s.s .
dragomir , reverses of schwarz , triangle and bessel inequalities in inner product spaces , _ preprint _ ` [ on line : http://www.mathpreprints.com/math/ ` ` preprint / sever/20030828.2/1/ ] . `
dragomir , new reverses of schwarz , triangle and bessel inequalities in inner product spaces , _ rgmia res .
, _ * 6*(2003 ) , _ supplement _ , article 20 , ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . `
dragomir and j. sndor , on bessel s and gram s inequalities in pre - hilbertian spaces .
hungar . , _ * 29 * ( 1994 ) , no .
3 , 197205 .
h. gunawan , a generalization of bessel s inequality and parseval s identity , _ period .
hungar . , _ * 44 * ( 2002 ) , no .
2 , 177181 .
n. ujevi , a generalisation of grss inequality in prehilbertian spaces , _ math .
& appl . , _ ( to appear ) .
let @xmath1361 be an inner product space over the real or complex number field @xmath2 .
if @xmath1362 are orthonormal vectors in the inner product space @xmath923 i.e. , @xmath1363 for all @xmath1364 where @xmath1025 is the kronecker delta , then we have the following inequality is well known in the literature as bessel s inequality ( see for example @xcite ) : @xmath1365 for other results related to bessel s inequality , see @xcite 5b.12 and chapter xv in the book @xcite . in 1941 , r.p .
boas @xcite and in 1944 , independently , r. bellman 1b.12 proved the following generalisation of bessel s inequality ( see also @xcite ) .
[ t1.1.12]if @xmath1366 are elements of an inner product space @xmath1367 then the following inequality : @xmath1368 , \label{1.2.12}\]]holds .
a recent generalisation of the boas - bellman result was given in mitrinovi - peari - fink @xcite where they proved the following : [ t1.2.12]if @xmath1366 are as in theorem [ t1.1.12 ] and @xmath1369 , then one has the inequality : @xmath1370 .\end{gathered}\ ] ] they also noted that if in ( [ 1.3.12 ] ) one chooses @xmath1371 , then this inequality becomes ( [ 1.2.12 ] ) .
for other results related to the boas - bellman inequality , see @xcite . in this section , by following @xcite , we point out some new results that may be related to both the mitrinovi - peari - fink and boas - bellman inequalities .
we start with the following lemma which is also interesting in itself 13nssd .
[ l2.1.12]let @xmath1372 and @xmath1373 .
then one has the inequality : @xmath1374 ^{\frac{1}{\gamma } } \left ( \sum\limits_{1\leq i\neq j\leq n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert ^{\delta } \right ) ^{\frac{1}{\delta } } , \\
\hfill \ \ \ \text{where \ } \gamma > 1,\ \
\frac{1}{\gamma } + \frac{1}{\delta } % = 1 ; \\ \\
\left [ \left ( \sum\limits_{i=1}^{n}\left\vert \alpha _ { i}\right\vert \right ) ^{2}-\sum\limits_{i=1}^{n}\left\vert \alpha
_ { i}\right\vert ^{2}\right ] \max\limits_{1\leq
i\neq j\leq n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert .% \end{array}% \right.\end{gathered}\ ] ] we observe that @xmath1375using hlder s inequality , we may write that @xmath1376by hlder s inequality for double sums we also have @xmath1377@xmath1378 ^{\frac{1}{\gamma } } \left ( \sum\limits_{1\leq i\neq j\leq n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert ^{\delta } \right ) ^{\frac{1}{\delta } } , \\
\hfill \ \ \ \text{where \ } \gamma > 1,\ \
\frac{1}{\gamma } + \frac{1}{\delta } % = 1 ; \\ \\
\left [ \left ( \sum\limits_{i=1}^{n}\left\vert \alpha _ { i}\right\vert \right ) ^{2}-\sum\limits_{i=1}^{n}\left\vert \alpha
_ { i}\right\vert ^{2}\right ] \max\limits_{1\leq i\neq j\leq n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert .%
\end{array}% \right.\]]utilising ( [ 2.3.12 ] ) and ( [ 2.4.12 ] ) in ( [ 2.2.12 ] ) , we may deduce the desired result ( [ 2.1.12 ] ) .
[ r2.2.12]inequality ( [ 2.1.12 ] ) contains in fact 9 different inequalities which may be obtained by combining the first 3 ones with the last 3 ones .
a particular case that may be related to the boas - bellman result is embodied in the following inequality @xcite .
[ c2.3.12]with the assumptions in lemma [ l2.1.12 ] , we have @xmath1379 @xmath1380 ^{\frac{1}{2}}}{% \sum_{i=1}^{n}\left| \alpha _ { i}\right| ^{2}}\left ( \sum\limits_{1\leq i\neq j\leq n}\left| \left ( z_{i},z_{j}\right ) \right| ^{2}\right ) ^{\frac{1}{2}% } \right\ } \notag \\ & \leq \sum_{i=1}^{n}\left| \alpha _ { i}\right| ^{2}\left\ { \max\limits_{1\leq i\leq n}\left\| z_{i}\right\| ^{2}+\left ( \sum\limits_{1\leq i\neq j\leq n}\left| \left ( z_{i},z_{j}\right ) \right| ^{2}\right ) ^{\frac{1}{2}}\right\ } . \notag\end{aligned}\ ] ] the first inequality follows by taking the third branch in the first curly bracket with the second branch in the second curly bracket for @xmath1381 the second inequality in ( [ 2.5.12 ] ) follows by the fact that @xmath1382 ^{\frac{1}{2% } } \leq \sum_{i=1}^{n}\left\vert \alpha _ { i}\right\vert ^{2}.\]]applying the following cauchy - bunyakovsky - schwarz type inequality @xmath1383we may write that @xmath1384and @xmath1385also , it is obvious that : @xmath1386consequently , we may state the following coarser upper bounds for @xmath1387 that may be useful in applications @xcite .
[ c2.4.12]with the assumptions in lemma [ l2.1.12 ] , we have the inequalities : @xmath1388 the proof is obvious by lemma [ l2.1.12 ] in applying the inequalities ( [ 2.7.12 ] ) ( [ 2.9.12 ] ) .
[ r2.5.12]the following inequalities which are incorporated in ( 2.10.12 ) are of special interest : @xmath1389 ; \label{2.11.12}\]]@xmath1390 , \end{gathered}\]]where @xmath1391 @xmath1392 and @xmath1393 . \label{2.13.12}\ ] ] we are now able to present the following result obtained in @xcite , which complements the inequality ( [ 1.3.12 ] ) due to mitrinovi , peari and fink @xcite .
[ t3.1.12]let @xmath1366 be vectors of an inner product space @xmath1361 and @xmath1394 @xmath1395 then one has the inequalities : @xmath1396@xmath1397 ^{\frac{1}{\gamma } } \\
\qquad \times \left ( \sum_{1\leq i\neq j\leq n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert ^{\delta } \right ) ^{\frac{1}{\delta } } , \\ \hfill \
\ \text{where \ } \gamma > 1,\ \
\frac{1}{\gamma } + \frac{1}{\delta } % = 1 ; \\ \\
\left [ \left ( \sum_{i=1}^{n}\left\vert c_{i}\right\vert \right ) ^{2}-\sum_{i=1}^{n}\left\vert c_{i}\right\vert ^{2}\right ] \max\limits_{1\leq i\neq j\leq n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert .% \end{array}% \right .
\label{3.1.12}\ ] ] we note that @xmath1398using schwarz s inequality in inner product spaces , we have : @xmath1399now using lemma [ l2.1.12 ] with @xmath1400 @xmath1401 @xmath1402 we deduce the desired inequality ( [ 3.1.12 ] ) . the following particular inequalities that may be obtained by the corollaries [ c2.3.12 ] and [ c2.4.12 ] and remark [ r2.5.12 ] hold @xcite .
[ c3.2.12]with the assumptions in theorem [ t3.1.12 ] , one has the inequalities : @xmath1403 [ r3.3.12]note that the first inequality in ( [ 3.2.12 ] ) is the result obtained by mitrinovi - peari - fink in @xcite . the other 3 provide similar bounds in terms of the @xmath548norms of the vector @xmath1404 if one chooses @xmath1405 @xmath1406 in ( [ 3.1.12 ] ) , then it is possible to obtain 9 different inequalities between the fourier coefficients @xmath1407 and the norms and inner products of the vectors @xmath1408 @xmath1409 we restrict ourselves only to those inequalities that may be obtained from ( [ 3.2.12 ] ) . as mitrinovi , peari and fink noted in @xcite , the first inequality in ( [ 3.2.12 ] ) for the above selection of @xmath1410 will produce the boas - bellman inequality ( [ 1.2.12 ] ) . from the second inequality in ( [ 3.2.12 ] ) for @xmath1411 we get @xmath1412taking the square root in this inequality we obtain : @xmath1413for any @xmath1366 vectors in the inner product space @xmath1414 if we assume that @xmath1415 is an orthonormal family in @xmath923 then by ( [ 4.1.12 ] ) we have @xmath1416from the third inequality in ( [ 3.2.12 ] ) for @xmath1411 we deduce @xmath1417for @xmath1391 @xmath1418 taking the square root in this inequality we get @xmath1419for any @xmath1420 @xmath1391 @xmath1418 the above inequality ( [ 4.3.12 ] ) becomes , for an orthornormal family @xmath1421@xmath1422finally , the choice @xmath1405 @xmath1423 will produce in the last inequality in ( [ 3.2.12 ] ) @xmath1424giving the following boas - bellman type inequality @xmath1425for any @xmath1426 it is obvious that ( [ 4.5.12 ] ) will give for orthonormal families the well known bessel inequality . in order the compare the boas - bellman result with our result ( [ 4.5.12 ] ) , it is enough to compare the quantities @xmath1427and @xmath1428consider the inner product space @xmath602 with @xmath1429 and choose @xmath1430 @xmath1431 , @xmath1432 @xmath1433 then @xmath1434denote @xmath1435
@xmath1436 @xmath1437 then @xmath1438firstly , if we assume that @xmath1439 then @xmath1440 @xmath1441 which shows that @xmath1442 now choose @xmath1443 and @xmath1444 then @xmath1445 and @xmath1446 showing that @xmath1447 consequently , in general , the boas - bellman inequality and our inequality ( [ 4.5.12 ] ) can not be compared .
r. bellman , almost orthogonal series , _ bull .
soc . , _ * 50 * ( 1944 ) , 517519 . r.p .
boas , a general moment problem , _ amer .
, _ * 63 * ( 1941 ) , 361370 . s.s .
dragomir , on the boas - bellman inequality in inner product spaces , _ rgmia res .
_ , * 6*(2003 ) , supplement , article 14 ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] ` .
dragomir and j. sndor , on bessel s and gram s inequality in prehilbertian spaces , _ periodica math . hung .
, _ * 29*(3 ) ( 1994 ) , 197205 .
dragomir and b. mond , on the boas - bellman generalisation of bessel s inequality in inner product spaces , _ italian j. of pure & appl . math .
, _ * 3 * ( 1998 ) , 2935 . s.s .
dragomir , b. mond and j.e .
peari , some remarks on bessel s inequality in inner product spaces , _ studia univ .
babe - bolyai , mathematica , _ * 37*(4 ) ( 1992 ) , 7786 .
mitrinovi , j.e .
peari and a.m. fink , _ classical and new inequalities in analysis , _ kluwer academic publishers , 1993 . in 1971 , e. bombieri @xcite ( see also @xcite ) gave the following generalisation of bessel s inequality .
[ t1.1.13]if @xmath1366 are vectors in the inner product space @xmath1367 then the following inequality : @xmath1448holds .
it is obvious that if @xmath1449 are orthonormal , then from ( [ 1.2.13 ] ) one can deduce bessel s inequality .
another generalisation of bessel s inequality was obtained by a. selberg ( see for example @xcite ) : [ t1.2.13]let @xmath1366 be vectors in @xmath1 with @xmath1450 @xmath1451 then one has the inequality : @xmath1452 in this case , also , if @xmath1449 are orthonormal , then from ( [ 1.3.13 ] ) one may deduce bessel s inequality .
another type of inequality related to bessel s result , was discovered in 1958 by h. heilbronn @xcite ( see also @xcite ) .
[ t1.3.13]with the assumptions in theorem [ t1.1.13 ] , one has @xmath1453 if in ( [ 1.4.13 ] ) one chooses @xmath1454 @xmath1402 where @xmath1415 are orthonormal vectors in @xmath923 then @xmath1455 in 1992 , j.e .
peari @xcite ( see also @xcite ) proved the following general inequality in inner product spaces .
[ t1.4.13]let @xmath1456 and @xmath1457 .
then @xmath1458 he showed that the bombieri inequality ( [ 1.2.13 ] ) may be obtained from ( [ 1.6.13 ] ) for the choice @xmath1405 ( using the second inequality ) , the selberg inequality ( [ 1.3.13 ] ) may be obtained from the first part of ( [ 1.6.13 ] ) for the choice @xmath1459while the heilbronn inequality ( [ 1.4.13 ] ) may be obtained from the first part of ( [ 1.6.13 ] ) if one chooses @xmath1460 for any @xmath1461 for other results connected with the above ones , see @xcite and 5b.13 .
we start with the following lemma which is also interesting in itself 14nssd .
[ l2.1.13]let @xmath1372 and @xmath1462 then one has the inequality : @xmath1463where @xmath1464@xmath1465 ^{\frac{1}{u}},\ \ \
p>1,\ \frac{1}{p}+% \frac{1}{q}=1 ; \\ \hfill \hfill t>1,\ \frac{1}{t}+\frac{1}{u}=1 ; \\ \left ( \sum\limits_{k=1}^{n}\left\vert \alpha _ { k}\right\vert ^{p}\right ) ^{% \frac{1}{p}}\sum\limits_{i=1}^{n}\left\vert \alpha _ { i}\right\vert \max\limits_{1\leq i\leq n}\left\ { \left ( \sum\limits_{j=1}^{n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert ^{q}\right ) ^{\frac{1}{q}}\right\ } , \ \ \
p>1,\ \frac{1}{p}+\frac{1}{q}=1;% \end{array}% \right.\]]and@xmath1466 ; \\ \\ \sum\limits_{k=1}^{n}\left\vert \alpha _ { k}\right\vert \ \left ( \sum\limits_{i=1}^{n}\left\vert \alpha _ { i}\right\vert ^{m}\right ) ^{\frac{1% } { m}}\left ( \sum\limits_{i=1}^{n}\left [ \max\limits_{1\leq j\leq n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert \right ] ^{l}\right ) ^{% \frac{1}{l}},\ \ \ m>1,\ \frac{1}{m}+\frac{1}{l}=1 ; \\ \\ \left ( \sum\limits_{k=1}^{n}\left\vert \alpha _ { k}\right\vert \right ) ^{2}\max\limits_{i,1\leq j\leq n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert .% \end{array}% \right.\ ] ] we observe that @xmath1467using hlder s inequality , we may write that @xmath1468for any @xmath35 giving @xmath1469by hlder s inequality we also have : @xmath1470and thus @xmath1471and the first 3 inequalities in ( [ 2.1.13 ] ) are obtained . by hlder
s inequality we also have : @xmath1472and the next 3 inequalities in ( [ 2.1.13 ] ) are proved .
finally , by the same hlder inequality we may state that : @xmath1473and the last 3 inequalities in ( [ 2.1.13 ] ) are proved .
if we would like to have some bounds for @xmath1474 in terms of @xmath1475 then the following corollaries may be used .
[ c2.1.a.13]let @xmath1476 and @xmath1477 be as in lemma [ l2.1.13 ] .
if @xmath1478 , @xmath1479 then one has the inequality @xmath1480 ^{\frac{1}{u% } } , \label{2.5.a.13}\]]where @xmath1481 @xmath1482 observe , by the monotonicity of power means , we may write that @xmath1483from where we get @xmath1484using the fifth inequality in ( [ 2.1.13 ] ) , we then deduce ( [ 2.5.a.13 ] ) .
[ r2.1.b.13]an interesting particular case is the one for @xmath1485 giving @xmath1486 [ c2.1.c.13]with the assumptions of lemma [ l2.1.13 ] and if @xmath1487 then @xmath1488 , \label{2.5.c.13}\]]where @xmath1418 since @xmath1489and @xmath1490then by the sixth inequality in ( [ 2.1.13 ] ) we deduce ( [ 2.5.c.13 ] ) . in a similar fashion
, one may prove the following two corollaries .
[ c2.1.1.13]with the assumptions of lemma [ l2.1.13 ] and if @xmath1491 then @xmath1492 ^{l}\right ) ^{\frac{1}{l}},\]]where @xmath1493 [ c2.1.e.13]with the assumptions of lemma [ l2.1.13 ] , we have : @xmath1494 the following lemma may be of interest as well @xcite .
[ l2.2.13]with the assumptions of lemma [ l2.1.13 ] , one has the inequalities @xmath1495 ; \\ \\
\left ( \sum_{i=1}^{n}\left\vert \alpha _ { i}\right\vert ^{2p}\right ) ^{\frac{1% } { p}}\left ( \left ( \sum_{j=1}^{n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert \right ) ^{q}\right ) ^{\frac{1}{q}},\ \\
\hfill p>1,\ \frac{1}{p}+\frac{1}{q}=1 ; \\ \\
\max\limits_{1\leq i\leq n}\left\vert \alpha
_ { i}\right\vert ^{2}\ \sum_{i , j=1}^{n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert .%
\end{array}% \right .
\notag\end{aligned}\ ] ] as in lemma [ l2.1.13 ] , we know that @xmath1496using the simple observation that ( see also @xcite ) @xmath1497we have @xmath1498 \\ & = \sum\limits_{i , j=1}^{n}\left\vert \alpha _ { i}\right\vert ^{2}\left\vert \left ( z_{i},z_{j}\right ) \right\vert , \end{aligned}\]]which proves the first inequality in ( [ 2.6.13 ] ) .
the second part follows by hlder s inequality and we omit the details . the first part in ( [ 2.6.13 ] )
is the inequality obtained by peari in @xcite .
we are now able to present the following result obtainend in @xcite , which complements the inequality ( [ 1.6.13 ] ) due to j.e .
peari @xcite ( see also @xcite ) .
[ t3.1.13]let @xmath1366 be vectors of an inner product space @xmath1361 and @xmath1394 .
then one has the inequalities : @xmath1499where@xmath1500 ^{\frac{1}{s}},\ \ \
r>1,\ \frac{1}{r}+\frac{1}{s}=1 ; \\ \\
\max\limits_{1\leq k\leq n}\left\vert c_{k}\right\vert \sum\limits_{k=1}^{n}\left\vert c_{k}\right\vert \max\limits_{1\leq i\leq n}\left ( \sum\limits_{j=1}^{n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert \right ) ; % \end{array}% \right.\]]@xmath1501 ^{\frac{1}{u } } , & p>1,\ \frac{1}{p}+\frac{1% } { q}=1 ; \\ & t>1,\ \frac{1}{t}+\frac{1}{u}=1 ; \\ &
\\ \left ( \sum\limits_{k=1}^{n}\left\vert c_{k}\right\vert ^{p}\right ) ^{\frac{1% } { p}}\sum\limits_{i=1}^{n}\left\vert c_{i}\right\vert \max\limits_{1\leq i\leq n}\left\ { \left ( \sum\limits_{j=1}^{n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert ^{q}\right ) ^{\frac{1}{q}}\right\ } , & p>1,\ \frac{1}{p}+\frac{1}{q}=1;% \end{array}% \right.\]]and@xmath1502 ; \\ \\ \sum_{k=1}^{n}\left\vert c_{k}\right\vert \ \left ( \sum_{i=1}^{n}\left\vert c_{i}\right\vert ^{m}\right ) ^{\frac{1}{m}}\left ( \sum_{i=1}^{n}\left [ \max\limits_{1\leq j\leq n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert % \right ] ^{l}\right ) ^{\frac{1}{l } } ,
\\ \hfill \ \ \
m>1,\ \frac{1}{m}+\frac{1}{l}=1 ; \\ \\
\left ( \sum_{k=1}^{n}\left\vert c_{k}\right\vert \right ) ^{2}\max\limits_{i,1\leq j\leq n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert .%
\end{array}% \right.\ ] ] we note that @xmath1503using schwarz s inequality in inner product spaces , we have @xmath1504finally , using lemma [ l2.1.13 ] with @xmath1400 @xmath1401 @xmath1402 we deduce the desired inequality ( [ 3.1.13 ] ) .
we omit the details
. the following corollaries may be useful if one needs bounds in terms of @xmath1505 [ c3.2.13]with the assumptions in theorem [ t3.1.13 ] and if @xmath1487 @xmath1479 @xmath1481 @xmath1506 one has the inequality : @xmath1507 ^{\frac{1}{u% } } , \end{gathered}\]]and , in particular , for @xmath1485@xmath1508 the proof is similar to the one in corollary [ c2.1.a.13 ] .
[ c3.3.13]with the assumptions in theorem [ t3.1.13 ] and if @xmath1487 then @xmath1509 ^{\frac{1}{q}},\]]where @xmath1418 the proof is similar to the one in corollary [ c2.1.c.13 ]
. the following two inequalities also hold .
[ c3.4.13]with the above assumptions for @xmath1510 and if @xmath1511 then @xmath1512 ^{l}\right ) ^{\frac{1}{l } } , \label{3.6.13}\]]where @xmath1493 [ c3.5.13]with the above assumptions for @xmath1513 one has @xmath1514 using lemma [ l2.2.13 ] , we may state the following result as well .
[ t3.6.13]with the assumptions of theorem [ t3.1.13 ] , one has the inequalities : @xmath1515 ; \\ \\
\left ( \sum_{i=1}^{n}\left\vert c_{i}\right\vert ^{2p}\right ) ^{\frac{1}{p}% } \left ( \sum_{i=1}^{n}\left ( \sum_{j=1}^{n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert \right ) ^{q}\right ) ^{\frac{1}{q } } , \\ \hfill \
p>1,\ \frac{1}{p}+\frac{1}{q}=1 ; \\
\\ \max\limits_{1\leq i\leq n}\left\vert c_{i}\right\vert ^{2}\ \sum_{i , j=1}^{n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert ; % \end{array}% \right.\end{aligned}\]]that provide some alternatives to peari s result ( [ 1.6.13 ] ) .
in this section we point out some inequalities of bombieri type that may be obtained from ( [ 3.1.13 ] ) on choosing @xmath1411 @xmath1451 if the above choice was made in the first inequality in ( [ 3.1.13 ] ) , then one can obtain : @xmath1516giving , by taking the square root , @xmath1517if the same choice for @xmath1410 is made in the second inequality in ( 3.1.13 ) , then one can get @xmath1518 ^{\frac{1}{s}},\end{gathered}\]]implying @xmath1519 ^{\frac{1}{2s}},\end{gathered}\]]where @xmath1520
@xmath1521 the other inequalities in ( [ 3.1.13 ] ) will produce the following results , respectively @xmath1522 ; \end{gathered}\]]@xmath1523 ^{% \frac{1}{2}},\end{gathered}\]]where @xmath1391 @xmath1392@xmath1524 ^{% \frac{1}{2u}},\end{gathered}\]]where @xmath1391 @xmath1481 @xmath1525 @xmath1526@xmath1527where @xmath1391 @xmath1392@xmath1528 ^{\frac{1}{2}}\max\limits_{1\leq i\leq n}\left\vert \left ( x , y_{i}\right ) \right\vert ^{\frac{1}{2 } } \label{4.7.13 } \\
\times \left ( \sum\limits_{i=1}^{n}\left [ \max\limits_{1\leq j\leq n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert \right ] \right ) ^{\frac{1% } { 2}};\end{gathered}\]]@xmath1529 ^{\frac{1}{2m}}\left [ \sum\limits_{i=1}^{n}\left [ \max\limits_{1\leq j\leq n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert ^{l}\right ] \right ] ^{\frac{1}{2l } } , \label{4.8.13}\]]where @xmath1530 @xmath1531 and @xmath1532 if in the above inequalities we assume that @xmath1533 where @xmath1534 are orthonormal vectors in the inner product space @xmath1535 then from ( [ 4.1.13 ] ) ( 4.9.13 ) we may deduce the following inequalities similar in a sense to bessel s inequality : @xmath1536@xmath1537where @xmath1538 @xmath1539@xmath1540@xmath1541where @xmath1542@xmath1543where @xmath1391 @xmath1525 @xmath1526@xmath1544@xmath1545@xmath1546 ^{\frac{1}{m}},\ \ \
m>1,\ \frac{1}{m}+\frac{1}{l}=1;\]]@xmath1547 corollaries [ c3.2.13 ] [ c3.5.13 ] will produce the following results which do not contain the fourier coefficients in the right side of the inequality . indeed , if one chooses @xmath1405 in ( 3.3.13 ) , then @xmath1548 ^{\frac{1}{u% } } , \]]giving the following bombieri type inequality : @xmath1549 ^{\frac{1}{u% } } , \]]where @xmath1550 @xmath1479 @xmath1481 @xmath1551 if in this inequality we consider @xmath1485 then @xmath1552for a different proof of this result see also @xcite .
in a similar way , if @xmath1405 in ( 3.6.13 ) , then @xmath1553 ^{l}\right ) ^{\frac{1}{l}},\]]where @xmath1530 @xmath1493 finally , if @xmath1405 @xmath1554 is taken in ( [ 3.7.13 ] ) , then @xmath1555 [ r4.1.13]let us compare bombieri s result @xmath1556with our result @xmath1557denote @xmath1558and @xmath1559 ^{\frac{1}{2}}.\ ] ] if we choose the inner product space @xmath602 , @xmath1560 and @xmath1561 then for @xmath1562 @xmath1563 @xmath1564 we have @xmath1565 assume that @xmath1566 then @xmath1567 showing that , in this case , the bound provided by ( [ 4.24.13 ] ) is better than the bound provided by ( [ 4.23.13 ] ) .
if @xmath1449 are orthonormal vectors , then @xmath1568 @xmath1569 , showing that in this case the bombieri inequality ( which becomes bessel s inequality ) provides a better bound than ( [ 4.24.13 ] ) .
r. bellman , almost orthogonal series , _ bull .
_ * 50 * ( 1944 ) , 517519 .
boas , a general moment problem , _ amer .
, _ * 63 * ( 1941 ) , 361370 .
e. bombieri , a note on the large sieve , _ acta arith .
_ , * 18*(1971 ) , 401 - 404 .
dragomir , on the bombieri inequality in inner product spaces , _ rgmia res . rep .
_ , * 6*(2003 ) , no .
3 , article 5 ` [ on line : http://rgmia.vu.edu.au/v6n3.html ] ` s.s .
dragomir and j. sndor , on bessel s and gaur s inequality in prehilbertian spaces , _ periodica math . hung .
, _ * 29*(3 ) ( 1994 ) , 197205 . s.s .
dragomir and b. mond , on the boas - bellman generalisation of bessel s inequality in inner product spaces , _ italian j. of pure & appl .
, _ * 3 * ( 1998 ) , 2935 .
dragomir , b. mond and j.e .
peari , some remarks on bessel s inequality in inner product spaces , _ studia univ .
babe - bolyai , mathematica , _ * 37*(4 ) ( 1992 ) , 7786 .
h. heilbronn , on the averages of some arithmetical functions of two variables , _ mathematica _ , * 5*(1958 ) , 1 - 7 .
mitrinovi , j.e . peari and a.m. fink ,
_ classical and new inequalities in analysis , _ kluwer academic publishers , 1993 .
peari , on some classical inequalities in unitary spaces , _ mat .
bilten _ ( scopje ) , * 16*(1992 ) , 63 - 72 . in 1992 , j.e .
peari @xcite proved the following inequality for vectors in complex inner product spaces @xmath1570 .
[ t1.1.14]suppose that @xmath1366 are vectors in @xmath1 and @xmath1571 are complex numbers . then the following inequalities @xmath1572hold .
he also showed that for @xmath1573 @xmath1574 one gets @xmath1575which improves bombieri s result @xcite ( see also @xcite ) @xmath1576note that ( [ 1.3.14 ] ) is in its turn a natural generalisation of _ bessel s inequality _
@xmath1577which holds for the orthornormal vectors @xmath1578 in this section , by following @xcite , we point out other related results to peari s inequality ( [ 1.1.14 ] ) than the ones stated in the previous sections
. some results of bombieri type are also mentioned .
we start with the following lemma that is interesting in its own right 15nssd .
[ l2.1.14]let @xmath1372 and @xmath1373 .
then one has the inequalities : @xmath1579where@xmath1580@xmath1581and@xmath1582where
@xmath1391 @xmath1418 we observe that @xmath1583if one uses the hlder inequality for double sums , i.e. , we recall it @xmath1584where @xmath1585 @xmath1481 @xmath1542 then @xmath1586and the first inequality in ( [ 2.1.14 ] ) is proved
. observe that @xmath1587giving @xmath1588similarly , we have @xmath1589using ( [ 2.2.14 ] ) and ( [ 2.5.14 ] ) ( [ 2.6.14 ] ) , we deduce the 9 inequalities in the second part of ( [ 2.1.14 ] ) . if we choose @xmath1590 then the following result holds @xcite .
[ c2.2.14]if @xmath1372 and @xmath1373 , then one has @xmath1591where@xmath1592@xmath1593and@xmath1594 we are now able to point out the following result obtained in @xcite , which complements and generalises the inequality ( [ 1.1.14 ] ) due to j. peari .
[ t3.1.14]let @xmath1366 be vectors of an inner product space @xmath1361 and @xmath1394 .
then one has the inequalities : @xmath1595where@xmath1596@xmath1597and@xmath1598for @xmath1391 @xmath1418 we note that @xmath1503using schwarz s inequality in inner product spaces , we have @xmath1504finally , using lemma [ l2.1.14 ] with @xmath1400 @xmath1401 @xmath1402 we deduce the desired inequality ( [ 3.1.14 ] ) .
[ r3.2.14]if in ( [ 3.1.14 ] ) we choose @xmath1590 we obtain amongst others , the result @xmath1599 due to j. peari .
the following results of bombieri type hold @xcite .
[ t4.1.14]let @xmath1426 then one has the inequality : @xmath1600 ^{\frac{1}{2p}}\left [ \sum\limits_{i=1}^{n}\left\vert \left ( x , y_{i}\right ) \right\vert ^{q}\left ( \sum\limits_{j=1}^{n}\left\vert \left ( y_{i},y_{j}\right ) \right\vert \right ) \right ] ^{\frac{1}{2q } } \notag \\ & \leq \left\vert x\right\vert \times \left\ { \begin{array}{c } j \\ k \\
l% \end{array}% \right .
, \notag\end{aligned}\]]where@xmath1601@xmath1602and@xmath1603for @xmath1604 the proof follows by theorem [ t3.1.14 ] on choosing @xmath1605 and taking the square root in both sides of the inequalities involved .
we omit the details .
we observe , by the last inequality in @xmath1606 , that @xmath1607where @xmath1604 if in this inequality we choose @xmath1590 then we recapture bombieri s result @xmath1608 e. bombieri , a note on the large sieve , _ acta arith . , _ * 18 * ( 1971 ) , 401 - 404 . s.s .
dragomir , on peari s inequality in inner product spaces , _ rgmia res .
coll_. , * 6*(2003 ) , supplement , article 17 ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . `
mitrinovi , j.e . peari and a.m. fink ,
_ classical and new inequalities in analysis , _ kluwer academic publishers , 1993 .
peari , on some classical inequalities in unitary spaces , _ mat .
bilten _ ( skopje ) , * 16 * ( 1992 ) , 63 - 72 .
we start by recalling some of the most important grss type discrete inequalities for real numbers that are available in the literature . 1
. \(1950 ) _ biernacki , pidek , ryll - nardzewski _ @xcite .
+ if @xmath1609 and @xmath1610 are @xmath0-tuples of real numbers such that there exists the real numbers @xmath1611 with @xmath1612then @xmath1613 \left ( 1-\frac{1}{n}\left [ \frac{% n}{2}\right ] \right ) \left ( a - a\right ) \left ( b - b\right ) \\ & = & \frac{1}{n^{2}}\left [ \frac{n^{4}}{4}\right ] \left ( a - a\right ) \left ( b - b\right ) \\ & \leq & \frac{1}{4}\left ( a - a\right ) \left ( b - b\right ) .\end{aligned}\ ] ] 2 .
\(1988 ) _ andrica - badea _ @xcite . + let @xmath1614 , @xmath1615 satisfy ( [ b.1.15.0 ] ) and @xmath1616 be an @xmath33tuple of nonnegative numbers with @xmath1617 if @xmath1618 is a subset of @xmath1619 that minimises the expression @xmath1620then @xmath1621where @xmath1622 3 .
\(2000 ) _ dragomir - booth _ @xcite .
+ if @xmath1614 , @xmath1615 are real @xmath33tuples and @xmath1623 is nonnegative with @xmath1624 then @xmath1625where @xmath1626 and @xmath1627 is the forward difference , @xmath1628 note that @xmath1629 in particular , we have @xmath1630the constant @xmath518 is best possible .
\(2002 ) _ dragomir _ @xcite .
+ with the assumptions in 3 , the following inequality holds @xmath1631where @xmath1391 @xmath1418 + in particular , we have @xmath1632the constant @xmath1633 is best possible .
\(2002 ) _ dragomir _ @xcite . + the following inequality holds , where @xmath1634 and @xmath1635 are as in assumption 3 , @xmath1636 in particular , we have @xmath1637the constant @xmath222 is sharp . 6 .
\(2002 ) _ cerone - dragomir _ @xcite .
+ if @xmath1614 , @xmath1615 are real @xmath33tuples and @xmath1623 is a positive @xmath33tuple and there exists @xmath555 such that @xmath1638then one has the inequality @xmath1639the constant @xmath222is best possible .
+ in particular , we have @xmath1640the constant @xmath222 is best possible .
the main aim of this chapter is to present some extensions of the above results holding in the general setting of @xmath0-tuples of vectors in an inner product space .
m. biernacki , h. pidek and c. ryll - nardzewski , c. , sur une ingalit entre des intgrales definies , _ ann .
mariae curie - skolodowska , _ * a4*(1950 ) , 1 - 4 .
d. andricaand c. badea , grss inequality for positive linear functionals , _ periodica math .
hungarica , _ * 19*(2)(1988 ) , 155 - 167 .
dragomir and g.l .
booth , on a grss - lupa type inequality and its application for the estimation of @xmath548moments of guessing mappings , _ math .
comm . , _ * 5*(2000 ) , 117 - 126 .
dragomir , another grss type inequality for sequences of vectors in normed linear spaces and applications , _ j. comp . analysis & appl .
, _ * 4*(2 ) ( 2002 ) , 157 - 172 .
dragomir , a grss type inequality for sequences of vectors in normed linear spaces , _ rgmia res .
, _ * 5*(2 ) ( 2002 ) , article 9 . (
online : ` http://rgmia.vu.edu.au/v5n2.html ` ) p. cerone and s.s .
dragomir , a refinement of grss inequality and applications , _ rgmia res .
, _ * 5*(2 ) ( 2002 ) , article 14 . (
online : ` http://rgmia.vu.edu.au/v5n2.html`)(2002 ) the following lemma is of interest in itself @xcite .
[ l2.1.15]let @xmath848 be an inner product space over the real or complex number field @xmath1641 @xmath1642 and @xmath1643 @xmath1644 such that @xmath1645 @xmath1646 if @xmath1647 are such that @xmath1648or equivalently,@xmath1649then we have the inequality @xmath1650the constant @xmath59 is sharp .
define @xmath1651and @xmath1652then @xmath1653and @xmath1654consequently , @xmath1655taking the real value in @xmath1656 we can state that @xmath1657which is also an identity of interest in itself.using the assumption @xmath1658 we can conclude , by @xmath1659 that @xmath1660it is known that if @xmath1661 then @xmath1662with equality iff @xmath1663 now , by @xmath1664 we can state that @xmath1665using @xmath1666 we obtain @xmath1667 to prove the sharpness of the constant @xmath373 let us assume that the inequality @xmath1668 holds with a constant @xmath232 i.e. , @xmath1669for all @xmath1670 and @xmath0 as in the hypothesis of lemma [ l2.1.15 ] .
assume that @xmath1561 @xmath1671 @xmath1672 and @xmath1673 with @xmath1674 and @xmath1675 then , obviously , @xmath1676which shows that the condition @xmath1677 holds .
if we replace @xmath1678 in @xmath1679 we obtain @xmath1680from where we deduce that @xmath386 which proves the sharpness of the constant @xmath1681 [ r2.2.15]the assumption @xmath1677 can be replaced by the more general condition @xmath1682and the conclusion @xmath1668 will still remain valid .
the following corollary is natural .
[ c2.3.15]let @xmath1683 @xmath1684 @xmath1406 @xmath1685 with @xmath1686 if @xmath1687 are such that @xmath1688 \geq 0\text { for all } i\in \left\ { 1,\dots , n\right\ } , \label{2.9.15}\]]then we have the inequality @xmath1689the constant @xmath59 is sharp .
the proof follows by the above lemma by choosing @xmath1690 @xmath1691 @xmath1692 @xmath1693 and @xmath1694 we omit the details .
[ r2.4.15]the condition @xmath1695 can be replaced by the more general assumption @xmath1696 \geq 0.\ ] ] the following grss type inequality holds @xcite .
[ t3.1.15]let @xmath848 be an inner product space over @xmath1697 @xmath1698 @xmath1423 @xmath1685 with @xmath1686 if @xmath67 and @xmath1647 are such that @xmath1688 \geq 0,\text { } \func{re}\left\langle x - x_{i},x_{i}-x\right\rangle \geq 0\text { } \label{3.1.15a}\]]for all @xmath1699 then we have the inequality @xmath1700the constant @xmath59 is sharp .
a simple computation shows that @xmath1701taking the norm in both parts of @xmath1702 and using the generalized triangle inequality , we obtain @xmath1703by the cauchy - bunyakovsky - schwartz discrete inequality for double sums , we obtain @xmath1704as a simple calculation reveals that @xmath1705and @xmath1706then , by @xmath1707 and @xmath1708 we conclude that @xmath1709however , from lemma [ l2.1.15 ] and corollary [ c2.3.15 ] , we know that @xmath1710and @xmath1711consequently , by using @xmath1712 we deduce the desired estimate @xmath1713 to prove the sharpness of the constant @xmath373 assume that @xmath1714 holds with a constant @xmath232 i.e. , @xmath1715for all @xmath1716 and @xmath0 as in the hypothesis of theorem [ t3.1.15 ] . if we choose @xmath1561 @xmath1717 @xmath1718 @xmath1719 @xmath1673 @xmath1720 and @xmath1671 then @xmath1721consequently , from @xmath1722 we deduce @xmath1723which implies that @xmath386 and the theorem is completely proved .
[ r3.2.15]the condition @xmath1724 can be replaced by the more general assumption @xmath1696 \geq 0\text { , } \sum_{i=1}^{n}p_{i}\func{re}% \left\langle x - x_{i},x_{i}-x\right\rangle \geq 0\]]and the conclusion @xmath1725 will still be valid .
the following corollary for real or complex numbers holds .
[ c3.3.15]let @xmath1726 @xmath1727 @xmath1643 @xmath1644 with @xmath1728 if @xmath728 are such that @xmath1688 \geq 0,\text { \ } \func{re}\left [ \left ( b - b_{i}\right ) \left ( \bar{b}% _ { i}-\bar{b}\right ) \right ] \geq 0 , \label{3.11.15}\]]then we have the inequality @xmath1729where the constant @xmath59 is sharp .
[ r3.5.15]if we assume that @xmath1730 are real numbers , then @xmath1731 is equivalent to @xmath1732and @xmath1733 becomes @xmath1734which is the classical grss inequality for sequences of real numbers .
let @xmath312 be an inner product space over @xmath2 and @xmath1735 be a sequence of vectors in @xmath26 for a given @xmath1736 define the _ discrete fourier transform _ as @xmath1737the following approximation result for the fourier transform @xmath1738 holds @xcite .
[ t4.1.15]let @xmath848 and @xmath1739 be as above .
if there exists the vectors @xmath1674 such that @xmath1740then we have the inequality @xmath1741
\times \frac{1}{n}\sum_{k=1}^{n}x_{k}\right\vert \label{4.3.15 } \\
\leq \frac{1}{2}\left\vert x - x\right\vert \left [ n^{2}-\frac{\sin ^{2}\left ( wmn\right ) } { \sin ^{2}\left ( wm\right ) } \right ] ^{\frac{1}{2}},\end{gathered}\]]for all @xmath1742 and @xmath1743 from the inequality @xmath1744 in theorem [ t3.1.15 ] , we can state that @xmath1745for all @xmath1746 however , the @xmath1747 satisfy @xmath1748 and therefore , by lemma [ l2.1.15 ] , we have @xmath1749consequently , by @xmath1750 and @xmath1751 we conclude that @xmath1752for all @xmath1753 we now choose in @xmath1754 @xmath1755 to obtain @xmath1756for all @xmath1757 as a simple calculation reveals that @xmath1758 \\ & = \exp \left ( 2wim\right ) \times \left [ \frac{\cos \left ( 2wmn\right ) + i\sin \left ( 2wmn\right ) -1}{\cos \left ( 2wm\right ) + i\sin \left ( 2wm\right ) -1}\right ] \\ & = \exp \left ( 2wim\right ) \times \frac{\sin \left ( wmn\right ) } { \sin \left ( wm\right ) } \left [ \frac{\cos \left ( wmn\right ) + i\sin \left ( wmn\right ) } { % \cos \left ( wm\right ) + i\sin \left ( wm\right ) } \right ] \\ & = \frac{\sin \left ( wmn\right ) } { \sin \left ( wm\right ) } \times \exp \left ( 2wim\right ) \left [ \frac{\exp \left ( iwmn\right ) } { \exp \left ( iwm\right ) } % \right ] \\ & = \frac{\sin \left ( wmn\right ) } { \sin \left ( wm\right ) } \times \exp \left [ w\left ( n+1\right ) im\right ] , \end{aligned}\]]@xmath1759and @xmath1760thus , from @xmath1761 we deduce the desired inequality @xmath1762 [ r4.2.15]the assumption @xmath1763 can be replaced by the more general condition @xmath1764and the conclusion @xmath1765 will still remain valid .
the following corollary is an obvious consequence of @xmath1766 [ c4.3.15]let @xmath1767 @xmath1451 if @xmath1687 are such that @xmath1688 \geq 0\text { for all } i\in \left\ { 1,\dots , n\right\ } , \label{4.9.15}\]]then we have an approximation of the fourier transform for the vector @xmath1768@xmath1769 \times \frac{1}{n}\sum_{k=1}^{n}a_{k}\right\vert \label{4.10.15 } \\
\leq \frac{1}{2}\left\vert a - a\right\vert \left [ n^{2}-\frac{\sin ^{2}\left ( wmn\right ) } { \sin ^{2}\left ( wm\right ) } \right ] ^{\frac{1}{2}},\end{gathered}\]]for all @xmath1742 and @xmath1770 so that @xmath1771 [ r4.5.15]if we assume that @xmath1772 then @xmath1773 is equivalent to @xmath1774consequently , with the assumption @xmath1775 we obtain the following approximation of the fourier transform @xmath1769 \times \frac{1}{n}\sum_{k=1}^{n}a_{k}\right\vert \\
\leq \frac{1}{2}\left ( a - a\right ) \left [ n^{2}-\frac{\sin ^{2}\left ( wmn\right ) } { \sin ^{2}\left ( wm\right ) } \right ] ^{\frac{1}{2}},\end{gathered}\]]for all @xmath1742 and @xmath1776 let @xmath312 be an inner product over @xmath938 and @xmath1735 be a sequence of vectors in @xmath26 define the _ mellin transform _ : @xmath1777of the sequence @xmath1778 the following approximation result holds @xcite .
[ t5.1.15]let @xmath1 and @xmath1779 be as above . if there exist the vectors @xmath1647 such that @xmath1780then we have the inequality @xmath1781 ^{\frac{1}{2}},\;m\in \left\ { 1,\dots , n\right\ } , \end{gathered}\]]where @xmath1782 is the @xmath548powered sum of the first @xmath0 natural numbers , i.e. , @xmath1783 we apply the inequality @xmath1784 to obtain @xmath1785 ^{\frac{1}{2 } } \\ & = \frac{1}{2}\left\vert x - x\right\vert \left [ ns_{2m-2}\left ( n\right ) -s_{m-1}^{2}\left ( n\right ) \right ] ^{\frac{1}{2}},\end{aligned}\]]and the inequality @xmath1786 is proved .
consider the following particular values of mellin transform @xmath1787and @xmath1788the following corollary holds .
[ c5.2.15]let @xmath1 and @xmath1779 be as in theorem [ t5.1.15 ] .
then we have the inequalities : @xmath1789 ^{\frac{1}{2 } } \label{5.5.15}\]]and @xmath1790 [ r5.3.15]if we assume that @xmath1791 is a probability distribution , i.e. , @xmath1792 @xmath1793 and @xmath1794 and @xmath1795 @xmath1796 then by @xmath1797 and @xmath1798 we get the inequalities @xmath1799 ^{\frac{1}{2}}\]]and
@xmath1800 let @xmath312 be an inner product space over @xmath2 and @xmath1801 be a sequence of vectors in @xmath26 define the polynomial @xmath1802 with the coefficients @xmath1803 by @xmath1804the following approximation result for the polynomial @xmath1805 holds @xcite .
[ t6.1.15]let @xmath1806 and @xmath1805 be as above .
if there exist the vectors @xmath1807 such that @xmath1808then we have the inequality @xmath1809
^{\frac{1}{% 2}}\end{gathered}\]]for all @xmath230 @xmath1810 using the inequality @xmath1754 we can state that @xmath1811 ^{\frac{1}{2 } } \notag \\ & = \frac{1}{2}\left\vert c - c\right\vert \left [ \left ( n+1\right ) \frac{%
\left\vert z\right\vert ^{2n+2}-1}{\left\vert z\right\vert ^{2}-1}-\frac{% \left\vert z\right\vert ^{2n+2}-2\func{re}\left ( z^{n+1}\right ) + 1}{% \left\vert z\right\vert ^{2}-2\func{re}\left ( z\right ) + 1}\right ] ^{\frac{1}{% 2 } } \notag\end{aligned}\]]and the inequality @xmath1812 is proved .
the following result for the complex roots of the unity also holds 16nssd .
[ t6.2.15]let @xmath1813 be the complex @xmath1814roots of the unity .
then we have the inequality @xmath1815where the coefficients @xmath1816 satisfy the assumption @xmath1817 from the inequality @xmath1818 we can state that @xmath1819 ^{\frac{1}{2}}\end{gathered}\]]for all @xmath1820 putting @xmath1821 and taking into account that @xmath1822 we get the desired result @xmath1823 the following corollary is a natural consequence of theorem [ t6.2.15 ] . [ c6.3.15]let @xmath1824 be a polynomial with real coefficients and @xmath1825 the @xmath1826-roots of the unity as defined above .
if @xmath1827 then we have the inequality : @xmath1828 let @xmath312 be as above and @xmath1829 a mapping defined on the inner product space @xmath1 with values in the normed linear space @xmath1830 which satisfy the _
lipschitzian condition : _
@xmath1831where @xmath1832 denotes the norm on @xmath1830 and @xmath1833 is the euclidean norm on @xmath1 .
the following theorem holds @xcite .
[ t7.1.15]let @xmath1829 be as above and @xmath1834 @xmath1644 with @xmath1835 .
if there exists two vectors @xmath1647 such that @xmath1836then we have the inequality @xmath1837 as @xmath949 is lipschitzian , we have @xmath1838 for all @xmath1839 choose @xmath1840 and @xmath1841 to get @xmath1842for all @xmath1843 . if we multiply @xmath1844 by @xmath1845 and sum over @xmath1846 from @xmath1847 to @xmath0 , we obtain @xmath1848using the generalized triangle inequality , we have @xmath1849 by the cauchy - buniakowsky - schwarz inequality , we also have @xmath1850 ^{\frac{1}{2}}p_{n}^{% \frac{1}{2 } } \notag \\ & = p_{n}^{\frac{1}{2}}\left [ \sum_{j=1}^{n}p_{j}\left [ \left\vert \frac{1}{% p_{n}}\sum_{i=1}^{n}p_{i}x_{i}\right\vert ^{2}-2\func{re}\left\langle \frac{1% } { p_{n}}\sum_{i=1}^{n}p_{i}x_{i},x_{j}\right\rangle + \left\vert x_{j}\right\vert ^{2}\right ] \right ] ^{\frac{1}{2 } } \notag \\ & = p_{n}^{\frac{1}{2}}\left [ p_{n}\left\vert \frac{1}{p_{n}}% \sum_{i=1}^{n}p_{i}x_{i}\right\vert ^{2}-2\func{re}\left\langle \frac{1}{% p_{n}}\sum_{i=1}^{n}p_{i}x_{i},\sum_{j=1}^{n}p_{j}x_{j}\right\rangle + \sum_{j=1}^{n}p_{j}\left\vert x_{j}\right\vert ^{2}\right ] ^{\frac{1}{2 } } \notag \\ & = p_{n}\left [ \frac{1}{p_{n}}\sum_{i=1}^{n}p_{i}\left\vert x_{i}\right\vert ^{2}-\left\vert \frac{1}{p_{n}}\sum_{i=1}^{n}p_{i}x_{i}\right\vert ^{2}% \right ] ^{\frac{1}{2}}. \notag\end{aligned}\]]combining the above inequalities @xmath1851 we deduce , by dividing with @xmath1852 , that @xmath1853 ^{\frac{1}{2}}.\end{gathered}\]]finally , using lemma [ l2.1.15 ] , we obtain the desired result
. [ r7.9.15]the condition @xmath1854 can be substituted by the more general condition @xmath1855and the conclusion @xmath1856 will still remain valid .
the following corollary is a natural consequence of the above findings .
[ c7.10.15]let @xmath1642 @xmath1644 and @xmath1647 be such that the condition @xmath1854 holds .
then we have the inequality @xmath1857 the proof follows by theorem [ t7.1.15 ] by choosing @xmath1858 , @xmath1859 which is lipschitzian with the constant @xmath1860 , as @xmath1861 for all @xmath24 .
dragomir , a grss type discrete inequality in inner product spaces and applications , _ j. math .
_ , * 250 * ( 2000 ) , 494 - 511 .
the following grss type inequality holds @xcite .
[ gtisvipsappt3.1]let @xmath1274 be an inner product space over @xmath1862 @xmath1863 @xmath1864 , @xmath1643 @xmath1865 @xmath1866 with @xmath1867 .
if @xmath1868 are such that @xmath1869 for all @xmath35 then we have the inequality @xmath1870the constant @xmath59 is sharp .
a simple calculation shows that @xmath1871taking the modulus in both parts of ( [ gtisvipsapp3.3 ] ) and using the generalized triangle inequality , we obtain @xmath1872by schwarz s inequality in inner product spaces we have @xmath1873for all @xmath1874 and therefore @xmath1875using the cauchy - bunyakovsky - schwarz inequality for double sums , we can state that @xmath1876and , as a simple calculation shows that , @xmath1877and @xmath1878we obtain @xmath1879using lemma [ l2.1.15 ] , we know that @xmath1880and @xmath1881and then , by ( [ gtisvipsapp3.8 ] ) , we deduce the desired inequality ( gtisvipsapp3.3 ) . to prove the sharpness of the constant @xmath59 , let us assume that ( [ gtisvipsapp3.2 ] ) holds with a constant @xmath334 , i.e. , @xmath1882 under the above assumptions for @xmath1883 and @xmath1884 . if we choose @xmath1885 , @xmath1886 @xmath1887 @xmath1888 and @xmath1889 , then @xmath1890and then @xmath1891choose @xmath1892 @xmath1893 @xmath1894 . then using ( [ gtisvipsapp3.9 ] ) , we derive @xmath1895which implies that @xmath1896 , and the theorem
is proved .
[ gtisvipsappr3.2]the condition ( [ gtisvipsapp3.1 ] ) can be replaced by the more general assumption @xmath1897and the conclusion ( [ gtisvipsapp3.2 ] ) still remains valid .
the following corollary for real or complex numbers holds .
[ gtisvipsappc3.3]let @xmath1898 @xmath1899 @xmath1900 @xmath1643 @xmath1644 with @xmath1728 if @xmath728 are such that @xmath1688 \geq 0,\;\;\func{re}\left [ \left ( b - b_{i}\right ) \left ( \bar{b}_{i}-% \bar{b}\right ) \right ] \geq 0 , \label{gtisvipsapp3.11}\]]then we have the inequality @xmath1901and the constant @xmath59 is sharp .
the proof is obvious by theorem [ gtisvipsappt3.1 ] applied for the inner product space @xmath1902 where @xmath1903 .
we omit the details .
[ gtisvipsappr3.4]the condition ( [ gtisvipsapp3.11 ] ) can be replaced by the more general condition @xmath1696 \geq 0,\;\;\sum_{i=1}^{n}p_{i}\func{re}\left [ \left ( b - b_{i}\right ) \left ( \bar{b}_{i}-\bar{b}\right ) \right ] \geq 0\]]and the conclusion of the above corollary will still remain valid .
[ gtisvipsappr3.5]if we assume that @xmath1898 @xmath1904 , @xmath1905 , @xmath1906 , @xmath102 , @xmath1830 are real numbers , then ( [ gtisvipsapp3.11 ] ) is equivalent to @xmath1907and ( [ gtisvipsapp3.12 ] ) becomes @xmath1734which is the classical grss inequality for sequences of real numbers .
let @xmath312 be a real inner product space and @xmath1908 a frchet differentiable convex mapping on @xmath1 .
then we have the _ gradient inequality _
@xmath1909for all @xmath24 , where @xmath1910 is the gradient operator associated to the differentiable convex function @xmath949 .
the following theorem holds @xcite .
[ gtisvipsappt4.1]let @xmath1908 be as above and @xmath1911 @xmath1644 .
suppose that there exists the vectors @xmath1647 such that @xmath1912 for all @xmath1913 and @xmath1914 such that @xmath1915 for all @xmath1913 .
then for all @xmath1643 @xmath1916 with @xmath1917 , we have the inequality @xmath1918 choose in ( [ gtisvipsapp4.1 ] ) , @xmath1919 and @xmath1920 to obtain @xmath1921for all @xmath1843 .
if we multiply ( [ gtisvipsapp4.3 ] ) by @xmath1845 and sum over @xmath1922 from @xmath1847 to @xmath1923 , we have @xmath1924dividing by @xmath1925 we obtain the inequality @xmath1926which is a generalisation for the case of inner product spaces of the result by dragomir - goh established in 1996 for the case of differentiable mappings defined on @xmath1927 @xcite .
applying theorem [ gtisvipsappt3.1 ] for real inner product spaces , and @xmath1928 , we easily deduce @xmath1929and then , by ( [ gtisvipsapp4.4 ] ) and ( [ gtisvipsapp4.5 ] ) we can conclude that the desired inequality ( [ gtisvipsapp4.2 ] ) holds .
let @xmath312 be an inner product space over @xmath2 , @xmath1930 , @xmath938 and @xmath1931 be a sequence of vectors in @xmath1 . for a given @xmath1932
, define the _ discrete fourier transform _ @xmath1933the complex number @xmath1934 is actually the usual fourier transform of the vector @xmath1935 and will be denoted by @xmath1936 the following result holds @xcite .
[ gtisvipsappt5.1]let @xmath1779 , @xmath1937 be sequences of vectors such that there exists the vectors @xmath1938 with the properties @xmath1939and @xmath1940then we have the inequality @xmath1941for all @xmath1742 .
the proof follows by theorem [ gtisvipsappt3.1 ] applied for @xmath1942 and for the sequences @xmath1943 and @xmath1944 @xmath1945 .
we omit the details .
we can also consider the _
mellin transform _
@xmath1946of the sequence @xmath1947 .
we remark that the complex number @xmath1948 is actually the mellin transform of the vector @xmath1949 and will be denoted by @xmath1950 the following theorem holds @xcite .
[ gtisvipsappt5.2]let @xmath1779 , @xmath1937 be sequences of vectors such that there exist the vectors @xmath1951 with the properties @xmath1952for all @xmath1953 and ( [ gtisvipsapp5.4 ] ) is fulfilled .
then we have the inequality @xmath1954for all @xmath1742 .
the proof follows by theorem [ gtisvipsappt3.1 ] applied for @xmath1942 and for the sequences @xmath1955 and @xmath1944 @xmath1956 .
we omit the details .
another result which connects the fourier transforms for different parameters @xmath1957 also holds @xcite . [ gtisvipsappt5.3]let @xmath1779 , @xmath1937 and @xmath1958 .
if there exist the vectors @xmath1959 such that @xmath1960and @xmath1961then we have the inequality : @xmath1962for all @xmath1742 .
the proof follows by theorem [ gtisvipsappt3.1 ] for the sequences @xmath1963 , @xmath1964 @xmath1793 .
we omit the details .
dragomir , a grss type inequality for sequences of vectors in inner product spaces and applications , _ journal of inequalities in pure and applied mathematics _ , * 1*(2 ) , 2000 , article 12 . `
[ on line : http://jipam.vu.edu.au/v1n2/002_00.pdf ] ` s.s .
dragomir and c.j .
goh , a counterpart of jensen s discrete inequality for differentiable convex mappings and applications in information theory , _ mathl .
modelling _ ,
* 24 * ( 2 ) ( 1996 ) , 1 - 11 . in the recent paper @xcite
, the author has obtained the following grss type inequality for forward difference .
[ t1.3.16]let @xmath1965 @xmath1966 and @xmath1967 be a probability sequence .
then one has the inequalities@xmath1968 \max\limits_{k=1,\dots , n-1}\left\vert \delta x_{k}\right\vert \max\limits_{k=1,\dots , n-1}\left\vert \delta y_{k}\right\vert ; \\ \\
\sum\limits_{1\leq j < i\leq n}p_{i}p_{j}\left ( i - j\right ) \left ( \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert ^{p}\right ) ^{% \frac{1}{p}}\left ( \sum\limits_{k=1}^{n-1}\left\vert \delta y_{k}\right\vert ^{q}\right ) ^{\frac{1}{q } }
\\ \hfill \text{if } p>1,\ \frac{1}{p}+\frac{1}{q}=1 ; \\
\\ \dfrac{1}{2}\left [ \sum\limits_{i=1}^{n}p_{i}\left ( 1-p_{i}\right ) \right ] \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert \sum\limits_{k=1}^{n-1}\left\vert \delta y_{k}\right\vert .% \end{array}% \right.\end{gathered}\]]the constants @xmath1969 @xmath1847 and @xmath222 in the right hand side of the inequality ( [ 1.6.16 ] ) are best in the sense that they can not be replaced by smaller constants .
if one chooses @xmath1970 @xmath1644 in ( 1.6.16 ) , then the following unweighted inequalities hold:@xmath1971here , the constants @xmath1972 @xmath1633 and @xmath222 are also best possible in the above sense . the following reverse of the cauchy - bunyakovsky - schwarz inequality for sequences of vectors in inner product spaces holds .
[ c1.4.16]with the assumptions in theorem [ t1.3.16 ] for @xmath1973 and @xmath1974 one has the inequalities@xmath1975 \max\limits_{k=\overline{1,n-1% } } \left\vert \delta x_{k}\right\vert ^{2 } ; \\ \\
\sum\limits_{1\leq j < i\leq n}p_{i}p_{j}\left ( i - j\right ) \left ( \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert ^{p}\right ) ^{% \frac{1}{p}}\left ( \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert ^{q}\right ) ^{\frac{1}{q } }
\\ \hfill \text{if } p>1,\ \frac{1}{p}+\frac{1}{q}=1 ; \\
\\ \dfrac{1}{2}\left [ \sum\limits_{i=1}^{n}p_{i}\left ( 1-p_{i}\right ) \right ] \left ( \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert \right ) ^{2}.% \end{array}% \right .
\notag\end{aligned}\]]the constants @xmath1969 @xmath1847 and @xmath222 are best possible in the above sense .
the following particular inequalities that may be deduced from ( [ 1.8.16 ] ) on choosing the equal weights @xmath1976 @xmath1977 are also of interest@xmath1978here the constants @xmath1972 @xmath1633 and @xmath222 are also best possible .
the main aim of this section is to present , by following @xcite , a different class of grss type inequalities for sequences of vectors in inner product spaces and to apply them for obtaining a reverse of jensen s inequality for convex functions defined on such spaces .
the following lemma holds ( see also @xcite ) .
[ l2.1.16]let @xmath583 be vectors in the inner product space @xmath1979 over the real or complex number field @xmath2 @xmath1980 with @xmath1981 the following statements are equivalent : 1 .
@xmath1982 2 .
@xmath1983 the following inequality of grss type for sequences of vectors in inner product spaces holds @xcite .
[ t2.3.16]let @xmath848 be an inner product over @xmath2 @xmath1984 , and @xmath1965 @xmath1985 @xmath1967 with @xmath1686 if @xmath1647 are such that@xmath1986or equivalently,@xmath1987then one has the inequality@xmath1988 ^{\frac{1}{2}}. \notag\end{aligned}\]]the constant @xmath222 is best possible in the first and second inequality in the sense that it can not be replaced by a smaller constant .
it is easy to see that the following identity holds true@xmath1989taking the modulus in ( [ 2.4.16 ] ) and using the schwarz inequality in the inner product space @xmath1990 we have@xmath1991and the first inequality in ( [ 2.3.16 ] ) is proved . using the cauchy - bunyakovsky - schwarz inequality for positive sequences and the calculation rules in inner product spaces , we have@xmath1992 ^{\frac{1}{2}}\]]and@xmath1993giving the second part of ( [ 2.3.16 ] ) . to prove the sharpness of the constant @xmath222 in the first inequality in ( [ 2.3.16 ] ) , let us assume that , under the assumptions of the theorem , the inequality holds with a constant @xmath78 i.e. , @xmath1994consider @xmath1885 and observe that@xmath1995and then , by ( [ 2.4a.16 ] ) , we deduce@xmath1996if we choose @xmath1997 @xmath1998 @xmath1999 and @xmath2000 @xmath2001 with @xmath2002 then ( [ 2.2.16 ] ) holds and from ( [ 2.5.16 ] ) we deduce
@xmath2003 the fact that @xmath222 is best possible in the second inequality may be proven in a similar manner and we omit the details .
[ r2.4.16]if @xmath2004 and @xmath2005 satisfy the assumptions of theorem [ t2.3.16 ] , or equivalently@xmath2006for each @xmath35 then by theorem [ t2.3.16 ] we may state the following sequence of inequalities improving the grss inequality ( [ 2.4.16])@xmath2007 in particular , for @xmath2008 @xmath1402 one has@xmath2009and the constant @xmath222 is best possible .
the following result also holds @xcite .
[ t2.5.16]let @xmath848 and @xmath2 be as above and @xmath2010 @xmath2011 and @xmath2012 a probability vector . if @xmath1647 are such that ( [ 2.1.16 ] ) or equivalently , ( [ 2.2.16 ] ) holds , then we have the inequality@xmath2013 ^{\frac{1}{2}}. \notag\end{aligned}\]]the constant @xmath222 in the first and second inequalities is best possible in the sense that it can not be replaced by a smaller constant .
we start with the following equality that may be easily verified by direct calculation@xmath2014if we take the norm in ( [ 2.10.16 ] ) , we deduce@xmath2015proving the inequality ( [ 2.9.16 ] ) .
the fact that the constant @xmath222 is sharp may be proven in a similar manner to the one embodied in the proof of theorem [ t2.3.16 ] .
we omit the details .
[ r2.6.16]if @xmath2004 and @xmath2016 satisfy the assumption @xmath2017for each @xmath35 then by theorem [ t2.3.16 ] we may state the following sequence of inequalities improving the grss inequality @xmath2018 [ r2.7.16]if in ( [ 2.9.16 ] ) we choose @xmath2019 and assume that @xmath2020 where @xmath2021 , then we get the following interesting inequality for complex numbers@xmath2022 ^{\frac{1}{2}}.\end{aligned}\ ] ] let @xmath312 be a real inner product space and @xmath1908 a frchet differentiable convex function on @xmath26 if @xmath2023 denotes the gradient operator associated to @xmath964 then we have the inequality@xmath2024for each @xmath12 the following result holds @xcite .
[ t3.1.16]let @xmath1908 be as above and @xmath2025 @xmath210 suppose that there exist the vectors @xmath2026 such that either@xmath2027or equivalently,@xmath2028if @xmath2029 @xmath2030 with @xmath2031 then we have the following converse of jensen s inequality@xmath2032 ^{\frac{1}{2}}. \notag\end{aligned}\ ] ] we know , see for example ( * ? ? ?
* eq . ( 4.4 ) ) , that the following reverse of jensen s inequality for frchet differentiable convex functions@xmath2033holds . now , if we use theorem [ t2.3.16 ] for the choices @xmath2034 @xmath2035 and @xmath2036 @xmath1574 we can state the inequality@xmath2037 ^{\frac{1}{2}}. \notag\end{aligned}\]]utilizing ( [ 3.5.16 ] ) and ( [ 3.6.16 ] ) , we deduce the desired result ( [ 3.4.16 ] ) .
if more information is available about the vector sequence @xmath2038 then we may state the following corollary .
[ c3.2.16]with the assumptions in theorem [ t3.1.16 ] and if there exist the vectors @xmath2039 such that either@xmath2040or , equivalently@xmath2041then we have the inequality@xmath2042 ^{\frac{1}{2 } } \notag \\ & \leq \frac{1}{4}\left\vert m - m\right\vert \left\vert z - z\right\vert . \notag\end{aligned}\ ] ] note that the inequality between the first term and the last term in ( 3.9.16 ) was first proved in ( * ? ? ? * theorem 4.1 ) .
consequently , the above corollary provides an improvement of the reverse of jensen s inequality established in @xcite .
dragomir , a grss type discrete inequality in inner product spaces and applications , _ j. math .
_ * 250 * ( 2000 ) , 494 - 511 .
dragomir , a grss type inequality for sequences of vectors in inner product spaces and applications , _ j. ineq .
pure & appl .
, _ * 1*(2 ) ( 2000 ) , article 12 . [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] .
dragomir , grss type inequalities for forward difference of vectors in inner product spaces , _ submitted .
dragomir , some grss type inequalities in inner product spaces , _ j. inequal .
pure & appl .
_ , * 4*(2003 ) , no .
2 , article 42 , [ ` on line : http://jipam.vu.edu.au/v4n2/032_03.html ` ] s.s . dragomir , grss type discrete inequalities in inner product spaces , revisited , _ preprint _ , [ ` on line ` ` http://www.mathpreprints.com/math/ ` ` preprint / sever/20030623.2/1/ ] . ` in @xcite , we have proved the following generalisation of the grss inequality . [ ta.16.0 ]
let @xmath2043 be an inner product space over @xmath1697 @xmath2044 and @xmath930 @xmath299 if @xmath1124 and @xmath24 are such that@xmath2045hold , then we have the inequality@xmath2046the constant @xmath59 is the best possible .
a grss type inequality for sequences of vectors in inner product spaces was pointed out in @xcite .
[ tb.16.0]let @xmath1 and @xmath2 be as in theorem [ ta.16.0 ] and @xmath2047 @xmath1683 @xmath1643 @xmath1406 @xmath1685 with @xmath1867 . if @xmath1687 and @xmath1647 are such that:@xmath2048 \geq 0,\ \ \ \func{re}\left\langle x - x_{i},x_{i}-x\right\rangle \geq 0\]]for any @xmath35 then we have the inequality@xmath2049the constant @xmath59 is best possible .
a complementary result for two sequences of vectors in inner product spaces is the following result that has been obtained in @xcite .
[ tc.16.0]let @xmath1 and @xmath2 be as above , @xmath2050
@xmath2051 @xmath1644 @xmath1685 with @xmath1645
. if @xmath1868 are such that:@xmath2052for all @xmath35 then we have the inequality@xmath2053the constant @xmath59 is best possible . in the general case of normed linear spaces ,
the following grss type inequality in terms of the forward difference is known , see @xcite .
[ td.16.0]let @xmath2054 be a normed linear space over @xmath2055 @xmath2056 @xmath2057 and @xmath1643 @xmath1644 such that @xmath1867 .
then we have the inequality@xmath2058 , \notag\end{aligned}\]]where @xmath2059 and @xmath2060 are the forward differences of the vectors having the components @xmath2061 and @xmath2062 respectively .
the inequality ( [ 1.7.16.0 ] ) is sharp in the sense that the multiplicative constant @xmath319 in the right hand side can not be replaced by a smaller one .
an important particular case is the one where all the weights are equal , giving the following corollary @xcite .
[ ce.16.0]under the above assumptions for @xmath2063 @xmath1423 we have the inequality@xmath2064the constant @xmath518 is best possible .
another result of this type was proved in @xcite .
[ tf.16.0]with the assumptions of theorem [ td.16.0 ] , one has the inequality@xmath2065the constant @xmath222 is best possible . as a useful particular case , we have the following corollary @xcite .
[ ch.16.0]if @xmath2063 @xmath1644 are as in theorem [ td.16.0 ] , then@xmath2066the constant @xmath222 is the best possible . finally , the following result is also known @xcite .
[ tg.16.0]with the assumptions in theorem [ td.16.0 ] , we have the inequality:@xmath2067where @xmath1391 @xmath1418 the constant @xmath423 in the right hand side of ( [ 1.11.16.0 ] ) is sharp .
the case of equal weights is embodied in the following corollary eb.16.0 .
[ cl.16.0]with the above assumptions for @xmath2063 @xmath1423 one has@xmath2068where @xmath1391 @xmath1418 the constant @xmath1633 is the best possible .
the main aim of this section is to establish some similar bounds for the absolute value of the difference@xmath2069provided that @xmath2070 @xmath1644 are vectors in an inner product space @xmath923 and @xmath1643 @xmath1644 with @xmath1686 we assume that @xmath564 is an inner product space over @xmath1697 @xmath1269 or @xmath2071 the following discrete inequality of grss type holds .
[ t2.1.16.0]if @xmath2050 @xmath1643 @xmath1406 with @xmath1867 , then one has the inequalities:@xmath2072 \max\limits_{k=1,\dots , n-1}\left\vert \delta x_{k}\right\vert \max\limits_{k=1,\dots , n-1}\left\vert \delta y_{k}\right\vert ; \\ \\
[ \sum\limits_{1\leq j < i\leq n}p_{i}p_{j}\left ( i - j\right ) \right ] \left ( \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert ^{p}\right ) ^{\frac{1}{p}}\left ( \sum\limits_{k=1}^{n-1}\left\vert \delta y_{k}\right\vert ^{q}\right ) ^{\frac{1}{q } } \\ \hfill \text{if } p>1,\ \frac{1}{p}+\frac{1}{q}=1 ; \\
\\ \dfrac{1}{2}\left [ \sum\limits_{i=1}^{n}p_{i}\left ( 1-p_{i}\right ) \right ] \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert \sum\limits_{k=1}^{n-1}\left\vert \delta y_{k}\right\vert .% \end{array}% \right .
\notag\end{aligned}\]]all the inequalities in ( [ 2.1.16.0 ] ) are sharp .
the following particular case for equal vectors holds .
[ c2.1.a.16.0]with the assumptions of theorem [ t2.1.16.0 ] , one has the inequalities@xmath2073 \max\limits_{k=1,\dots , n-1}\left\vert \delta x_{k}\right\vert ^{2 } ; \\ \\
\sum\limits_{1\leq j < i\leq n}p_{i}p_{j}\left ( i - j\right ) \left ( \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert ^{p}\right ) ^{% \frac{1}{p}}\left ( \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert ^{q}\right ) ^{\frac{1}{q } }
\\ \hfill \text{if } p>1,\ \frac{1}{p}+\frac{1}{q}=1 ; \\
\\ \dfrac{1}{2}\sum\limits_{i=1}^{n}p_{i}\left ( 1-p_{i}\right ) \left ( \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert \right ) ^{2}.% \end{array}% \right.\end{aligned}\ ] ] the following particular case for equal weights may be useful in practice .
[ c2.2.16.0]if @xmath1864 @xmath1644 , then one has the inequalities:@xmath2074@xmath2075the constants @xmath518 , @xmath1633 and @xmath222 are best possible
. in particular , the following corollary holds .
[ c2.3.16.0]if @xmath1642 @xmath1402 then one has the inequality@xmath2076the constants @xmath518 , @xmath1633 and @xmath222 are best possible .
it is well known that , the following identity holds in inner product spaces:@xmath2077we observe , for @xmath2078 we can write that @xmath2079taking the modulus in ( [ 3.1.16.0 ] ) and by the use of ( [ 3.2.16.0 ] ) and schwarz s inequality in inner product spaces , i.e. , we recall that @xmath2080 @xmath2081 we have:@xmath2082it is obvious that@xmath2083and@xmath2084giving that@xmath2085and since@xmath2086the first inequality in ( [ 2.1.16.0 ] ) is proved . using hlder s discrete inequality , we can state that@xmath2087and@xmath2088for @xmath1391 @xmath1481 giving that:@xmath2089and the second inequality in ( [ 2.1.16.0 ] ) is proved .
also , observe that@xmath2090and thus@xmath2091since@xmath2092 \\ & = \frac{1}{2}\left ( 1-\sum\limits_{k=1}^{n}p_{k}^{2}\right ) \\ & = \frac{1}{2}\sum\limits_{i=1}^{n}p_{i}\left ( 1-p_{i}\right ) , \end{aligned}\]]the last part of ( [ 2.1.16.0 ] ) is also proved .
now , assume that the first inequality in ( [ 2.1.16.0 ] ) holds with a constant @xmath232 i.e.,@xmath2093 \max\limits_{k=1,\dots , n-1}\left\vert \delta x_{k}\right\vert \max\limits_{k=1,\dots , n-1}\left\vert \delta y_{k}\right\vert\end{aligned}\]]and choose @xmath1885 to get@xmath2094for any @xmath2095 and @xmath2096 if in ( [ 3.5.16.0 ] ) we choose @xmath1998 @xmath1999 and @xmath2097 then we deduce @xmath2098 which proves the sharpness of the constant in the first inequality in ( [ 2.1.16.0 ] ) . in a similar way one may show that the other two inequalities are sharp , and the theorem is completely proved .
let @xmath312 be a real inner product space and @xmath1908 a frchet differentiable convex function on @xmath26 if @xmath2023 denotes the gradient operator associated to @xmath964 then we have the inequality@xmath2099for each @xmath12 the following result holds .
[ t4.1.16.0]let @xmath1908 be as above and @xmath2100 @xmath210 if @xmath2029 @xmath2101 with @xmath2102 then we have the following reverse of jensen s inequality@xmath2103 \max\limits_{k=1,\dots , n-1}\left\vert \delta \left ( \triangledown f\left ( z_{i}\right ) \right ) \right\vert \max\limits_{k=1,\dots , n-1}\left\vert \delta z_{i}\right\vert ; \\ \\
\left [ \sum\limits_{1\leq j < i\leq n}q_{i}q_{j}\left ( i - j\right ) \right ] \left ( \sum\limits_{i=1}^{n-1}\left\vert \delta \left ( \triangledown f\left ( z_{i}\right ) \right ) \right\vert ^{p}\right ) ^{\frac{1}{p}}\left ( \sum\limits_{i=1}^{n-1}\left\vert \delta z_{i}\right\vert ^{q}\right ) ^{% \frac{1}{q } } \\
\hfill \text{if } p>1,\ \frac{1}{p}+\frac{1}{q}=1 ; \\
\\ \dfrac{1}{2}\left [ \sum\limits_{i=1}^{n}q_{i}\left ( 1-q_{i}\right ) \right ] \sum\limits_{i=1}^{n-1}\left\vert \delta \left ( \triangledown f\left ( z_{i}\right ) \right ) \right\vert \sum\limits_{i=1}^{n-1}\left\vert \delta z_{i}\right\vert .% \end{array}% \right .
\notag \\ & \notag\end{aligned}\ ] ] we know , see for example ( * ? ? ?
* eq . ( 4.4 ) ) , that the following reverse of jensen s inequality for frchet differentiable convex functions@xmath2104holds .
now , if we apply theorem [ t2.1.16.0 ] for the choices @xmath2105 and @xmath2106 then we may state@xmath2107 \max\limits_{k=1,\dots , n-1}\left\vert \delta \left ( \triangledown f\left ( z_{k}\right ) \right ) \right\vert \max\limits_{k=1,\dots , n-1}\left\vert \delta z_{k}\right\vert ; \\ \\
\left [ \sum\limits_{1\leq j < i\leq n}q_{i}q_{j}\left ( i - j\right ) \right ] \left ( \sum\limits_{k=1}^{n-1}\left\vert \delta \left ( \triangledown f\left ( z_{k}\right ) \right ) \right\vert ^{p}\right ) ^{\frac{1}{p}}\left ( \sum\limits_{k=1}^{n-1}\left\vert \delta z_{k}\right\vert ^{q}\right ) ^{% \frac{1}{q } } \\
\hfill \text{if } p>1,\ \frac{1}{p}+\frac{1}{q}=1 ; \\
\\ \dfrac{1}{2}\left [ \sum\limits_{i=1}^{n}q_{i}\left ( 1-p_{i}\right ) \right ] \sum\limits_{k=1}^{n-1}\left\vert \delta \left ( \triangledown f\left ( z_{k}\right ) \right ) \right\vert \sum\limits_{k=1}^{n-1}\left\vert \delta z_{k}\right\vert .% \end{array}% \right .
\notag\end{aligned}\]]finally , on making use of the inequalities ( [ 4.3.16.0 ] ) and ( 4.4.16.0 ) , we deduce the desired result ( [ 4.2.16.0 ] ) .
the unweighted case may useful in application and is incorporated in the following corollary .
let @xmath1908 be as above and @xmath2025 @xmath2108 then we have the inequalities@xmath2109 s.s .
dragomir , a generalisation of grss inequality in inner product space and applications , _ j. math .
appl . , _ * 237 * ( 1999 ) , 74 - 82 .
dragomir , a grss type discrete inequality in inner product spaces and applications , _ j. math .
_ * 250 * ( 2000 ) , 494 - 511 .
dragomir , a grss type inequality for sequences of vectors in inner product spaces and applications , _ j. ineq .
pure & appl .
, _ * 1*(2 ) ( 2000 ) , article 12 . ` [ on line : http://jipam.vu.edu.au ] ` .
dragomir and g.l .
booth , on a grss - lupa type inequality and its applications for the estimation of @xmath2110-moments of guessing mappings , _ math .
_ * 5 * ( 2000 ) , 117 - 126 .
dragomir , another grss type inequality for sequences of vectors in normed linear spaces and applications , _ j. comp .
_ * 4*(2 ) ( 2002 ) , 155 - 172 .
dragomir , a grss type inequality for sequences of vectors in normed linear spaces and applications , _ rgmia res .
, _ * 5*(2002 ) , no.2 , article 9 ` [ on line : http://rgmia.vu.edu.au/v5n2.html ] ` j.e .
peari , f. proschan and y.l .
tong , _ convex functions , partial orderings and statistical applications _ , academic press , san diego , 1992 .
mitrinovi , j.e . peari and a.m. fink ,
_ classical and new inequalities in analysis , _ kluwer academic publishers , 1993 .
let @xmath312 be an inner product over the real or complex number field @xmath2 .
for @xmath2111 and @xmath2112 @xmath2113 define the _
_ ebyev functional__@xmath2114where @xmath2115 the following grss type inequality has been obtained in @xcite .
[ t1.1.16.1]let @xmath923 @xmath2116 be as above and @xmath2117 @xmath2030 with @xmath2118 i.e. , @xmath2119 is a probability sequence .
if @xmath2120 are such that@xmath2121for each @xmath35 or , equivalently , ( see ssd2.16.1)@xmath2122for each @xmath35 then we have the inequality@xmath2123the constant @xmath59 is best possible in the sense that it can not be replaced by a smaller constant . in @xcite , the following grss type inequality for the forward difference of vectors was established .
[ t1.2.16.1]let @xmath2124 @xmath2125 and @xmath2126 be a probability sequence .
then one has the inequality:@xmath2127 \max\limits_{1\leq k\leq n-1}\left\vert \delta x_{k}\right\vert \max\limits_{1\leq k\leq n-1}\left\vert \delta y_{k}\right\vert ; \\ \\
\sum\limits_{1\leq j < i\leq n}p_{i}p_{j}\left ( i - j\right ) \left ( \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert ^{p}\right ) ^{% \frac{1}{p}}\left ( \sum\limits_{k=1}^{n-1}\left\vert \delta y_{k}\right\vert ^{q}\right ) ^{\frac{1}{q } } \\ \hfill \text { \ if \ } p>1,\ \frac{1}{p}+\frac{1}{q}+1 \\ \\
\dfrac{1}{2}\left [ \sum\limits_{i=1}^{n}p_{i}\left ( 1-p_{i}\right ) \right ] \sum\limits_{k=1}^{n-1}\left\vert \delta x_{k}\right\vert \sum\limits_{k=1}^{n-1}\left\vert \delta y_{k}\right\vert .% \end{array}% \right .
\notag\end{aligned}\]]the constants @xmath2128 and @xmath222 in the right hand side of inequality ( [ 1.4.16.1 ] ) are best in the sense that they can not be replaced by smaller constants . another result is incorporated in the following theorem ( see @xcite ) .
[ t1.3.16.1]let @xmath2116 and @xmath2119 be as in theorem [ t1.2.16.1 ] .
if there exist @xmath1647 such that@xmath2129or , equivalently,@xmath2130then one has the inequality@xmath2131 ^{\frac{1}{2}}. \notag\end{aligned}\]]the constant @xmath222 is best possible in the first and second inequalities in the sense that it can not be replaced by a smaller constant . if @xmath2132 and @xmath2133 satisfy the assumptions of theorem t1.1.16.1 , then we have the following sequence of inequalities improving the grss inequality ( [ 1.3.16.1]):@xmath2134 now , if we consider the ebyev functional for the uniform probability distribution @xmath2135@xmath2136then , with the assumptions of theorem [ t1.1.16.1 ] , we have@xmath2137 theorem [ t1.2.16.1 ] will provide the following inequalities@xmath2138here the constants @xmath1972 @xmath1633 and @xmath222 are best possible in the above sense . finally , from ( [ 1.8.16.1 ]
) , we have@xmath2139 it is the main aim of this section to point out other bounds for the ebyev functionals @xmath2140 and @xmath2141 for @xmath2142 and @xmath2143 we define@xmath2144and the vectors@xmath2145for @xmath2146 the following result holds .
[ t2.1.16.1]let @xmath1274 be an inner product space over @xmath2 , @xmath2147 and @xmath2148 then we have the identities@xmath2149where @xmath2150 @xmath2151 is the forward difference .
we use the following summation by parts formula for vectors in inner product spaces@xmath2152where @xmath2153 @xmath2154 are vectors in @xmath923 @xmath2155 ( @xmath2156 @xmath2157 are natural numbers ) . if we choose in (
[ 2.2.16.1 ] ) , @xmath2158 @xmath2159 @xmath2160 and @xmath2161 @xmath2162
then we get@xmath2163proving the first identity in ( [ 2.1.16.1 ] ) .
the second and third identities are obvious and we omit the details .
the following lemma is of interest in itself .
[ l2.2.16.1]let @xmath2164 and @xmath2165 then we have the equality@xmath2166for each @xmath2146 define , for @xmath2167 the vector@xmath2168we have@xmath2169using the summation by parts formula , we have@xmath2170and@xmath2171using ( [ 2.5.16.1 ] ) and ( [ 2.6.16.1 ] ) , we have@xmath2172 a_{i+1}+p_{i}\sum_{j = i+1}^{n-1}p_{j+1}a_{j+1}-\bar{p}% _ { i}\sum_{j=1}^{i+1}p_{j}a_{j } \\ & = p_{n}p_{i+1}a_{i+1}+p_{i}\sum_{j = i+1}^{n-1}p_{j+1}a_{j+1}-\bar{p}% _ { i}\sum_{j=1}^{i+1}p_{j}a_{j } \\ & = \left ( p_{i}+\bar{p}_{i}\right ) p_{i+1}a_{i+1}+p_{i}\sum_{j = i+1}^{n-1}p_{j+1}a_{j+1}-\bar{p}% _ { i}\sum_{j=1}^{i+1}p_{j}a_{j } \\ & = p_{i}\sum_{j = i+1}^{n-1}p_{j}a_{j}-\bar{p}_{i}\sum_{j=1}^{i}p_{j}a_{j } \\ & = p_{i}\bar{a}_{i}\left ( \mathbf{p}\right ) -\bar{p}_{i}a_{i}\left ( \mathbf{p% } \right ) \\ & = p_{i}a_{n}\left ( \mathbf{p}\right ) -p_{n}a_{i}\left ( \mathbf{p}\right ) , \end{aligned}\]]and the identity is proved .
we are able now to state and prove the second identity for the ebyev functional .
[ t2.3.16.1]with the assumptions of theorem [ t2.1.16.1 ] , we have the identity@xmath2173 follows by theorem [ t2.1.16.1 ] and lemma [ l2.2.16.1 ] and we omit the details .
the following result holds .
[ t3.1.16.1]let @xmath1274 be an inner product space over the real or complex number field @xmath2174 ; @xmath2142 and @xmath2175 then we have the inequalities@xmath2176all the inequalities in ( [ 3.1.16.1 ] ) are sharp in the sense that the constants @xmath1847 can not be replaced by smaller constants .
using the first identity in ( [ 2.1.16.1 ] ) and schwarz s inequality in @xmath923 i.e. , @xmath2177 @xmath738 we have successively:@xmath2178using hlder s inequality , we deduce the desired result ( [ 3.1.16.1 ] ) .
let us prove , for instance , that the constant 1 in the second inequality is best possible . assume , for @xmath232 we have that@xmath2179for @xmath1391 @xmath1481 @xmath2180 if we choose @xmath1561 then we get@xmath2181also , for @xmath1561@xmath2182and @xmath2183and then , from ( [ 3.2.16.1 ] ) , for @xmath1561 we deduce@xmath2184if in ( [ 3.3.16.1 ] ) we choose @xmath2185 @xmath2186 and @xmath2187 @xmath2188 we deduce @xmath2098 proving that 1 is the best possible constant in that inequality .
the following corollary for the uniform distribution of the probability @xmath2189 holds .
[ c3.2.16.1]with the assumptions of theorem [ t3.1.16.1 ] for @xmath2190 and @xmath2191 we have the inequalities@xmath2192 the following result may be stated as well .
[ t3.3.16.1]with the assumptions of theorem [ t3.1.16.1 ] and if @xmath2193 @xmath1402 then we have the inequalities@xmath2194all the inequalities in ( [ 3.5.16.1 ] ) are sharp in the sense that the constant 1 can not be replaced by a smaller constant . using the second equality in ( [ 2.1.16.1 ] ) and schwarz s inequality , we have@xmath2195using hlder s weighted inequality , we deduce ( [ 3.5.16.1 ] ) .
the sharpness of the constant may be proven in a similar manner to the one in theorem [ t3.1.16.1 ] .
we omit the details .
the following corollary containing the unweighted inequalities holds .
[ c3.4.16.1]with the above assumptions for @xmath2196 and @xmath2197 one has@xmath2198the inequalities ( [ 3.6.16.1 ] ) are sharp in the sense mentioned above .
another type of inequality may be stated if ones used the third identity in ( [ 2.1.16.1 ] ) and hlder s weighted inequality with the weights : @xmath2199 @xmath2200 [ t3.5.16.1]with the assumptions in theorem [ t3.1.16.1 ] and if @xmath2201 @xmath2202 @xmath2167 then we have the inequalities@xmath2203 in particular , if @xmath1976 @xmath35 then we have@xmath2204the inequalities in ( [ 3.7.16.1 ] ) and ( [ 3.8.16.1 ] ) are sharp in the above mentioned sense .
a different approach may be considered if one uses the representation in terms of double sums for the ebyev functional provided by theorem [ t2.3.16.1 ] .
the following result holds .
[ t3.6.16.1]with the above assumptions of theorem [ t3.1.16.1 ] , we have the inequalities@xmath2205the inequalities are sharp in the sense mentioned above .
the proof follows by the identity ( [ 2.7.16.1 ] ) on using hlder s inequality for double sums and we omit the details .
now , define@xmath2206using the elementary inequality@xmath2207we deduce@xmath2208for @xmath2209 consequently , we deduce@xmath2210 we may now state the following corollary of theorem [ t3.6.16.1 ] .
[ c3.7.16.1]with the assumptions of theorem [ t3.1.16.1 ] for @xmath2190 and @xmath2191 we have the inequality@xmath2211the constant @xmath59 can not be replaced in general by a smaller constant .
[ r3.8.16.1]the inequality ( [ 3.10.16.1 ] ) is better than the third inequality in ( [ 1.10.16.1 ] ) .
consider now , for @xmath2212 the number@xmath2213 ^{q}\right ) ^{\frac{1}{q}}.\]]we observe , by symmetry of the terms under the summation symbol , we have that@xmath2214that may be computed exactly if @xmath2215 or another natural number .
since , as above,@xmath2216 ^{q}\leq \frac{1}{4^{q}}\left ( n-\left\vert i - j\right\vert \right ) ^{2q},\]]we deduce@xmath2217 ^{\frac{1}{% q } } \\ & = \frac{1}{4}\left ( n-1\right ) ^{\frac{2}{q}}.\end{aligned}\]]consequently , we may state the following corollary as well .
[ c3.9.16.1]with the assumptions of theorem [ t3.1.16.1 ] for @xmath2190 and @xmath2191 we have the inequalities@xmath2218provided @xmath2219 , @xmath1418 the constant @xmath59 can not be replaced in general by a smaller constant . finally ,
if we denote@xmath2220 , \]]then we observe , for @xmath2221 @xmath2222 that@xmath2223and by theorem [ t3.6.16.1 ] , we deduce the inequality@xmath2224note that , the above inequality has been discovered using a different method in @xcite .
the constant @xmath518 is best possible .
dragomir , a grss type inequality for sequences of vectors in inner product spaces and applications , _ j. inequal .
pure and appl .
, _ * 1*(2 ) ( 2000 ) , article 12 . [ online : ` http://jipam.vu.edu.au/v1n2 ` ] s.s .
dragomir , grss type discrete inequalities in inner product spaces , revisited , _ the mathematics preprint server , _ [ ` online : http://www.mathpreprints.com / math / preprint / sever/20030623.2/1/ ` ] s.s .
dragomir , grss type inequalities for forward difference of vectors in inner product spaces , _ the mathematics preprint server , _
[ ` online : http://www.mathpreprints.com / math / preprint / sever/20030926.1/1/ ` ]
h. iki and t. iki in 2001 , @xcite , by the use of an argument based on orthogonal projections in inner product spaces have observed that ostrowski s inequality may be naturally stated in an abstract setting as follows : [ t.2.17 ] let @xmath848 be a real or complex inner product space and @xmath793 two linearly independent vectors . if @xmath5 is such that@xmath2231then one has the inequality@xmath2232with equality if and only if@xmath2233 in the present section , by the use of elementary arguments and schwarz s inequality in inner product spaces , we show that ostrowski s inequality @xmath2234 holds true for a larger class of elements @xmath2235 the case of equality is analyzed .
applications for complex sequences and integrals are also provided .
[ t.3.17 ] let @xmath848 be a real or complex inner product space and @xmath793 two linearly independent vectors . if @xmath5 is such that @xmath2236then one has the inequality@xmath2237the equality holds in @xmath2238 if and only if@xmath2239where @xmath2240 is such that @xmath2241 if we apply @xmath2245 for @xmath2246where @xmath2247 and @xmath2248 we have@xmath2249with equality iff there is a scalar @xmath2250 such that@xmath2251since simple calculations show that@xmath2252and@xmath2253then , by @xmath2254 , we deduce@xmath2255 \left [ \left\vert d\right\vert ^{2}\left\vert c\right\vert ^{2}-\left\vert \left\langle d , c\right\rangle \right\vert ^{2}\right ] \label{2.7.17 } \\
\geq \left\vert \left\langle z , d\right\rangle \left\vert c\right\vert ^{2}-\left\langle z , c\right\rangle \left\langle c , d\right\rangle \right\vert ^{2 } , \notag\end{gathered}\]]with equality if and only if there is a @xmath2250 such that @xmath2256 holds . if @xmath2257 satisfy @xmath2258 then by @xmath2259 and @xmath2260 for the choices @xmath2261 and @xmath2262 we deduce the inequality @xmath2238 with equality iff there exists a @xmath2263 such that @xmath2239and , by the second condition in @xmath2264@xmath2265since @xmath2266 is clearly equivalent to @xmath2267 , the theorem is completely proved . 1 . if @xmath2268 where @xmath2269 with @xmath2270 linearly independent and@xmath2271then one has the inequality@xmath2272with equality iff @xmath2273 , \ \
i\in \mathbb{n}\]]and @xmath2263 with the property@xmath2274 2 . if @xmath2275 , where @xmath2276 is a measurable space and @xmath2277 with @xmath526 linearly independent and@xmath2278then one has the inequality@xmath2279with equality iff@xmath2280\]]for @xmath2281 a.e .
@xmath2282 and @xmath2283 with@xmath2284 [ t1.2.18]let @xmath2196 , @xmath2285 and @xmath2132 be @xmath33tuples of real numbers with @xmath2226 and@xmath2286then@xmath2287if @xmath2196 and @xmath2285 are not proportional , then the equality holds in ( [ 1.5.18 ] ) iff@xmath2288 ^{\frac{1}{2}}},\ \
k\in \left\ { 1,\dots , n\right\ } , \]]with @xmath2289 in the present section , by the use of an elementary argument based onschwarz s inequality , a natural generalisation in inner - product spaces of ( [ 1.5.18 ] ) is given .
the case of equality is analyzed .
applications for sequences and integrals are also provided .
[ t2.1.18]let @xmath640 be a real or complex inner product space and @xmath793 two linearly independent vectors .
if @xmath5 is such that@xmath2290then@xmath2291the equality holds in ( [ 2.1.18 ] ) iff@xmath2292where @xmath2283 @xmath2293 is such that@xmath2294 ^{\frac{1}{2}}}.\ ] ] we use schwarz s inequality in the inner product space @xmath923 i.e.,@xmath2295with equality iff there is a scalar @xmath16 such that@xmath2296if we apply ( [ 2.4.18 ] ) for @xmath2297 @xmath2298 , where @xmath2247 and @xmath2248 then we deduce the inequality@xmath2255 \left [ \left\vert d\right\vert ^{2}\left\vert c\right\vert ^{2}-\left\vert \left\langle d , c\right\rangle \right\vert ^{2}\right ] \label{2.6.18 } \\
\geq \left\vert \left\langle z , d\right\rangle \left\vert c\right\vert ^{2}-\left\langle z , c\right\rangle \left\langle c , d\right\rangle \right\vert ^{2}\end{gathered}\]]with equality iff there is a @xmath2250 such that@xmath2299if in ( [ 2.6.18 ] ) we choose @xmath2300 @xmath2301 and @xmath2302 where @xmath1905 and @xmath14 statisfy ( i ) , then we deduce@xmath2303 \geq \left [ \left\langle
x , b\right\rangle \left\vert a\right\vert ^{2}\right ] ^{2}\]]which is clearly equivalent to ( [ 2.1.18 ] ) . 1 . if @xmath2196 , @xmath2285 , @xmath2305 @xmath2055 where @xmath2306with @xmath2196 , @xmath2285 linearly independent and@xmath2307then@xmath2308the equality holds in ( [ 2.9.18 ] ) iff@xmath2309 , \ \ \
i\in \left\ { 1,2,\dots \right\}\]]with @xmath2283 such that@xmath2310 ^{\frac{1}{2}}}.\ ] ] 1 .
if @xmath2311 where @xmath2276 is an @xmath2281measurable space and @xmath2312with @xmath526 being linearly independent and@xmath2313then@xmath2314the equality holds in ( [ 2.13.18 ] ) iff@xmath2315 \text { \ for \ a.e .
\ } x\in \omega\]]and @xmath2283 with@xmath2316 ^{\frac{1}{2}}}.\ ] ] s.s .
dragomir and a.c .
goa , a generalisation of an ostrowski inequality in inner product spaces , _ rgmia res .
_ , * 6*(2003 ) , _ no .
2 _ , article 20 . `
[ on line : http://rgmia.vu.edu.au/v6n2.html ] . ` for @xmath2321
we recapture the cauchy - bunyakovsky - schwarz s inequality , i.e. , ( see for example @xcite ) @xmath2322with equality if and only if there exists a real number @xmath562 such that @xmath2323 for each @xmath2324 [ 2.1.19 ] let @xmath2326 and @xmath2327 be two @xmath0-tuples of vectors in @xmath26 then for any @xmath2328 $ ] one has the inequality @xmath2329 ^{2 } \label{a.2.19}\]]@xmath2330 % \left [ \sum_{k=1}^{n}\left\vert y_{k}\right\vert ^{2}+2\alpha \cdot \sum_{1\leq i < j\leq n}\func{re}\left\langle y_{i},y_{j}\right\rangle \right ] .\ ] ] following the proof by p. flor @xcite , we may consider the function @xmath2331 , given by @xmath2332then @xmath2333 t^{2 } \label{a.4.19 } \\ + 2\left [ \left ( 1-\alpha \right ) \cdot
\sum_{k=1}^{n}\func{re}\left\langle x_{k},y_{k}\right\rangle + \alpha \cdot \func{re}\left\langle \sum_{k=1}^{n}x_{k},\sum_{k=1}^{n}y_{k}\right\rangle \right ] t
\\ + \left [ \left ( 1-\alpha \right ) \cdot
\sum_{k=1}^{n}\left\vert y_{k}\right\vert ^{2}+\alpha \cdot \left\vert \sum_{k=1}^{n}y_{k}\right\vert ^{2}\right ] .\end{gathered}\]]observe that @xmath2334and @xmath2335also @xmath2336using @xmath2337 we deduce that @xmath2338 t^{2 } \label{a.8.19 } \\ + 2\left [ \sum_{k=1}^{n}\func{re}\left\langle x_{k},y_{k}\right\rangle + \alpha \cdot \sum_{1\leq i\neq j\leq n}\func{re}\left\langle x_{i},y_{j}\right\rangle \right ] t
\\ + \left [ \sum_{k=1}^{n}\left\vert y_{k}\right\vert ^{2}+2\alpha \cdot \sum_{1\leq i < j\leq
n}\func{re}\left\langle y_{i},y_{j}\right\rangle \right ] .\end{gathered}\]]since , by @xmath2339 @xmath2340 for any @xmath2341 it follows that the discriminant of the quadratic function given by @xmath2342 is negative , which is clearly equivalent with the desired inequality @xmath2343 [ 2.2.19 ] assume that @xmath2344 are orthogonal , i.e. , @xmath2345 for any @xmath2346 @xmath2347 and @xmath2348 are also orthogonal in the real inner product space @xmath2349 then @xmath2350 } \left [ \sum_{k=1}^{n}\left\langle x_{k},y_{k}\right\rangle + \alpha \cdot \sum_{1\leq i\neq j\leq n}\left\langle x_{i},y_{j}\right\rangle \right ] ^{2}\leq \sum_{k=1}^{n}\left\vert x_{k}\right\vert ^{2}\sum_{k=1}^{n}\left\vert y_{k}\right\vert ^{2}.\ ] ] 1 .
if we assume that @xmath2351 with the inner product @xmath2352 then by @xmath2353 we may deduce the following wagner type inequality for complex numbers @xmath2354 ^{2 } \\
\leq \left [ \sum_{k=1}^{n}\left\vert a_{k}\right\vert ^{2}+2\alpha \cdot \sum_{1\leq i < j\leq n}\func{re}\left ( a_{i}\bar{a}_{j}\right ) \right ] \\
\times \left [ \sum_{k=1}^{n}\left\vert b_{k}\right\vert ^{2}+2\alpha \cdot \sum_{1\leq
i < j\leq n}\func{re}\left ( b_{i}\bar{b}_{j}\right ) \right ] , \end{gathered}\]]where @xmath2328 $ ] and @xmath2355 , @xmath2356 .
2 . consider the hilbert space @xmath2357 where @xmath151 is a @xmath406-measurable space and @xmath2358 $ ] is a positive measure on @xmath159 then for @xmath2359 and since the inner product generating the norm is given by @xmath2360we get the inequality @xmath2361 ^{2 } \\
\leq \left [ \sum_{k=1}^{n}\int_{\omega } \left\vert f_{k}\left ( x\right ) \right\vert ^{2}d\mu \left ( x\right ) + 2\alpha \cdot \sum_{1\leq i < j\leq n}\int_{\omega } \func{re}\left ( f_{i}\left ( x\right ) \bar{f}_{j}\left ( x\right ) \right ) d\mu \left ( x\right ) \right ] \\ \times \left [ \sum_{k=1}^{n}\int_{\omega } \left\vert g_{k}\left ( x\right ) \right\vert ^{2}d\mu \left ( x\right ) + 2\alpha \cdot \sum_{1\leq i < j\leq n}\int_{\omega } \func{re}\left ( g_{i}\left ( x\right ) \bar{g}_{j}\left ( x\right ) \right ) d\mu \left ( x\right ) \right ] , \end{gathered}\]]where @xmath2362 and @xmath2328 .$ ] s.s .
dragomir , a generalisation of wagner s inequality for sequences of vectors in inner product spaces , _ rgmia res .
_ , * 6*(2003 ) , supplement , article 3 ` [ on line : http://rgmia.vu.edu.au/v6(e).html ] . `
if @xmath2371 denotes the semi - norm associated to this form and @xmath2372 is an orthonormal family of vectors in @xmath433 , that is , @xmath2373 , then one has @xcite : @xmath2374which is well known in the literature as bessel s inequality .
[ tnew]let @xmath433 be a linear space and @xmath2375 two hermitian forms on @xmath433 such that @xmath2376 is greater than or equal to @xmath2377 , that is , @xmath2378 for all @xmath2379 .
assume that @xmath2380 is an orthonormal family in @xmath2381 and @xmath2382 is an orthonormal family in @xmath2383 such that for any @xmath1068 there exists a finite @xmath2384 such that @xmath2385then one has the inequality : @xmath2386 [ lemnew]let @xmath433 be a linear space endowed with a positive hermitian form @xmath2387 and @xmath2388 @xmath2389 be an orthonormal family in @xmath2390 then @xmath2391for all@xmath2392 @xmath2393and@xmath2370 let @xmath1 be a finite subset of @xmath950 . since @xmath2394 is greater than @xmath2395 , we have : @xmath2396since , by ( [ e2.20 ] ) , we may state that for any @xmath2397 there exists a finite @xmath2384 with @xmath2398we have , for all @xmath2379 @xmath2399applying lemma [ lemnew ] , we can conclude that @xmath2400where @xmath2401consequently , we have @xmath2402for all @xmath2379 and @xmath1 a finite subset of @xmath950 , from which ( [ e3.20 ] ) results . 1 .
let @xmath2405 be an inner product space and @xmath2372 an orthonormal family in @xmath2406 .
assume that @xmath2407 is a linear operator such that @xmath2408 for all @xmath2379 and @xmath2409 for all @xmath2410 .
then one has the inequality @xmath2411 2 .
if @xmath2407 is such that @xmath2412 for all @xmath2379 , then , with the previous assumptions , we also have @xmath2413 3 .
suppose that @xmath2407 is a symmetric positive definite operator with @xmath2414 for all @xmath2379 .
if @xmath2372 is an orthonormal family in @xmath433 such that @xmath2409 for all @xmath2410 , then one has the inequality @xmath2415for any @xmath2370 it is obvious that if @xmath2417 where @xmath1415 are orthornormal vectors in @xmath923 i.e. , @xmath2418 @xmath2419 where @xmath1025 is the kronecker delta , then ( [ 1.1.23 ] ) provides bessel s inequality @xmath2420 we observe that @xmath2422firstly , we have @xmath2423secondly , by the hlder inequality for double sums , we obtain @xmath2424 ^{% \frac{1}{p}}\left ( \sum\limits_{i , j=1}^{n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert ^{q}\right ) ^{\frac{1}{q } } \\ & = \left ( \sum_{i=1}^{n}\left\vert \alpha _ { i}\right\vert ^{p}\sum_{j=1}^{n}\left\vert \alpha _ { j}\right\vert ^{p}\right ) ^{\frac{1}{p}% } \left ( \sum\limits_{i , j=1}^{n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert ^{q}\right ) ^{\frac{1}{q } } \\ & = \left ( \sum\limits_{i=1}^{n}\left\vert \alpha _ { i}\right\vert ^{p}\right ) ^{\frac{2}{p}}\left ( \sum\limits_{i , j=1}^{n}\left\vert \left ( z_{i},z_{j}\right ) \right\vert ^{q}\right ) ^{\frac{1}{q}},\end{aligned}\]]where @xmath1391 @xmath1418 [ r2.9.23]observe , that by the monotonicity of power means , we may write @xmath2442taking the square in both sides , one has @xmath2443giving @xmath2444using ( [ 2.10.23 ] ) and the second inequality in ( [ 2.7.23 ] ) we may deduce the following result @xmath2445for @xmath1550 @xmath1418 to do that , denote @xmath2449and @xmath2450consider the inner product space @xmath602 , @xmath2451 , @xmath1885 and @xmath2452 @xmath2453 then @xmath2454assume that @xmath2455 @xmath2456 , $ ] @xmath2457.$ ] utilizing maple 6 , one may easily see by plotting the function @xmath2458that it has positive and negative values in the box @xmath2459 \times \left [ 1,2\right ] $ ] , showing that the inequalities ( [ 2.11.23 ] ) and ( [ 2.12.23 ] ) can not be compared .
this means that one is not always better than the other .
let @xmath2460 be two functions defined and integrable on @xmath2461 $ ] .
assume that@xmath2462for each @xmath2463 $ ] , where @xmath2464 are given real constants .
then we have the following inequality which is well known in the literature as the grss inequality ( @xcite)@xmath2465 in this inequality , g. grss has proven that the constant @xmath2466 is the best possible in the sense that it can not be replaced by a smaller one , and this is achieved when@xmath2467 [ ta.24]let @xmath2468 be an inner product space over @xmath2469 and @xmath2470 . if @xmath573 are real or complex numbers and @xmath568 are vectors in @xmath1 such that the conditions@xmath2471hold , then we have the inequality@xmath2472the constant @xmath2466 is best possible in sense that it can not be replaced by a smaller constant . [ t3.1.5.24]let @xmath2043 be an inner product space over @xmath2479 and @xmath2480 if @xmath573 are real or complex numbers and @xmath2481 are vectors in @xmath1 such that either the conditions@xmath2482or equivalently , @xmath2483hold , then we have the inequality @xmath2484 ^{\frac{1}{2}}\left [ \func{re}\left\langle \gamma e - y , y-\gamma e\right\rangle \right ] ^{\frac{1}{2 } } \\ & \leq \left ( \frac{1}{4}\left\vert \phi -\varphi \right\vert \cdot \left\vert \gamma -\gamma \right\vert \right ) .\end{aligned}\]]the constant @xmath59 is best possible .
[ tb.24]let @xmath2485 be a family of orthonormal vectors in @xmath1 , @xmath949 a finite part of @xmath950 , @xmath2486 , @xmath966 and @xmath14 a vector in @xmath1 such that either the condition@xmath2487or equivalently,@xmath2488holds , then we have the following reverse of bessel s inequality@xmath2489the constant @xmath2466 is best possible .
[ tc.24]let @xmath917 be a family of orthornormal vectors in @xmath923 @xmath949 a finite part of @xmath1033 @xmath2490 @xmath1042 , and @xmath1839 if either@xmath2491or equivalently,@xmath995hold true , then@xmath2492the constant @xmath59 is best possible in the sense that it can not be replaced by a smaller constant .
the main aim here is to provide some similar inequalities which , giving refinements of the usual grss inequality , are known in the literature as pre - grss type inequalities .
applications for lebesgue integrals in general measure spaces are also given .
[ t1.24]let @xmath2468 be an inner product space over @xmath2493 and @xmath2470 .
if @xmath2494 are real or complex numbers and @xmath568 are vectors in @xmath1 such that either the condition@xmath2495or equivalently,@xmath2496holds true , then we have the inequalities@xmath2497and@xmath2498 it is obvious that:@xmath2499using schwarz s inequality in inner product spaces @xmath2500 for the vectors @xmath2501 and @xmath2502 we deduce:@xmath2503now , the inequality ( [ 12.24 ] ) is a simple consequence of ( [ 4.24 ] ) for @xmath759 , or of lemma [ lb.24 ] and ( [ 11.24 ] ) . since ( see for instance @xcite ) ,
@xmath2504then making use of the elementary inequality @xmath2505 with @xmath2506 we can state that@xmath2507using ( [ 16.24 ] ) and ( [ 17.24 ] ) we have@xmath2508taking into account the inequalities ( [ 15a.24 ] ) and ( [ 18.24 ] ) , we get that@xmath2509finally , using the elementary inequality for positive real numbers : @xmath2510we have : @xmath2511giving the desired inequality ( [ 13.24 ] ) .
[ t2.24]let @xmath2512 be a family of orthonormal vectors in @xmath1 , @xmath949 a finite part of @xmath950 , @xmath2486 , @xmath966 and @xmath568 vectors in @xmath1 such that either the condition@xmath2487or equivalently,@xmath2513holds .
then we have inequalities@xmath2514and@xmath2515 it is obvious ( see for example @xcite ) that:@xmath2516using schwarz s inequality in inner product spaces , we have:@xmath2517 in a similar manner to the one in the proof of theorem [ t1.24 ] we may conclude that ( [ 23.24 ] ) holds true .
let @xmath2520 be a measure space consisting of a set @xmath151 , @xmath149 a @xmath150algebra of parts and @xmath152 a countably additive and positive measure on @xmath149 with values in @xmath2521 .
denote by @xmath2522 the hilbert space of all real or complex valued functions @xmath155 defined on @xmath151 and 2integrable on @xmath151 , i. e. @xmath2523the following proposition holds @xcite .
now , consider the family @xmath1022 of functions in @xmath2537 with the properties that@xmath2538where @xmath1025 is @xmath1026 if @xmath1027 and @xmath1028 if @xmath1029 @xmath1030 is an orthornormal family in @xmath2539 [ p5.1.24]let @xmath1022 be an orthornormal family of functions in @xmath2540 @xmath949 a finite subset of @xmath1033 @xmath1034 @xmath1035 and @xmath2541 such that either@xmath2542 d\mu \left ( s\right ) \geq 0 \label{34.24}\]]or equivalently,@xmath2543holds .
then we have the inequalities@xmath2544and@xmath2545 | the main aim of this monograph is to survey some recent results obtained by the author related to reverses of the schwarz , triangle and bessel inequalities .
some grss type inequalities for orthonormal families of vectors in real or complex inner product spaces are presented as well .
generalizations of the boas - bellman , bombieri , selberg , heilbronn and peari inequalities for finite sequences of vectors that are not necessarily orthogonal are also provided .
two extensions of the celebrated ostrowski s inequalities for sequences or real numbers and the generalization of wagner s inequality in inner product spaces are pointed out .
finally , some grss type inequalities for n - tuples of vectors in inner product spaces and their natural applications for the approximation of the discrete fourier and mellin transforms are given as well . |
soft gamma ray repeaters ( sgrs ) and anomalous x - ray pulsars ( axps ) are a class of compact objects that show interesting observational properties see e.g. mereghetti ( 2008)@xcite : rotational periods in the range @xmath0@xmath1 s , a narrow range with respect to the wide range of ordinary pulsars @xmath2@xmath3 s ; spin - down rates @xmath4@xmath5 , larger than ordinary pulsars @xmath6 ; strong outburst of energies @xmath7@xmath8 erg , and for the case of sgrs , giant flares of even large energies @xmath9@xmath10 erg , not observed in ordinary pulsars .
the observation of sgr 0418 + 5729 with a rotational period of @xmath11 s , an upper limit of the first time derivative of the rotational period @xmath12 rea et .
al.(2010)@xcite , and an x - ray luminosity of @xmath13 erg s@xmath14 can be considered as the rosetta stone for alternative models of sgrs and axps . the magnetar model , based on a neutron star of fiducial parameters @xmath15 , @xmath16 km and a moment of inertia @xmath17 g @xmath18 , needs a magnetic field larger than the critical field for vacuum polarization @xmath19 g in order to explain the observed x - ray luminosity in terms of the release of magnetic energy see @xcite for details .
the inferred upper limit of the surface magnetic field of sgr 0418 + 5729 @xmath20 g describing it as a neutron star see rea et .
al.@xcite for details , is well below the critical field , which has challenged the power mechanism based on magnetic field decay in the magnetar scenario .
alternatively , it has been recently pointed out how the pioneering works of morini et .
al.@xcite and paczynski@xcite on the description of 1e 2259 + 586 as a white dwarf ( wd ) can be indeed extended to all sgrs and axps .
these wds were assumed to have fiducial parameters @xmath21 , @xmath22 km , @xmath23 g @xmath18 , and magnetic fields @xmath24 g see @xcite for details inferred from the observed rotation periods and spin - down rates .
the energetics of sgrs and axps including their steady emission , glitches , and their subsequent outburst activities have been shown to be powered by the rotational energy of the wd @xcite .
the occurrence of a glitch , the associated sudden shortening of the period , as well as the corresponding gain of rotational energy , can be explained by the release of gravitational energy associated with a sudden contraction and decrease of the moment of inertia of the uniformly rotating wd , consistent with the conservation of their angular momentum . by describing sgr 0418 + 5729 as a wd , we calculated @xcite an upper limit for the magnetic field @xmath25 g and
show that the x - ray luminosity observed from sgr 0418 + 5729 can be well explained as originating from the loss of rotational energy of the wd leading to a theoretical prediction for the spin - down rate @xmath26 where the lower limit was established by assuming that the observed x - ray luminosity of sgr 0418 + 5729 coincides with the rotational energy loss of the wd .
as we will show below , these predictions can be still improved by considering realistic wd parameters @xcite instead of fiducial ones .
the situation has become even more striking considering the recent x - ray timing monitoring with swift , rxte , suzaku , and xmm - newton satellites of the recently discovered sgr swift j1822.31606 @xcite .
the rotation period @xmath27 s , and the spin - down rate @xmath28 have been obtained . assuming a ns of fiducial parameters , a magnetic field @xmath29 g is inferred , which is again in contradiction with a magnetar explanation for this source .
we have recently computed in @xcite general relativistic configurations of uniformly rotating white dwarfs within hartle s formalism @xcite .
we have used the relativistic feynman - metropolis - teller equation of state @xcite for wd matter , which we have shown generalizes the traditionally used equation of state of salpeter@xcite .
it has been there shown that rotating wds can be stable up to rotation periods of @xmath30 s ( see @xcite and sec .
[ sec:3 ] for details ) .
this range of stable rotation periods for wds amply covers the observed rotation rates of sgrs and axps @xmath0@xmath1 s. the aim of this article is to describe one of the so - called _ low magnetic field magnetars _ , sgr 0418 + 5729 as a massive fast rotating highly magnetized wd . in doing
so we extend the work of malheiro et .
al.@xcite by using precise wd parameters recently obtained by boshkayev et .
al.@xcite for general relativistic uniformly rotating wds .
the loss of rotational energy associated with the spin - down of the wd is given by @xmath31 where @xmath32 is the moment of inertia of the wd in units of @xmath33 g @xmath18 .
this rotational energy loss amply justifies the steady x - ray emission of all sgrs and axps see @xcite for details .
the upper limit on the magnetic field see e.g. @xcite obtained by requesting that the rotational energy loss due to the dipole field be smaller than the electromagnetic emission of the magnetic dipole , is given by @xmath34 where @xmath35 is the mean radius of the wd in units of @xmath36 cm . the mean radius
is given by @xmath37 see e.g. @xcite with @xmath38 and @xmath39 the equatorial and polar radius of the star .
it is clear that the specific values of the rotational energy loss and the magnetic field depend on observed parameters , such as @xmath40 and @xmath41 , as well as on model parameters , such as the mass , moment of inertia , and mean radius of the rotating wd .
the rotational stability of fast rotating wds was implicitly assumed by malheiro et .
al.@xcite . the crucial question of whether rotating wds can or can not attain rotation periods as short as the ones observed in sgrs and axps has been recently addressed by boshkayev et .
the properties of uniformly rotating wds were computed within the framework of general relativity through hartle s formalism @xcite .
the equation of state for cold wd matter is based on the relativistic feynman - metropolis - teller treatment @xcite , which generalizes the equation of state of salpeter@xcite .
the stability of rotating wds was analyzed taking into account the mass - shedding limit , inverse @xmath42-decay instability , and secular axisymmetric instability , with the latter determined by the turning point method of friedman et .
al.@xcite ; see fig .
[ fig : consomj ] and @xcite , for details .
= constant sequences , where the static case @xmath43 the thickest one .
the color thin - dashed curves correspond to @xmath44=constant sequences .
the keplerian sequence is the red thick dashed curve , the blue thick dotted - dashed curve is the inverse @xmath42 instability boundary , and the green thick dotted curve is the axisymmetric instability line .
the gray - shaded region is the stability region of rotating white dwarfs @xcite .
the numbers show the period @xmath40 in seconds . ]
the minimum rotation period @xmath45 of wds is obtained for a configuration rotating along the keplerian ( the mass shedding ) sequence , at the critical inverse @xmath42-decay density , namely this is the configuration lying at the crossing point between the mass - shedding and inverse @xmath42-decay boundaries .
the numerical values of the minimum rotation period @xmath46 s and the maximum masses @xmath47 were found for helium ( he ) , carbon ( c ) , oxygen ( o ) , and iron ( fe ) wds , respectively ( see boshkayev et .
al.@xcite , for details ) . as a byproduct ,
these values show that indeed all sgrs and axps can be described as rotating wds because their rotation periods are in the range @xmath48 s. the relatively long minimum period of rotating fe wds , @xmath49 s , lying just at the lower edge of the observed range of rotation periods of sgrs and axps , reveals crucial information on the chemical composition of sgrs and axps , namely they are very likely made of elements lighter than fe , such as c or o. it can be seen from fig .
[ fig : consomj ] that every @xmath44=constant or equivalently @xmath40=constant sequence intersects the stability region of general relativistic uniformly rotating wds ( @xmath50-@xmath38 curves inside the shaded region of fig .
[ fig : consomj ] ) in two points .
these two points determine the minimum(maximum ) and maximum(minimum ) @xmath51(@xmath52 ) , respectively , for the stability of a wd with the given rotation angular velocity @xmath53 .
associated with the boundary values @xmath51 and @xmath52 , we can obtain the corresponding bounds for the moment of inertia of the wd , @xmath54 , respectively .
we turn now to a specific analysis of sgr 0418 + 5729 .
sgr 0418 + 5729 has a rotational period of @xmath11 s , and the upper limit of the spin - down rate @xmath12 was obtained by rea et .
the corresponding rotation angular velocity of the source is @xmath55 rad s@xmath14 .
we show in table [ tab : sgr0418 ] bounds for the mass , equatorial radius , mean radius , and moment of inertia of sgr 0418 + 5729 obtained by the request of the rotational stability of the rotating wd , as described in section [ sec:4 ] , for selected chemical compositions . hereafter we shall consider general relativistic rotating carbon wds .
[ tab : sgr0418 ] introducing the values of @xmath40 and the upper limit @xmath41 into eq .
( [ eq : edot ] ) we obtain an upper limit for the rotational energy loss @xmath56 which for any possible mass is larger than the observed x - ray luminosity of sgr 0418 + 5729 , @xmath13 erg s@xmath14 , assuming a distance of 2 kpc @xcite . the corresponding upper limits on the surface magnetic field of sgr 0418 + 5729 , obtained from eq . ( [ eq : bmax ] )
are ( see also table [ tab : sgr0418 ] ) @xmath57 it is worth noting that the above maximum possible value of the surface magnetic field of sgr 0418 + 5729 obtained for the maximum possible mass of a wd with rotation period @xmath58 s , @xmath59 g , is even more stringent and improves the previously value given by @xcite , @xmath60 g , based on fiducial wd parameters .
the electron cyclotron frequency expected from such a magnetic field is @xmath61 corresponding to wavelengths 9.04 and 0.31 @xmath62 m , respectively .
assuming that the observed x - ray luminosity of sgr 0418 + 5729 equals the rotational energy loss @xmath63 , we obtain the lower limit for the spin - down rate @xmath64 which in the case of the wd with the maximum possible mass is more stringent than the value reported by malheiro et .
@xcite , @xmath65 , for a massive wd of fiducial parameters .
the recent observations of sgr 0418 + 5729 @xcite , @xmath11 and @xmath12 challenge the description of these sources as ultramagnetized nss of the magnetar model of sgrs and axps .
based on the recent work of malheiro et .
al.@xcite , we have shown here that , instead , sgr 0418 + 5729 is in full agreement with a description based on massive fast rotating highly magnetic wds . from the analysis of the rotational stability of the wd using the results of boshkayev et .
@xcite , we have predicted the wd parameters . in particular , bounds for the mass , radius , moment of inertia , and magnetic field of sgr 0418 + 5729 . we have improved the theoretical prediction of the lower limit for the spin - down rate of sgr 048 + 5729 , for which only the upper limit , @xmath12 rea et .
( 2010)@xcite , is currently known .
based on a wd of fiducial parameters , malheiro et .
al.@xcite predicted for sgr 0418 + 5729 the lower limit @xmath66 .
our present analysis based on realistic general relativistic rotating wds allows us to improve such a prediction , see eq .
( [ eq : pdotminsgr0418 ] ) for the new numerical values .
we have given in eq .
( [ eq : fcycsgr0418 ] ) an additional prediction of the electron cyclotron frequencies of sgr 0418 + 5729 .
the range we have obtained for such frequencies fall into the optical and infrared bands .
0 m. malheiro , j. a. rueda , & r. ruffini , _ publ .
. japan _ 64 , 56 ( 2012 ) .
s. mereghetti , _ _ , 15 , 225 ( 2008 ) .
k. boshkayev , j. a. rueda , r. ruffini & i. siutsou _ _ , 762 , 14 ( 2013 ) ; arxiv:1204.2070 .
r. c. duncan & c. thompson , _ _ , 392 , l9 ( 1992 ) . c. thompson , & r. c. duncan , _ _ , 275 , 255 ( 1995 ) .
a. ferrari & r. ruffini , _ _ , 158 , l71 ( 1969 ) .
j. l. friedman , j. r. ipser & r. d. sorkin , _ _ , 325 , 722 ( 1988 ) .
j. b. hartle , _ _ , 150 , 1005 ( 1967 ) .
j. b. hartle , & k. s. thorne , _ _ , 153 , 807 ( 1968 ) .
m. morini , n. r. robba , a. smith , & m. van der klis , _ _ , 333 , 777 ( 1988 ) .
b. paczynski , _ _ , 365 , l9 ( 1990 ) .
n. rea , p. esposito , r. turolla , et al .
, _ science _
, 330 , 944 ( 2010 ) . n. rea , g. l. israel , p. esposito , et al . _ in press _ ( 2012 ) ; arxiv:1203.6449 .
m. rotondo , j. a. rueda , r. ruffini , & s .- s .
xue , _ _ , 83 , 045805 ( 2011 ) .
e. e. salpeter , _ _ , 134 , 669 ( 1961 ) . | we describe one of the so - called _ low magnetic field magnetars _ sgr 0418 + 5729 , as a massive fast rotating highly magnetized white dwarf following malheiro et .
al.@xcite .
we give bounds for the mass , radius , moment of inertia , and magnetic field for these sources , by requesting the stability of realistic general relativistic uniformly rotating configurations . based on these parameters ,
we improve the theoretical prediction of the lower limit of the spin - down rate of sgr 0418 + 5729 . in addition , we compute the electron cyclotron frequencies corresponding to the predicted surface magnetic fields . |
the specifications of the receiver system are summarized in table 4 and a block diagram of the receiver system is shown in figure 6 .
the operation frequency range is from 72 to 128 ghz , which is limited by the frequency range of the lo chain .
the if frequency range of 4.08.0 ghz with the bandwidth of 4.0 ghz is adopted .
this receiver is composed of an omt and two 2sb mixers , which are both based on a waveguide technique . the 2sb mixers are hereafter referred to as mixer - a and mixer - b .
the photograph of the receiver in the dewar is shown in figure 7 .
( 160mm,118mm)figure6.eps ( 80mm,60mm)figure7.eps by using an omt , it is possible to receive signals in the dual linear polarizations independently and simultaneously .
the omt adopted herein was developed at the australia telescope national facility ( atnf ) , commonwealth scientific and industrial research organization ( csiro ) , and consists of a square to double ridged guide transition followed by a junction of two side arms with the central guide ( moorey et al .
measurement of the omt performance by atnf has revealed a return loss of @xmath4 20 db from 71 to 118 ghz and an insertion loss at room temperature of @xmath9 0.16 db for both polarizations .
( 80mm,80mm)figure8.eps the detailed structure of a split - block waveguide unit for our 2sb mixer is described in asayama et al .
the basic design of the present sideband - separating sis mixer is similar to that described in claude et al .
the split - block waveguide unit contains an rf quadrature hybrid , two lo directional couplers , an lo power divider , and 4 k cold image terminations .
we also integrated two dsb mixers on the split - block waveguide unit through the waveguide taper transformers .
the rf and lo signals are fed to the feed point through a linearly tapered waveguide impedance transformer , which uses a full height to 1/5 reduced height waveguide for waveguide - to - stripline transition of the sis mixer .
the linearly tapered waveguide impedance transformer was designed using the lumped - gap - source port provided by hfss@xmath10 ( asayama et al .
the sis junctions adopted herein were developed at the nro .
a six - series array was composed of nb / alo@xmath11/nb junctions with a normal state resistance of approximately 8090 @xmath12 , and each junction area was 1.9 @xmath13 1.9 @xmath6m@xmath14 .
the reason for using the series junction is that a wider bandwidth of rf frequency can be achieved . in the present system ,
the required fractional bandwidth totals @xmath15 40% . the parallel connected twin junction ( pctj ) designed by asayama et al .
( 2004 ) is not suited for this receiver because the designed rf frequency range is 90115 ghz , which corresponds to a fractional bandwidth of only approximately 25% .
moreover , the series junction barely saturates and the intensity can be calibrated with high accuracy ( kerr 2002 ) .
the if signals from the two dsb mixers are combined in a commercial quadrature hybrid ( nihon tsushinki inc . ) .
ll tuning range & 72128 ghz + mixer type & sis ( nb / alo@xmath11/nb ) six - series junction + & mounted on fixed - tuned waveguide - type 2sb mixer block + receiving mode & ssb ( 2sb ) operation + polarization & dual linear polarization + receiver noise temperature ( dsb ) & @xmath7 50 k ( @xmath16 = 80115 ghz ) + & + & + receiver noise temperature ( ssb ) & @xmath17 100 k ( @xmath18 = 75120 ghz ) + & @xmath7 68 k ( minimum value at @xmath18 = 90 ghz in the usb ) + & lsb : 86.8 k ( average value at @xmath18 = 75106 ghz ) + & usb : 88.2 k ( average value at @xmath18 = 79120 ghz ) + image rejection ratio & @xmath4 10 db ( @xmath18 = 80123 ghz ) + & lsb : 14.9 db ( average value at @xmath18 = 80111 ghz ) + & usb : 14.3 db ( average value at @xmath18 = 84123 ghz ) + & + & + receiver noise temperature ( ssb ) & @xmath17 100 k ( @xmath18 = 75120 ghz ) + & @xmath7 49 k ( minimum value at @xmath18 = 95 ghz in the lsb ) + & lsb : 74.5 k ( average value at @xmath18 = 75106 ghz ) + & usb : 72.7 k ( average value at @xmath18 = 79120 ghz ) + image rejection ratio & @xmath4 10 db ( @xmath18 = 80100 ghz ) + & lsb : 13.4 db ( average value at @xmath18 = 80100 ghz ) + & usb : 12.2 db ( average value at @xmath18 = 84110 ghz ) + & + if frequency band & 4.08.0 ghz ( 4.0 ghz bandwidth ) + if amplifier & cooled low noise hemt amplifier + & ( typical noise temperature of 8 k and gain of @xmath1930 db ) + before installing the receiver system in the telescope , we evaluated its performance
. the noise temperature of the 2sb sis receiver was measured by a standard y - factor method using hot ( 300 k ) and cold ( 77 k ) loads in the laboratory .
the mixer was mounted on a 4 k cold stage in a dewar .
the first - stage if amplifier is a 4 k cooled hemt at the 4.08.0 ghz band .
the equivalent noise temperature and the gain of the hemt amplifier associated with an isolator were approximately 8 k and + 30 db , respectively .
the following - stage amplifiers work at room temperature . the measured dsb receiver noise temperatures of the sis mixers by 4.08.0 ghz if by a power meter are approximately 50 k over the lo frequency range of 80115 ghz .
the ssb receiver noise temperature , including the noise contribution from the vacuum window , the feed horn , and the if amplifier chain , were measured by a spectrum analyzer . for a certain lo frequency ,
the output spectra were smoothed to a resolution of 500 mhz and sampled at eight if frequency points from 4.25 ghz to 7.75 ghz at intervals of 0.5 ghz .
we then swept the lo frequency from 80 ghz to 120 ghz by a step of 5 ghz to cover the rf frequency range from 72.25127.75 ghz .
finally , the measured ssb receiver noise temperatures were corrected for the irr , in order to obtain the ssb receiver noise temperature .
since the irrs were measured with the different measurement system from the one used to measure the noise temperature , we assumed the irrs to be 10 db , that is nearly equal to the smallest value among the actual irrs ( see section 3.3 ) , for all the measured points .
therefore , it is to be noted that the actual ssb receiver noise temperature will be better than the values presented in this paper for most of the measured points .
the ssb receiver noise temperatures were measured to be lower than approximately 100 k over the rf range of 75120 ghz , and the minimum value of @xmath768 k for mixer - a in the usb and @xmath749 k for mixer - b in the lsb are achieved at approximately 90 ghz and 95 ghz , respectively .
the mean value and standard deviation between the rf frequency range of 75106 ghz in the lsb and 79120 ghz in the usb are 87@xmath2011 k and 88@xmath2015 k , respectively , for mixer - a ( figure 8(a ) ) and are 74@xmath2014 k and 73@xmath2010 k , respectively , for mixer - b ( figure 8(b ) ) .
( 80mm,80mm)figure9.eps the irrs were measured by the relative amplitudes of the if responses in the usb and lsb when injecting a continuous wave ( cw ) signal ( kerr et al .
2001 ) from a signal generator .
the irr measured by 4.08.0 ghz if was greater than 10 db over the rf frequency range of 80123 ghz for mixer - a ( figure 9(a ) ) .
the mean value between the rf frequency range of 80111 ghz in the lsb and 84123 ghz in the usb are 15 db and 14 db , respectively .
the irr of mixer - b was lower than 10 db at @xmath21105 ghz ( figure 9(b ) ) .
however , both the mean value between the rf frequency range of 80102 ghz in the lsb and that between the rf frequency range of 84110 ghz in the usb were 12 db .
therefore , it is most desirable to use mixer - b for observation at lower rf frequency .
there is a possibility in the future that the mixer will be replaced by a new mixer with better performance .
( 80mm,107mm)figure10.eps the receiver system was installed in the 45-m telescope in early december 2007 .
figure 10 shows the receiver dewar in the receiver cabin .
the sis and hemt amplifier biases , lo oscillators , and if chains are placed around the dewar .
the lo signals for each 2sb mixer of dual - polarization are independently generated by multiplying the output of the signal generators ( @xmath71220 ghz ) with 2@xmath133 multipliers .
therefore , the four if signals from two 2sb mixers can cover different regions of the rf frequency band .
hereafter , the polarization for mixer - a is referred to as pol-1 , and the polarization for mixer - b is referred to as pol-2 .
the performance of the receiver may change when it is installed in the telescope because the environment of the receiver is different from that in the laboratory .
therefore , we installed the receiver system in the telescope and measured the performance of the receiver system .
note that the noise temperatures in this subsection are not corrected for irr . even after correcting for irr ,
the noise temperature increases only @xmath17 10% .
the ssb receiver noise temperature , including the ellipsoidal mirror before the horn , is measured by a standard y - factor method to be approximately 60 k in the usb at @xmath16 = 109 ghz ( @xmath18 = 115 ghz ) for both 2sb mixers .
the ssb noise temperatures of the system , including the atmosphere , are approximately 180 k in the lsb at @xmath16 = 109 ghz ( @xmath18 = 103 ghz ) for pol-1 and at @xmath16 = 104 ghz ( @xmath18 = 98 ghz ) for pol-2 at an elevation of 80@xmath22 . the system noise temperature , including the atmosphere , became approximately half of that of the previous receiver system ( s100 ) .
we estimated the beam size and the main - beam efficiency of the 45-m telescope based on observations of the quasar , 3c273 , and the saturn in continuum .
note that the following data is obtained at the beginning of the commissioning of the new receiver system , and thus the results are preliminary .
we scanned 3c273 in the azimuth and elevation directions , while recording the total power of the if output .
the beam pattern of the antenna is obtained from the resultant map of 3c273 by rotating 180@xmath22 along the peak position , because 3c273 can be assumed as a point source .
figure 11 shows the resultant map of 3c273 at 115 ghz .
the half power beam width ( hpbw ) of the telescope is estimated to be 15.@[email protected] for pol-1 and 17.@[email protected] for pol-2 assuming that the beam pattern has a gaussian shape in both polarizations .
the main - beam efficiency , @xmath24 , was calculated to be 34 % and 35 % for pol-1 and pol-2 , respectively , assuming that the brightness distribution of saturn as a uniform disk of apparent diameter of 18@xmath23 with a brightness temperature of 150 k ( mangum 1993 ) .
( 80mm,40mm)figure11.eps the first astronomical signal after the installation , @xmath0co ( @xmath25 = 10 ) spectra at 115.271 ghz in the usb from the w51 giant molecular cloud , was obtained on december 11 , 2007 ( figure 12(a ) and ( b ) ) .
we also obtained hco@xmath3 ( @xmath25 = 10 ) spectra at 89.182 ghz in the lsb from the w51 on december 17 , 2007 ( figure 12(c ) and ( d ) ) .
the irrs of the two 2sb mixers were calculated from these spectra , resulting in @xmath4 20 db in both the usb at @xmath18 @xmath7115 ghz and the lsb at @xmath18 @xmath789 ghz for pol-1 ( figure 12 ( a ) and ( c ) ) and @xmath714 db in the usb at @xmath18 @xmath7115 ghz and @xmath712 db in the lsb at @xmath18 @xmath789 ghz for pol-2 ( figure 12 ( b ) and ( d ) ) .
these values are as good as those obtained in the laboratory .
moreover , we successfully observed the molecular lines toward the sagittarius b2 region by using six digital spectrometers ( sorai et al .
2000 ) with bandwidths of 512 mhz .
figure 13 shows an example of the results , which are obtained by a single pointing .
we detected ch@xmath26c@xmath27h and h@xmath27cs in the lsb and @xmath0co in the usb , respectively , for pol-1 , and cs in the lsb and so@xmath27 , c@xmath2o , hnco , @xmath1co and ch@xmath26cn in the usb , respectively , for pol-2 .
since four if signals can be observed independently and simultaneously , we obtained these spectra by integrating the signal only for five minutes .
this was the first astronomical observation using the waveguide - type dual - polarization sideband - separating sis receiver system in the 100 ghz band .
( 80mm,80mm)figure12.eps ( 80mm,135mm)figure13.eps
a waveguide - type dual - polarization sideband - separating sis receiver system in the 100 ghz band was developed for and installed in the nro 45-m telescope .
this receiver system is composed of an ortho - mode transducer and two sideband - separating sis mixers , both of which are based on the waveguide type .
the four if bands , 4.08.0 ghz , in each of the polarizations and sidebands can be observed simultaneously and separately . over the rf frequency range of 80120 ghz , the single - sideband receiver noise temperatures are 50100 k and the image rejection ratios are greater than 10 db .
we have developed new optics for the receiver and new if chains for the four if signals obtained by the receiver system .
the new receiver system was installed in the 45-m telescope , and we successfully observed the @xmath0co , @xmath1co , and c@xmath2o emission lines simultaneously toward the sagittarius b2 region .
this is the first astronomical observation using a waveguide - type dual - polarization sideband - separating sis receiver system in the 100 ghz band .
the ssb noise temperatures of the system , including the atmosphere , are approximately 180 k in the lsb at @xmath16 = 109 ghz ( @xmath18 = 103 ghz ) for one polarization , and at @xmath16 = 104 ghz ( @xmath18 = 98 ghz ) for the other polarization at an elevation of 80@xmath22 .
the ssb system noise temperature , are measured to be @xmath7180 k at @xmath18 @xmath7100 ghz for both polarizations , and therefore became approximately half of that of the previous receiver system .
the irrs of the two 2sb mixers were calculated from the @xmath0co ( @xmath25 = 10 ) spectra at 115.271 ghz in the usb and hco@xmath3 ( @xmath25 = 10 ) spectra at 89.182 ghz in the lsb from the w51 giant molecular cloud , resulting in @xmath4 20 db for one polarization and @xmath4 12 db for the other polarization .
the authors would like to thank takafumi kojima and yasuhiro abe for their contributions to this project .
we are also grateful to chieko miyazawa , akira mori , and the entire staff of the nobeyama radio observatory for their useful discussions and support .
this study was supported in part by a grant - in - aid for scientific research on priority areas ( 15071205 ) from mext . | we developed a waveguide - type dual - polarization sideband - separating sis receiver system of the 100-ghz band for the 45-m radio telescope at the nobeyama radio observatory , japan .
this receiver is composed of an ortho - mode transducer and two sideband - separating sis mixers , which are both based on the waveguide technique . the receiver has four intermediate frequency bands of 4.08.0 ghz . over the radio frequency range of 80120 ghz , the single - sideband receiver noise temperatures are 50100 k and the image rejection ratios are greater than 10 db .
we developed new matching optics for the telescope beam as well as new if chains for the four if signals .
the new receiver system was installed in the telescope , and we successfully observed the @xmath0co , @xmath1co and c@xmath2o emission lines simultaneously toward the sagittarius b2 region to confirm the performance of the receiver system .
the ssb noise temperature of the system , including the atmosphere , became approximately half of that of the previous receiver system .
the image rejection ratios ( irrs ) of the two 2sb mixers were calculated from the @xmath0co and hco@xmath3 spectra from the w51 giant molecular cloud , resulting in @xmath4 20 db for one polarization and @xmath4 12 db for the other polarization .
( 80mm,60mm)figure1.eps .physical parameters of the optics [ cols="<,>",options="header " , ] we designed the new ellipsoidal mirror with an edge level of @xmath554 db so that the spillover loss would be small . in this design ,
more than 99% of the total power can be transmitted .
we fabricated the ellipsoidal mirror using an nc milling cutter and checked the surface error using a 3d vision measurement system .
as a result , a surface error of 3.8 @xmath6 m r.m.s . is achieved , which is sufficiently small at the observation wavelength of @xmath73 mm .
this mirror is fixed on the side of a receiver dewar by a holder having a positional adjustment system .
( 80mm,31mm)figure4.eps ( 80mm,65mm)figure5.eps next , we used the technique of physical optics for the evaluation of the optical system obtained by gaussian optics . in this calculation
, we used the grasp9 analysis software package of the reflector antenna ( ticra ) . as a result , an antenna directivity of 92.15 dbi , which is equivalent to an aperture efficiency @xmath8 of 0.74 for the 45-m telescope , a beam size of 15.8 arcsec , and the first side lobe level of @xmath517 db were obtained ( figure 3 )
however , the actual performance may be somewhat worse by several percent because the calculation was made without blocking by the sub - reflector stays , surface error , or ohmic loss of the optical elements .
corrugated horns are commonly used with reflector antenna systems . since
corrugated horns can reduce the edge diffraction , improved pattern symmetry and reduced cross polarization can be obtained ( clarricoats et al . 1984 ) .
such performance is needed for the horn used for the wide range of the rf frequency .
we repeatedly calculated the suitable corrugation pattern from the basic design .
the physical dimensions of the corrugated horn are shown in table 3 and figure 4 . as a result
, we obtained a return loss lower than @xmath523 db , a maximum cross polarization level lower than @xmath530 db , and good similarity between the calculated beam profiles of the e - plane and h - plane ( figure 5 ) .
we fabricated the horn by the directly - dig method , rather than by the electro - forming method ( kimura et al .
2008 ) . |
the finite volume element ( fve ) method has been widely used in numerically solving partial deferential equations .
the main feature of fve method is that it inherits some physical conservation laws of original problems locally , which are very desirable in practical applications . during the last decades
, many research works have been presented for fve methods solving various partial differential equations , see @xcite and the references cited therein .
superconvergence of numerical solutions has been an active research area for finite element method ( fem ) since its practical importance in enhancing the accuracy of finite element approximation .
but the study of superconvergence properties of fve methods is far behind that of fems . for elliptic problems in two - dimensional domain , the early superconvergence results of linear and bilinear fve solutions are of this form @xcite @xmath4 where @xmath3 is the set of optimal stress points of interpolation function on partition @xmath5
, @xmath6 is the total number of points in @xmath3 , and @xmath7 denotes the average gradient on elements containing point @xmath8 .
later , lv and li in @xcite extended result ( [ 1.1 ] ) to the isoparametric bilinear fve on quadrilateral meshes under the @xmath9-uniform mesh condition .
recently , zhang and zou in @xcite also derived some superconvergence results for the bi - complete @xmath10-order fve on rectangular meshes , and in the case of bilinear fve ( @xmath11 ) , their result is @xmath12 where @xmath13 is the gauss point of element ( the midpoint of element ) .
moreover , by using the postprocessing technique , chou and ye @xcite obtain the superconvergence estimate : @xmath14 where @xmath15 is the postprocessed linear fve solution obtained by the @xmath16-projection method ; zhang and sheng @xcite further derive the superconvergence estimate : @xmath17 where @xmath18 is the reconstructed gradient of bilinear fve solution obtained by using the patch interpolation recovery method . in this paper
, we consider the bilinear fve method to solve the following problem @xmath19 where @xmath20 is a rectangular domain with boundary @xmath21 , coefficient matrix @xmath22 .
our main goal is to give some piecewise - point gradient superconvergence for the bilinear fve approximation to problem ( [ 1.5 ] ) . to the authors best knowledge , in existing literatures ,
only the midpoints of elements are proved to be the superconvergence point of gradient approximation @xcite ( also see ( [ 1.2 ] ) ) , we here will prove that except the midpoints of elements , all interior mesh points and midpoints of interior edges are also the superconvergence points .
generally speaking , the analysis of bilinear fve on rectangular meshes is more difficult then that of linear fve on triangle meshes , the reason is that @xmath23 is not constant in the former case . by calculating exactly some integrals on element and its boundary ,
we first establish the superclose weak estimate for the bilinear form of the fve method , @xmath24 where @xmath25 is the bilinear interpolation operator , @xmath26 is the projection operator from the trial function space @xmath27 to the test function space
. it is well known that such weak estimate plays an important role in the superconvergence analysis of fems @xcite . by means of this weak estimate and some integral estimates on element ,
we further derive the following superconvergence result for the gradient approximation @xmath28 where @xmath3 is the optimal stress point set of bilinear interpolation function which is composed of the interior mesh points , the midpoints of interior edges and the midpoints of elements . this paper is organized as follows . in section 2 , we introduce the fve method and give some lemmas . in section 3 ,
we establish the superconvergence weak estimate .
section 4 is devoted to the piecewise - point gradient supconvergence estimate on the set @xmath3 .
finally , in section 5 , numerical experiments are provided to illustrate our theoretical analysis . throughout this paper ,
we adopt the notations @xmath29 to indicate the usual sobolev spaces on domain @xmath30 equipped with the norm @xmath31 and semi - norm @xmath32 , and if @xmath33 , we set @xmath34 , @xmath35 .
when @xmath36 , we omit the index @xmath37 . we will use letter @xmath38 to represent a generic positive constant , independent of the mesh size @xmath39 .
consider problem ( [ 1.5 ] ) . as usual , we assume that there exist positive constants @xmath40 and @xmath41 such that @xmath42 we further assume that @xmath43^{2\times 2}$ ] , @xmath44 and @xmath45 . let @xmath46 be a rectangular partition of domain @xmath47 so that @xmath48 , where @xmath49 , @xmath50 is the diameter of element @xmath51 .
we assume that partition @xmath5 is regular , that is , there exists a positive constant @xmath52 such that @xmath53 where @xmath54 denotes the diameter of the biggest ball included in @xmath51 .
associated with partition @xmath5 , we construct the central dual partition @xmath55 by connecting the center of each element to the midpoints of edges by straight lines .
thus , for each nodal point @xmath2 in @xmath5 , there exists a rectangle @xmath56 surrounding @xmath2 , @xmath57 is called the dual element or the control volume at point @xmath2 , see fig.1 .
+ to partition @xmath5 and @xmath58 , we associate the following trial function space @xmath27 and test function space @xmath59 , respectively , @xmath60 where @xmath61 is the set of all bilinear polynomials on @xmath51 and @xmath62 is the set of all nodal points of @xmath5 . the conventional weak form for problem ( [ 1.5 ] ) is that find @xmath63 such that @xmath64 where @xmath65 weak form ( [ 2.3 ] ) usually is adopted for the finite element method .
but for the fve method , we need a new weak form .
let @xmath66 be the solution of problem ( [ 1.5 ] ) , then by using green s formula , we have @xmath67 where @xmath68 is the outward unit normal vector on the boundary concerned .
motivated by this weak formula , we introduce the following bilinear form @xmath69 and define the fve approximation of problem ( [ 1.5 ] ) by finding @xmath70 such that @xmath71 let @xmath72 be the interpolation operator defined by @xmath73 where @xmath74 is the characteristic function of the dual element @xmath56 . since @xmath26 is a one to one mapping from @xmath27 onto @xmath59 , then the equivalent form of problem ( [ 2.7 ] ) is that find @xmath70 such that @xmath75 this is the fve scheme to be used in our analysis . from ( [ 2.5 ] ) we know that scheme ( [ 2.8 ] ) is consistent and the following error equation holds .
@xmath76 let @xmath77 be the usual bilinear interpolation function of continuous function @xmath66 . in our analysis , the following approximation property , trace inequality and inverse inequality will be used frequently . for @xmath78 , we have @xmath79 furthermore , the following two lemmas hold @xcite . +
* lemma 2.1 * @xmath80 * lemma 2.2*_for @xmath39 small , we have _
@xmath81 lemma 2.2 implies that the solution @xmath82 of problem ( [ 2.8 ] ) uniquely exists .
it is well known that the interpolation weak estimate plays an important role in the superconvergence analysis @xcite . to fem defined on regular rectangular meshes ,
the following interpolation weak estimate has been established @xcite @xmath83 in this section , we will give a similar weak estimate for the fve method , which is very useful in our superconvergence analysis . in order to utilize the known result ( [ 3.1 ] ) , we need to give the difference between the bilinear form of fve and that of fem . + * lemma 3.1 * + _ proof_by green s formula , we have @xmath84 and ( see fig.1 ) @xmath85 substituting this two identities into the definitions of @xmath86 and @xmath87 ( see ( [ 2.4 ] ) and ( [ 2.6 ] ) ) , the proof is completed.@xmath88 let @xmath89 be a rectangular element , @xmath90 , see fig.2 .
the following lemma gives some exact calculations for integrals on @xmath91 .
fig.2.rectangular element @xmath92 + * lemma 3.2 * @xmath93 _ proof_we only prove ( [ 3.3 ] ) , the other are similar . noting that @xmath94 and @xmath95 are linear on line segment @xmath96 ( @xmath97 is its middle point )
, we have from the definition of @xmath98 that @xmath99 on the other hand , using the composite trapezoidal rule : @xmath100 we obtain by taking @xmath101 that @xmath102 the difference of above two equalities implies the first equality in ( [ 3.3 ] ) .
the second equality comes from ( [ 2.13 ] ) , noting that @xmath103 is constant on @xmath96.@xmath88 let @xmath104 denote the piecewise constant approximation of function @xmath105 on @xmath5 , @xmath106 * theorem 3.1*_let @xmath5 be a regular rectangular mesh and @xmath107 .
then we have _ @xmath108 _
proof_let @xmath109 be the value of @xmath110 at the midpoint @xmath97 of edge @xmath111 , see fig.2 . from lemma 3.1
, we have @xmath112 using ( [ 2.10])-([2.11 ] ) and ( [ 2.14])-([2.15 ] ) , we obtain @xmath113 for @xmath114 , set @xmath115 . since @xmath116 then we have @xmath117 obviously , @xmath118 .
now , let @xmath119 , see fig.2 .
since @xmath120 hence @xmath121 therefore , we have from ( [ 2.13 ] ) that @xmath122 using the approximation property , we obtain @xmath123 so that @xmath124 now we need to estimate @xmath125 noting that @xmath126 .
let @xmath127 be an interior edge , that is , a common edge of two adjacent element @xmath51 and @xmath128 .
since @xmath129 and @xmath130 is continuous across edge @xmath127 , then we have @xmath131 without loss of generality , let @xmath132 be a rectangular element and the common edge @xmath133 , see fig.2 . from lemma 3.2
we have @xmath134 then , from ( [ 3.14 ] ) , ( [ 3.4 ] ) and ( [ 3.11 ] ) , and noting that @xmath135 , @xmath136 is continuous acrose @xmath137 , we obtain @xmath138 let @xmath139 , be an approximation of function @xmath105 , which satisfies @xmath140 for example , for @xmath33 we may take @xmath141 , @xmath142 is the @xmath16-projection operator ; for @xmath143 , we may take @xmath144 .
then , from ( [ 3.15 ] ) and the inverse inequality , we have @xmath145 where we have used the fact that @xmath146 . combining ( [ 3.13 ] ) with ( [ 3.16 ] ) , we can conclude that @xmath147 substituting estimates @xmath148 into ( [ 3.9 ] ) , we complete the proof by using ( [ 3.1]).@xmath88 from theorem 3.1 , we immediately obtain the following superclose result . + * theorem 3.2*_let @xmath66 and @xmath82 be the solutions of problem ( [ 1.5 ] ) and ( [ 2.8 ] ) , respectively , @xmath149
. then we have _ @xmath150 _
proof_from lemma 2.2 , error equation ( [ 2.9 ] ) and weak estimate ( [ 3.8 ] ) , we have @xmath151 this gives the conclusion.@xmath152 by using the embedding theory and weak embedding inequality of finite element space @xcite , a direct result from theorem 3.2 is the following optimal @xmath153-error estimates , @xmath154
* definition 4.1 * it is well known that for the bilinear interpolation @xmath77 on rectangular meshes , the optimal stress point set @xmath3 is composed of all interior mesh points , midpoints of interior edges and the midpoints of elements @xcite . in this section
, we will derive the @xmath155-superconvergence result for the bilinear finite volume solution @xmath82 on set @xmath3 .
we first give some lemmas .
let rectangular element @xmath156 . for @xmath157 , set @xmath158 .
introduce the discrete @xmath159-norm by @xmath160 where @xmath161 * lemma 4.1 * @xmath162 _ proof_let @xmath163\times [ y_1,y_1+h_y ] , \xi=(x - x_1)/h_x,\;\eta=(y - y_1)/h_y$ ] . for @xmath164 , we may write @xmath165 set @xmath166 then , we have from ( [ 4.5])-([4.7 ] ) that @xmath167 hence , it follows from the cauchy inequality and the regularity condition ( [ 2.2 ] ) that @xmath168 combining this with ( [ 4.3 ] ) , the proof is completed.@xmath152 let @xmath169 be the piecewise constant approximation of matrix function @xmath110 . + * lemma 4.2 * , @xmath170 _ proof_from lemma 3.2 we obtain @xmath171 now , we need to introduce the regularized green function @xcite . for any given @xmath172 , let @xmath173 be the smooth @xmath174-function which satisfies , @xmath175 for any appointed direction @xmath176 , define the direction derivative @xmath177
then , there exists a regularized green function of derivative type @xmath178 such that @xmath179 let @xmath180 be the finite element approximation of @xmath181 such that @xmath182 clearly , we have @xmath183 assume that partition @xmath5 is quasi - uniform so that the inverse inequality holds on finite element space @xmath27 . under this condition ,
the following boundness estimates were given in @xcite @xmath184 where @xmath38 is a positive constant independent of @xmath185 .
+ * theorem 4.1*_let partition @xmath5 be quasi - uniform , and @xmath66 and @xmath82 be the solutions of problem ( [ 1.5 ] ) and ( [ 2.8 ] ) , respectively , @xmath186 .
then we have _ @xmath187
_ proof_from ( [ 4.9 ] ) , error equation ( [ 2.9 ] ) , theorem 3.1 and ( [ 4.10 ] ) , we obtain @xmath188 below we need to estimate ( see lemma 3.1 ) @xmath189 first , using ( [ 2.11 ] ) , ( [ 2.12 ] ) and ( [ 2.15 ] ) , theorem 3.2 and ( [ 4.10 ] ) , we have @xmath190 next , it follows from lemma 4.2 and inverse inequality ( [ 2.12 ] ) that @xmath191 from ( [ 4.5 ] ) and lemma 4.1 , we have @xmath192 substituting this estimate into @xmath193 and using theorem 3.2 , we obtain @xmath194 similarly , by lemma 4.1 and theorem 3.2 , we have ( noting that @xmath195 ) @xmath196 for @xmath197 , using theorem 3.2 , ( [ 2.14 ] ) and ( [ 4.10 ] ) to obtain @xmath198 substituting estimates @xmath199 into ( [ 4.13 ] ) , we complete the proof by ( [ 4.12]).@xmath152 now we can give the main superconvergence result . +
* theorem 4.2*_let partition @xmath5 be quasi - uniform , and @xmath66 and @xmath82 be the solutions of problem ( [ 1.5 ] ) and ( [ 2.8 ] ) , respectively , @xmath186 .
then we have @xmath200 where @xmath201 denotes the average values of gradient @xmath202 in elements containing point @xmath2 and @xmath3 is the optima stress point set composed of all interior mesh points , the midpoints of interior edges and the midpoints of elements . _
+ _ proof_from ( [ 4.1 ] ) and theorem 4.1 , we immediately obtain the conclusion by using the triangle inequality.@xmath152
in this section , we will present some numerical results to illustrate the theoretical analysis . in the numerical experiments , we first partition domain @xmath47 into a rectangular mesh @xmath5 ,
then the refined meshes are obtained by connecting the midpoints of each edge of elements in @xmath5 .
thus , a rectangular mesh sequence is generated with successively halving mesh size @xmath206 .
denote by @xmath207 the computation error and the numerical convergence rate is computed by using the formula @xmath208 .
table 1 gives the error and convergence rate with successively halved mesh sizes .
we see that the convergence rate is just about @xmath209 , as the theoretical prediction . | we study the gradient superconvergence of bilinear finite volume element ( fve ) solving the elliptic problems .
first , a superclose weak estimate is established for the bilinear form of the fve method .
then , we prove that the gradient approximation of the fve solution has the superconvergence property : @xmath0 , where @xmath1 denotes the average gradient on elements containing point @xmath2 and @xmath3 is the set of optimal stress points composed of the mesh points , the midpoints of edges and elements . |
the chromosphere of the quiet sun is one of the most complex stellar atmospheric regions ( e.g. , harvey 2006 , 2009 ; judge 2006 , 2009 ) . lying between the thin photosphere and the extended @xmath5 k corona
, it is here where the magnetic field becomes the globally dominating factor ruling the outer solar atmosphere .
measuring the magnetic field vector in the solar chromospheric plasma is , however , notoriously difficult , especially outside sunspots and related active regions .
quantitative information can be obtained through spectropolarimetry , but the measurement and physical interpretation of the weak polarization signatures that the hanle and zeeman effects produce in the few chromospheric lines that can be observed from ground - based telescopes is not an easy task ( see trujillo bueno 2010 , for a recent review ) .
it is thus important to develop novel methods of measuring " the chromospheric magnetic field , ideally based on the action of the hanle and zeeman effects in spectral lines whose line - center intensity images reveal the fine - scale thermodynamic structuring of the solar chromosphere .
here we show that the scattering polarization observed in the ir triplet of ca ii contains valuable information on the thermal and magnetic structure of the quiet " chromosphere , from the shock dominated region of the lower chromosphere " to the upper chromosphere " below the interface region to the @xmath5 k solar corona .
the only way to obtain direct empirical information on the intensity and topology of the magnetic fields of the solar chromosphere is via the measurement and interpretation of polarization signals in chromospheric spectral lines . in regions with high concentrations of magnetic flux , such as in sunspots and plages ,
the polarization patterns are dominated by the zeeman effect .
hence , diagnostic techniques based on this effect are quite useful ( e.g. , socas - navarro , trujillo bueno & ruiz cobo 2000 ; socas - navarro 2005 ) . however , in the quiet sun , which covers most of the solar disk at any given time during the solar magnetic activity cycle , the zeeman splitting between the @xmath6 ( @xmath7 ) and @xmath8 ( @xmath9 ) components ( @xmath10 being the magnetic quantum number ) is only a small fraction of the width of the spectral lines formed in the chromosphere . as a result , the transverse zeeman effect makes a very small or insignificant contribution to the emergent linear polarization amplitudes . on the other hand , the circular polarization of the zeeman effect as a diagnostic tool is of limited practical interest for the exploration of the magnetism of the upper chromosphere because the response function of the emergent stokes @xmath11 profiles of strong lines like h@xmath12 and ca ii 8542 is significant only in the photosphere and/or lower chromosphere ( socas - navarro & uitenbroek 2004 ; uitenbroek 2006 ) .
there is yet another mechanism producing linear polarization in the spectral lines that can be exploited to diagnose the solar chromosphere .
atoms align due to the optical pumping caused by the incident anisotropic radiation field ; i.e. , the individual magnetic @xmath10 sublevels of energy levels with total angular momentum @xmath13 are unevenly populated ( in such a way that the populations of substates with different values of @xmath14 are unequal ) and coherences between them may appear .
this , in turn , gives rise to a sizable amount of linear polarization in chromospheric spectral lines , either by selective emission or absorption of polarization components in the spectral line .
a remarkable example is provided by the ir triplet of ca ii .
its linear polarization pattern was considered `` enigmatic '' for a while ( stenflo , keller & gandorfer 2000 ) because it was impossible to achieve even a qualitative fit to the observed relative polarization amplitudes considering that scattering line polarization results only from selective emission processes caused by the population imbalances of the line s upper level .
it was later realized that selective absorption of polarization components caused by the presence of populations imbalances in the line s lower level is a very efficient mechanism for producing linear polarization in the spectral lines of a stellar atmosphere ( trujillo bueno & landi deglinnocenti 1997 ; trujillo bueno 1999 ; trujillo bueno et al .
in fact , manso sainz & trujillo bueno ( 2003 ) demonstrated quantitatively that the physical origin of the linear polarization pattern observed in the ca ii lines at 8542 and 8662 is indeed zero - field " dichroism , i.e. , selective absorption of polarization components caused by the presence of atomic alignment in the metastable lower levels @xmath15 and @xmath16 ( see fig .
1 ) , while that observed in the 8498 line is controled by both dichroism and by the selective emission of polarization components that result from the population imbalances between the sublevels of its upper level @xmath17 .
the alignment of the atomic @xmath18-levels and the ensuing emergent linear polarization is modified by the presence of a magnetic field through the hanle effect , even if the magnetic field turns out to have a complex , tangled topology , with mixed polarities at subresolution scales ( e.g. , landi deglinnocenti & landolfi 2004 ) . in the absence of collisional depolarization the magnetic field strength @xmath19 ( in g ) that is sufficient to produce a significant change is @xmath20 where @xmath21 and @xmath22 are , respectively , the lifetime ( in seconds ) and the land factor of the @xmath18 level under consideration .
interestingly enough , since the lifetimes of the metastable lower levels of the ca ii ir triplet are about two orders of magnitude larger than the upper level lifetimes and levels with @xmath23 or @xmath24 can not carry atomic alignment , the linear polarization of the ca ii line at 8662 ( whose upper level has @xmath25 ) is expected to be sensitive to mg fields only , while the scattering polarization of the 8498 and 8542 lines ( which share the same upper level whose @xmath26 ) can in principle be sensitive also to magnetic fields in the gauss range . in the quiet solar chromosphere depolarization of the ca ii atomic levels by elastic collisions with neutral hydrogen atoms
is insignificant for the upper levels of the h and k lines , while such collisions have a rather small impact on the atomic polarization of the metastable lower levels ( manso sainz & trujillo bueno 2003 ; derouich , trujillo bueno & manso sainz 2007 ) . actually , the mere fact that the ca ii 8662 line shows a clear @xmath3 polarization signal ( see stenflo et al .
2000 ) implies that its lower level ( whose @xmath27 ) is significantly polarized in the quiet solar chromosphere ( manso sainz & trujillo bueno 2003 ) .
the aim of this paper is to investigate the thermal and magnetic sensitivity of the scattering polarization @xmath3 and @xmath4 profiles of the ca ii ir triplet . after formulating in 2 the multilevel problem of scattering polarization and the hanle effect ,
4 discusses the unmagnetized case in the hot " and cool " chromospheric models described in 3 .
the impact of the hanle effect on the linear polarization profiles of the ca ii ir triplet is discussed in 5 , showing for each solar atmospheric model detailed results for the cases of disk center ( 5.1 ) and close to the limb observations ( 5.2 ) .
the ensuing information is summarized in 5.3 in the form of hanle effect diagrams for each of the spectral lines of the ir triplet . finally ,
6 summarizes our main conclusions with an outlook to future research .
the relevant radiative transfer problem here implies calculating , at each spatial point of the ( generally magnetized ) astrophysical plasma model under consideration , the values of the diagonal and non - diagonal elements of the atomic density matrix corresponding to each atomic level of total angular momentum @xmath18 , which quantify its overall population , the population imbalances between its magnetic sublevels , and the quantum coherences between each pair of them .
the values of such density - matrix elements have to be consistent with the intensity , polarization and symmetry properties of the radiation field generated within the medium .
this requires solving jointly the rt equations for the stokes parameters and the statistical equilibrium equations for the density - matrix elements .
once such a self - consistent solution is obtained it is straightforward to compute the emergent stokes profiles via the formal solution of the stokes - vector transfer equation for any desired line of sight ( los ) .
the aim of this section is to formulate such equations with emphasis on the case of the ir triplet of ca ii .
the excitation state of the atomic system is described through the density matrix ( fano 1957 ) .
we consider the populations ( @xmath28 ) , and coherences ( @xmath29 ) among the @xmath30 magnetic sublevels @xmath10 of each given atomic level with total angular momentum @xmath18 . alternatively
, we use the multipolar components of the atomic density matrix ( omont 1977 ) : @xmath31 where @xmath32 , @xmath33 , and the expression in brackets is the wigner 3j - symbol ( e.g. , brink & satchler 1968 ) .
the @xmath34 elements with @xmath35 are real numbers measuring population imbalances between sublevels , while @xmath36 is the total population of the level .
the @xmath37 elements are combinations of the complex non - diagonal components @xmath29 .
hermiticity of the density matrix ( @xmath38 ) implies @xmath39^*$ ] ( @xmath40 meaning complex conjugate ) . in the absence of net circular polarization in the incident radiation @xmath41 ,
i.e. , atoms _ align _ but they do not _
orientate_. hence , odd-@xmath42 @xmath43 elements vanish . therefore , the 5-level atomic model of ca ii considered here ( fig .
1 ) requires @xmath44 @xmath43 elements to fully characterize the excitation state of the atomic system .
the transfer of spectral line polarization is described by the following equations for the stokes parameters [ rte ] @xmath45 \\
\frac{\rm d}{{\rm d}s}q&=\epsilon_q-\eta_qi-\eta_iq-\rho_vu+\rho_uv , \label{rte2 } \displaybreak[0 ] \\
\frac{\rm d}{{\rm d}s}u&=\epsilon_u-\eta_ui+\rho_vq-\eta_iu-\rho_qv , \label{rte3 } \displaybreak[0 ] \\
\frac{\rm d}{{\rm d}s}v&=\epsilon_v-\eta_vi-\rho_uq+\rho_qu-\eta_iv.\label{rte4}\end{aligned}\ ] ] in the hanle effect regime the magnetic splitting of @xmath46 and @xmath6 components of the line transitions are negligible and the spectral line emissivity thus depends exclusively on the excitation state of the upper level of the transition ( i.e. , on the @xmath47 values ) .
in particular , linearly polarized emission is due to the presence of alignment and coherences between atomic sublevels of the upper level of the transition .
moreover , since levels are not oriented , @xmath48 . more explicitly ( landi deglinnocenti & landolfi 2004 ; see also manso sainz & trujillo bueno 1999 ) : [ epsilons ] @xmath49 - \sin \chi { \rm im}[{{{\rho}}^2_1 } ] )
\\ + \frac{\sqrt{3}}{2 } ( 1-\mu^2 ) ( \cos 2\chi \ , { \rm re}[{{{\rho}}^2_2}]-\sin 2\chi \ , { \rm im}[{{{\rho}}^2_2 } ] ) \big{\}},\end{gathered}\ ] ] @xmath50 - \sin \chi { \rm im}[{{\rho}}^2_1 ] ) \\ - \frac{\sqrt{3}}{2 } ( 1+\mu^2 ) ( \cos 2\chi \ , { \rm re}[{{\rho}}^2_2]-\sin 2\chi \ , { \rm im}[{{\rho}}^2_2 ] ) \big{\}},\end{gathered}\ ] ] @xmath51+\cos \chi { \rm im}[{{\rho}}^2_1 ] ) + \mu ( \sin 2\chi \ , { \rm re}[{{\rho}}^2_2 ] + \cos 2\chi \ , { \rm im}[{{\rho}}^2_2 ] ) \big{\}},\ ] ] where @xmath52 ( with @xmath53 the total number of atoms per unit volume , @xmath54 the einstein coefficient for spontaneous emission and @xmath55 the voigt profile , with @xmath56 and @xmath57 the transition s frequency ) , @xmath58 and @xmath59 are the inclination and azimuth of the ray , and @xmath60 is a numerical coefficient that depends only on @xmath61 and @xmath62 in particular , @xmath63 , @xmath64 , and @xmath65 ( see table 10.1 in landi deglinnocenti & landolfi 2004 ) . the absorption coefficient @xmath66 , and the dichroism coefficients @xmath67 and @xmath68 are given by analogous expressions but with @xmath69 instead of @xmath70 , @xmath71 instead of @xmath60 ( @xmath72 , @xmath73 ) , and with the @xmath74 values of the _ lower _ level of the line transition instead of those of the upper level ( we neglect stimulated emissions ) . likewise , since we are assuming that the atomic levels are not oriented , @xmath75 ( when neglecting the wavelength shifts between the @xmath46 components caused by the zeeman effect ) .
finally , the anomalous dispersion terms ( @xmath76 ) have the same functional dependence on the @xmath74 values of the _ lower _ level , but with an antisymmetric dispersive dependence on frequency , i.e. , @xmath77 , with @xmath78 the dispersion profile . additionally , we consider an unpolarized background continuum providing an extra absorption coefficient @xmath79 and emissivity @xmath80 ( where @xmath81 is the planck function ) , a very good approximation toward the red part of the solar spectrum ( e.g. , gandorfer 2000 ) .
therefore , in eqs .
( 3 ) , @xmath82 , @xmath83 , while all other emission , dichroism , and anomalous dispersion coefficients have only line contributions ( @xmath84 , @xmath85 , @xmath86 , @xmath87 , @xmath88 , @xmath89 ) . note that @xmath90 and @xmath91 depend on both the population imbalances ( @xmath92 ) and on the coherences ( @xmath93 , with @xmath94 ) , while @xmath95 and @xmath96 depend _ only _ on the quantum coherences is the _ perpendicular _ to the plane formed by the propagation direction and the quantization z - axis of total angular momentum ] .
the @xmath97 elements produce the dominant contribution to the stokes @xmath98 profile , while the @xmath93 elements ( the _ alignment _ components ) contribute to the _ linear _ polarization signals , which we quantify by the stokes parameters @xmath1 and @xmath2 . although @xmath99 ( because , as mentioned above , @xmath100 ) according to eq .
( 3d ) it is possible to generate stokes @xmath11 in the spectral line due to the @xmath101 ( second - order ) terms .
this contribution is however negligible compared to the longitudinal zeeman effect signal produced by the splitting of the @xmath46-components of the emission profile which we do not consider here .
since the degree of anisotropy of the radiation field in solar - like atmospheres is weak , the population imbalances and coherences of each @xmath18-level are only a small fraction of its overall population .
therefore , second - order terms can be safely neglected in the stokes - vector transfer equation and the relevant equations for calculating @xmath3 and @xmath4 are @xmath102 -\eta_i q , \\
\frac{\rm d}{{\rm d}s}u & \approx [ \epsilon_u-\eta_ui ] -\eta_i u. \end{aligned}\ ] ] nevertheless , all the calculations of this paper were carried out by solving numerically the full stokes vector transfer equations ( 3 ) through the application of the formal solver mentioned in appendix a. the complication with respect to the unmagnetized case considered by manso sainz & trujillo bueno ( 2003 ) is that now the presence of a magnetic field of given strength @xmath19 , inclination @xmath103 and azimuth @xmath104 at each grid point of the stellar atmosphere model under consideration implies that we have to consider also the quantum coherences ( @xmath93 , with @xmath105 ) between each pair of magnetic substates pertaining to each @xmath18 level , in addition to their individual populations ( @xmath92 ) .
for this reason , not only stokes @xmath1 but also stokes @xmath2 is now non - zero , in general .
the expressions above for the components of the emission vector and propagation matrix ( eqs . 4 and paragraphs following them ) simplify considerably for several cases of practical interest .
for a one - dimensional atmosphere , either unmagnetized , or permeated by a _
microturbulent _ and _ isotropically distributed _ magnetic field of strength
@xmath19 , or in the presence of a deterministic magnetic field with a fixed inclination but with a random azimuth below the line s photon mean - free - path , or in the presence of a deterministic magnetic field with a fixed orientation but with a strength in the saturation regime of the hanle effect , the quantum coherences ( i.e. the real and imaginary parts of the @xmath106 and @xmath107 components ) vanish in a reference system whose quantization axis must be the parallel to the magnetic field vector for the deterministic field case and the the normal to the stellar surface for the three other cases : @xmath108},\ ] ] @xmath109 @xmath110 where the @xmath34 ( @xmath111 ) values are those of the _ upper _ level of the line transition under consideration , and where the orientation of the ray is specified by the azimuthal angle @xmath59 and by @xmath112 ( with @xmath113 the angle between the direction of the radiation beam and the quantization axis , which must be the magnetic field direction itself for the deterministic field case and the normal to the surface of the stellar atmospheric model for the two other cases ) .
likewise , the only non - zero elements of the propagation matrix are @xmath114 and @xmath91 which are given by identical expressions ( i.e. by @xmath115 and @xmath116 ) , but with @xmath117 instead of @xmath70 , @xmath71 instead of @xmath60 and with the @xmath34 values of the _ lower _ level of the line transition instead of those of the upper level ( for the case in which stimulated emissions are neglected ) .
under such assumptions , and taking into account the low polarization level in solar - like atmospheres ( i.e. , that @xmath118 and @xmath119 ) , an approximate formula can be derived to estimate the emergent fractional linear polarization amplitude at the center of a strong spectral line ( trujillo bueno 2003 ) : @xmath120,\ ] ] where @xmath121 must be evaluated at height in the model atmosphere where the line - center optical distance along the line of sight is unity .
its first term is due to _ selective emission _ of polarization components ( caused by the population imbalances of the upper level ) , while its second term accounts for `` zero - field '' dichroism .
note that since @xmath63 and @xmath72 the scattering polarization of @xmath122 is exclusively due to `` zero - field '' dichroism .
general equations for the multivel atom model which neglects quantum interferences between the sublevels pertaining to different @xmath18-levels are derived in landi deglinnocenti & landolfi ( 2004 ) . for the particular case of the 5-level atom of fig .
1 the following rate equations for the density matrix elements follow : @xmath123 \rho^0_0(1 ) + a_{41}\rho^0_0(4 ) + \sqrt{2 } a_{51}\rho^0_0(5 ) + \sum_{i\neq 1 } c_{i1 } \sqrt{\frac{2j_i+1}{2 } } \rho^0_0(i ) , \end{split } \displaybreak[0 ] \\
\begin{split}\label{see02 } \frac{\rm d}{{\rm d}t } \rho^0_0(2 ) = & -\biggl[\sum_{u=4}^5 b_{2u}j^0_0(2\rightarrow u ) + \sum_{i\neq 2 } c_{2i } \biggr ] \rho^0_0(2 ) - \sum_{q'=-2}^2 \biggl ( \frac{1}{\sqrt{2 } } b_{24}j^2_{q'}(2\rightarrow 4 ) -\frac{2\sqrt{2}}{5 } b_{25}j^2_{q'}(2\rightarrow 5 ) \biggr ) \rho^2_{q'}(2 ) \\ & + \frac{1}{\sqrt{2 } } a_{42}\rho^0_0(4 ) + a_{52}\rho^0_0(5 ) + \sum_{i\neq 2 } c_{i2 } \frac{\sqrt{2j_i+1}}{2 } \rho^0_0(i ) , \end{split } \displaybreak[0 ] \\
\begin{split}\label{see03 } \frac{\rm d}{{\rm d}t } \rho^2_q(2 ) = & -\biggl [ 2\pi{\rm i}\nu_{\rm l}g_2 q + \sum_{u=4}^5 b_{2u}j^0_0(2\rightarrow
u ) + \sum_{i\neq 2 } c_{2i } -d_2^{(2)}\biggr ]
\rho^2_q(2 ) \\ & - \biggl ( \frac{1}{\sqrt{2 } } b_{24}j^2_{-q}(2\rightarrow 4 ) -\frac{2\sqrt{2}}{5 } b_{25}j^2_{-q}(2\rightarrow 5 ) \biggr)(-1)^q \rho^0_0(2 ) \\ & + \frac{1}{5 } a_{52}\rho^2_q(5 ) + \sum_{i=3 , 5 } c_{i2}^{(2 ) } \frac{\sqrt{2j_i+1}}{2 } \rho^2_q(i ) , \end{split } \displaybreak[0 ] \\ \begin{split}\label{see04 } \frac{\rm d}{{\rm d}t } \rho^0_0(3 ) = & -\biggl[b_{35}j^0_0(3\rightarrow 5 ) + \sum_{i\neq 3 } c_{3i } \biggr ] \rho^0_0(3 ) - \sum_{q=-2}^2 \frac{\sqrt{7}}{5 } b_{35}j^2_q(3\rightarrow 5 ) \rho^2_q(3 ) \\ & + \sqrt{\frac{2}{3 } } a_{53 } \rho^2_q(5 ) + \sum_{i\neq 3 } c_{i3 } \sqrt{\frac{2j_i+1}{6 } } \rho^0_0(i ) , \end{split } \displaybreak[0 ] \\
\begin{split}\label{see05 } \frac{\rm d}{{\rm d}t } \rho^2_q(3 ) = & -\biggl [ 2\pi{\rm i}\nu_{\rm l}g_3 q + b_{35}j^0_0(3\rightarrow 5 ) + \sum_{i\neq 3 } c_{3i } -d_3^{(2)}\biggr ] \rho^2_q(3 ) - b_{35}(-1)^q j^2_{-q}(3\rightarrow 5)\frac{\sqrt{7}}{5 } \rho^0_0(3 ) \\ & - \sum_{q , q=-2}^2 b_{35}j^2_{q}(3\rightarrow 5 ) \biggl [ -\sqrt{\frac{5}{7 } } \biggl ( \begin{matrix } 2 & 2 & 2 \\ q & -q ' & q \end{matrix } \biggr ) ( -1)^{q ' } \rho^2_{q'}(3 ) + \frac{9}{2}\sqrt{\frac{3}{35 } } \biggl ( \begin{matrix } 2 & 4 & 2 \\ q & -q ' & q \end{matrix } \biggr ) ( -1)^{q ' } \rho^4_{q'}(3 ) \biggr ] \\ & + \frac{2}{5}\sqrt{\frac{7}{3 } } a_{53 } \rho^2_q(5 ) + \sum_{i=2 , 5 } c_{i3}^{(2 ) } \sqrt{\frac{2j_i+1}{6 } } \rho^2_0(i ) , \end{split } \displaybreak[0 ] \\ \begin{split}\label{see06 } \frac{\rm d}{{\rm d}t } \rho^4_q(3 ) = & -\biggl [ 2\pi{\rm i}\nu_{\rm l}g_3 q + b_{35}j^0_0(3\rightarrow 5 ) + \sum_{i\neq 3 } c_{3i } -d_3^{(4)}\biggr ] \rho^4_q(3 ) \\ & - \sum_{q , q'=-2}^2 b_{35 } j^2_q(3\rightarrow 5 ) \biggl [ \frac{9}{2}\sqrt{\frac{3}{35 } } \biggl ( \begin{matrix } 4 & 2 & 2 \\ q & -q ' & q \end{matrix } \biggr)(-1)^{q ' } \rho^2_{q'}(3 ) + 3\sqrt{\frac{11}{70 } } \biggl ( \begin{matrix } 4 & 4 & 2 \\ q & -q ' & q \end{matrix } \biggr)(-1)^{q ' } \rho^4_{q'}(3 ) \biggr ] . \end{split } \displaybreak[0 ] \\ \begin{split}\label{see07 } \frac{\rm d}{{\rm d}t } \rho^0_0(4 ) = & -\biggl [ \sum_{\ell=1}^2 a_{4\ell } + \sum_{i\neq 4 } c_{4i } \biggr ] \rho^0_0(4 ) + \sum_{\ell=1}^2 b_{\ell 4 } j^0_0(\ell \rightarrow 4 ) \sqrt{\frac{2j_\ell+1}{2 } } \rho^0_0(\ell ) \\ & + \sum_{q=-2}^2 b_{24 } j^2_q(\ell \rightarrow 4 ) \rho^2_q(2 ) + \sum_{i\neq 4 } c_{i4 } \sqrt{\frac{2j_i+1}{2 } } \rho^0_0(i ) , \end{split } \displaybreak[0 ] \\
\begin{split}\label{see08 } \frac{\rm d}{{\rm d}t } \rho^0_0(5 ) = & -\biggl [ \sum_{\ell=1}^3 a_{5\ell } + \sum_{i\neq 5 } c_{5i } \biggr ] \rho^0_0(5 ) + \sum_{\ell=1}^3 b_{\ell 5 } j^0_0(\ell \rightarrow 5 ) \frac{\sqrt{2j_\ell+1}}{2 } \rho^0_0(\ell ) \\ & + \sum_{q=-2}^2 \biggl ( -\frac{2\sqrt{2}}{5 } b_{25 } j^2_q(2\rightarrow 5 ) \rho^2_q(2 ) + \frac{\sqrt{42}}{10 } b_{35 } j^2_q(3\rightarrow 5 ) \rho^2_q(3 )
\biggr ) + \sum_{i\neq 5 } c_{i5 } \frac{\sqrt{2j_i+1}}{2 } \rho^0_0(i ) , \end{split } \displaybreak[0 ] \\
\begin{split}\label{see09 } \frac{\rm d}{{\rm d}t } \rho^2_q(5 ) = & -\biggl [ 2\pi{\rm i}\nu_{\rm l}g_5
q + \sum_{\ell=1}^3 a_{5\ell } + \sum_{i\neq 5 } c_{5i } -d_5^{(2)}\biggr ] \rho^2_q(5 ) + \frac{1}{5}b_{25}j^0_0(2\rightarrow 5)\rho^2_q(2 ) + \frac{\sqrt{21}}{5}b_{35}j^0_0(3\rightarrow 5)\rho^2_q(3 ) \\ & + -\frac{2\sqrt{2}}{5 } b_{25 } j^2_{-q}(2\rightarrow 5 ) \rho^0_0(2 ) + \sqrt{\frac{3}{100 } } b_{35 } j^2_{-q}(3\rightarrow 5 ) \rho^0_0(3 ) \\ &
+ \sum_{q'=-2}^2 \biggl ( 2\sqrt{\frac{7}{5 } } b_{25 } j^2_{q'}(2\rightarrow 5 ) \rho^2_{q+q'}(2 ) -\sqrt{\frac{3}{5 } } b_{35 } j^2_{q'}(3\rightarrow 5 ) \rho^2_{q+q'}(3 ) \biggr ) \biggl ( \begin{matrix } 2 & 2 & 2 \\ -q & q - q ' & -q ' \end{matrix } \biggr ) \\ & + \sum_{i=2 , 3 } c_{i5}^{(2 ) } \frac{\sqrt{2j_i+1}}{2 } \rho^2_0(i ) .
\end{split}\end{aligned}\ ] ] in the previous equations @xmath124 and @xmath125 are the einstein coefficients for spontaneous emission and absorption , respectively ( see table 1 ) .
@xmath126 ( @xmath127 ; @xmath128 ) are the radiation field tensors introduced by landi deglinnocenti ( 1984 ) .
they are integrals over frequency and solid angle of the stokes parameters ( for their explicit expressions appropiate for computational work see manso sainz & trujillo bueno 1999 ) .
in particular , @xmath129 is the mean intensity integrated over the absorption profile , while @xmath130\,.\ ] ] the contribution of other @xmath131 components is very small here , where only the magnetic field breaks the symmetry of the radiation field .
thus , @xmath132 throughout the whole outer atmosphere in the five transitions relevant for our problem .
@xmath133 and @xmath134 are the excitation and desexcitation collisional rates , respectively ( see table 1 ) , between levels @xmath135 and @xmath136 , while @xmath137 and @xmath138 are collisional transfer rates for alignment between polarizable levels 2 , 3 , and 5 .
derouich et al .
( 2007 ) have computed values for these transitions in ca ii .
alternatively one can follow the approach suggested by landi deglinnocenti & landolfi ( 2004 ) , and consider @xmath139 where @xmath140 is just a numerical factor depending on the quantum numbers of the transition .
@xmath141 is the depolarization rate of the @xmath42-th multipole of level @xmath142 due to elastic collisions with neutral hydrogen .
theoretical values for ca ii are given by derouich et al .
( 2007 ) , who confirmed that such @xmath141 terms of the statistical equilibrium equations have only a very small impact on the atomic polarization of the metastable lower levels ( cf .
, manso sainz & trujillo bueno 2003 ) . the first term on the r.h.s .
( [ see03 ] ) , ( [ see05])-([see06 ] ) , and ( [ see09 ] ) represents the relaxation of coherences in the presence of a magnetic field with larmor frequency @xmath143 ( @xmath19 expressed in gauss and @xmath144 in hz ) , and @xmath145 the land factor of the @xmath18-level under consideration .
equations ( [ see01])-([see09 ] ) are expressed in a reference system with the quantization axis along the magnetic field .
the radiative and collisional terms remain formally invariant in any other reference system although clearly , their actual numerical values may change due to the different values of @xmath131 in the new reference system . in an arbitrary reference system where the magnetic field has an inclination @xmath103 with respect to the quantization axis and an azimuth @xmath104
, the magnetic relaxation term reads : @xmath146 , \end{gathered}\ ] ] where the term multiplying @xmath147 ( @xmath148 ) is not considered when @xmath149 ( @xmath150 ) .
the statistical equilibrium equations ( [ see01])-([see09 ] ) have been derived under several hypotheses .
collisions are considered under the impact approximation .
the interaction time of the collision is negligible and hence , collisional and radiative terms can be added independently .
the colliders are assumed to be isotropically distributed ( they are maxwellian ) , and they can not generate atomic polarization , though they can transfer it between levels .
radiative rates have been derived within the framework of the quantum theory of spectral line polarization described in landi deglinnocenti and landolfi ( 2004 ) , assuming that the radiation field is _ spectrally flat _
( i.e. , lacking spectral structure ) , in spectral ranges of the order of the energy separation between levels with coherences .
this is an excellent approximation for the case under consideration since we are considering relatively weak fields ( less than @xmath151100 g ) and we neglect coherences between different @xmath18 levels .
neglecting @xmath18-level interferences is a good approximation for modeling the ca ii ir triplet and the core of the h- and k - lines , but it can not account for the general @xmath3 pattern of the uv doublet ( see stenflo 1980 ) , which is beyond the scope of this work .
finally , we neglect doppler correlations due to the thermal movement of the atoms .
all these hypotheses together account for what is called complete frequency redistribution ( crd ) in the laboraroty frame , since correlations between the incident and scattered radiation fields are completly neglected .
we calculate the 29 @xmath74 unknowns at each grid point of the chosen stellar atmosphere model assuming statistical equilibrium @xmath152 the resulting system of equations is linearly dependent and we substitute one of the equations , say , the one for the ground - level population , by the trace equation of the density matrix , which establishes the conservation of the number of particles @xmath153 we have developed a general multilevel radiative transfer computer program for the numerical solution of the non - lte problem of the second kind , which we have applied in this investigation to obtain the self - consistent solution of the previous equations .
see appendix a for details .
cccccccc @xmath154 ( ) & @xmath136 & @xmath135 & @xmath124 ( s@xmath155 ) & + & & & & 3000 k & 5000 k & 7000 k & 9000 k + + k & 5 & 1 & @xmath156 & 3.60@xmath157 & 2.92@xmath157 & 2.59@xmath157 & 2.39@xmath157 + h & 4 & 1 & @xmath156 & 3.60@xmath157 & 2.92@xmath157 & 2.59@xmath157 & 2.39@xmath157 + 8498 & 5 & 2 & @xmath158 & 3.68@xmath157 & 2.93@xmath157 & 2.51@xmath157 & 2.23@xmath157 + 8542 & 5 & 3 & @xmath159 & 1.59@xmath160 & 1.29@xmath160 & 1.12@xmath160 & 1.01@xmath160 + 8662 & 4 & 2 & @xmath161 & 1.61@xmath160 & 1.31@xmath160 & 1.15@xmath160 & 1.04@xmath160 + + & 5 & 4 & & 3.94@xmath157 & 3.05@xmath157 & 2.58@xmath157 & 2.27@xmath162 + & 4 & 3 & & 2.69@xmath157 & 2.08@xmath157 & 1.76@xmath157 & 1.55@xmath157 + & 3 & 1 & & 2.20@xmath157 & 1.71@xmath157 & 1.44@xmath157 & 1.27@xmath157 + & 3 & 2 & & 5.51@xmath157 & 4.27@xmath157 & 3.61@xmath157 & 3.19@xmath157 + & 2 & 1 & & 2.20@xmath157 & 1.71@xmath157 & 1.44@xmath157 & 1.27@xmath157 +
high spatial and temporal resolution intensity images taken at the line center of strong absorption lines , like h@xmath12 and ca ii 8542 , show that the chromosphere of the quiet sun is a highly inhomogeneous and dynamic medium ( e.g. , review by rutten 2007 ) .
the spatial and dynamic complexity is such that it is obvious that the two one - dimensional ( 1d ) atmospheric models we have chosen for this investigation ( i.e. , the hot " fal - c model of fontenla et al .
1993 and the cool " m - co model of avrett 1995 ) should only be considered as rough representations of the stratification of the kinetic temperature and density at two different phases of inter - network oscillations ( see fig .
2 ) . nevertheless ,
as we shall see below , our solution of the multilevel hanle - effect problem for the ir triplet of ca ii in such semi - empirical models of the solar atmosphere allows us to reach the main goal of this paper , namely to demonstrate the diagnostic potential of the linear polarization signals produced by atomic polarization in the levels of the ca ii ir triplet .
we consider a 5-level atomic model of ca ii including the ground level , the metastable @xmath163 levels and the @xmath164 upper levels of the @xmath165 and @xmath42 lines ( see fig .
the excitation state in such a model is given by 29 statistical tensor elements @xmath74 ( @xmath166 , 2 , 4 ; @xmath167 ) , including the total populations of the five levels .
the radiative transition probabilities are those of edln & risberg ( 1956 ) compiled at the nist database .
( note that these transition probabilities for the infrared triplet are slightly larger than those used by manso sainz & trujillo bueno ( 2003a ) for the unmagnetized case ; correspondingly , the @xmath3 values for the zero - field case are slightly larger here than in our previous work . )
inelastic collisional rates and line broadening parameters have been taken from shine & linsky ( 1974 ; see table 1 ) .
elastic depolarizing collisions are treated according to lamb & ter haar ( 1971 ) , which gives results similar to those obtained using the collisional rates given by derouich et al .
the collisional pumping of @xmath168 statistical tensors are treated following landi deglinnocenti & landolfi ( 2004 ) .
we have calculated the number density of ca ii ions at each atmospheric height of the model under consideration by solving the standard non - lte radiative transfer problem for a realistic atomic model that includes also the bound - free transitions from all the bound levels of ca ii to the ground level of ca iii . with the resulting number densities of ca ii ions we then solved the non - lte problem of the second kind for the atomic model of fig . 1 ( see the appendix ) , which is sufficiently realistic for investigating the polarization signatures produced by atomic level polarization and the hanle effect in the lines of the ca ii ir triplet .
we point out that while @xmath169 ( see eq .
1 ) is of the order of a few gauss for the upper level of the 8498 and 8542 lines , it lies in the milligauss range for the lower ( metastable ) levels of the ca ii ir triplet .
for symmetry reasons , in a one - dimensional model atmosphere without magnetic fields the only non - zero stokes parameters are @xmath170 and @xmath171 ( choosing the polarization directions as in eqs .
( [ epsilons ] ) ) .
figure 3 shows the wavelength variation of the emergent fractional linear polarization ( @xmath3 ) in the ca ii ir triplet , calculated for a line of sight with @xmath172 in the hot " fal - c model ( middle panels ) and in the cool " m - co model ( bottom panels ) . while @xmath3 in the 8542 and 8662 lines are positive in both models , the @xmath3 profile of the 8498 line is positive in the hot " model but negative in the
cool " model .
moreover , the 8498 @xmath3 profiles calculated in such atmospheric models have even different shapes , and their amplitudes are one order of magnitude smaller than those of the 8542 and 8662 lines .
the ( positive ) @xmath3 amplitudes of the 8542 and 8662
lines turn out to be significantly larger in the cool " model .
this dependence of the amplitude of the @xmath3 profiles on the chosen model atmosphere is not at all surprising , because the atomic polarization that anisotropic pumping processes induce in the atomic energy levels sensitively depends on the temperature structure of the model .
in particular , the steeper the temperature gradient , the larger the anisotropy factor and hence , the induced atomic level polarization ( e.g. , see figure 4 in trujillo bueno 2001 ) .
the upper panels of fig . 4 show the spatial variation of the degree of anisotropy " in each of the considered atmospheric models , indicating for each los characterized by @xmath173 the atmospheric height where the line - center optical depth is unity along the los .
note that at the height corresponding to a los with @xmath172 the anisotropy is larger in the m - co model .
the decrease of the anisotropy factor at very low optical depths is due to the rapid widening of the @xmath174 profile with which the radiation field is averaged ( e.g. , see eq . (
18 ) for the definition of the @xmath175 radiation field tensor component ) .
this is a consequence of the temperature rising in the upper chromosphere - transition region .
the two lower panels of fig .
4 show the behavior of the fractional atomic alignment of the levels of fig .
1 having @xmath13 ( i.e. , the two metastable levels , which are the lower levels of the ca ii ir triplet , and the upper level of the k - line ) .
the fact that the sign of the @xmath3 profile of @xmath176 ( i.e. , the weakest line of the ca ii ir triplet ) is very sensitive to the thermal structure of the lower chromosphere can be easily understood by combining the information provided in the lower panels of fig .
4 and the approximate eddington - barbier formula ( 8) .
3 suggests that the scattering polarization of the ca ii ir triplet can be used as a sensitive thermometer of the quiet " regions of the solar chromosphere , as already noted by manso sainz & trujillo bueno ( 2001 ) . in this respect , a particularly useful quantity is the center - limb variation of the @xmath3 line - center amplitudes ( see fig . 5 ) . as seen in this figure ,
the most noteworthy feature is that in the hot " fal - c model the ( positive ) @xmath3 amplitude of the 8498 line is larger at @xmath177 than at @xmath178 and that it becomes negative for line of sights with @xmath173 values significantly smaller than 0.1 .
the aim of this section is to show how the hanle effect modifies the emergent linear polarization of the ir triplet of ca ii . to this end
, we consider the following two geometries . in the absence of magnetic fields and horizontal atmospheric inhomogeneities the polarization of the atomic levels
do not produce any linear polarization in the spectral line radiation observed at disc center ( @xmath179 ; see fig .
the same is true in the presence of a vertical magnetic field , because in a one - dimensional stellar atmospheric model the vertical direction coincides with the symmetry axis of the anisotropic radiation field that induces atomic level polarization .
however , in the presence of an inclined magnetic field the symmetry of the problem is broken and , as a result , the ensuing atomic level polarization can generate linear polarization even for a line of sight with @xmath179 . in this case
, the linear polarization is created by the hanle effect of the inclined field ( see a demonstrative observational example in trujillo bueno et al .
2002 ) .
figure 6 shows the fractional linear polarization of the ir triplet of ca ii generated in forward scattering geometry by the hanle effect of a horizontal magnetic field in the fal - c ( top panels ) and m - co ( bottom panels ) models . in the ( hot " ) fal - c model , the linear polarization amplitudes are at the level of @xmath180 when the magnetic strength of the horizontal field is similar or larger than only 0.1 gauss , with @xmath181 ( i.e. , parallel to the horizontal magnetic field ) for the 8498 line and with @xmath182 ( i.e. , perpendicular to the horizontal magnetic field ) for the 8542 and 8662 lines . in the ( cool " ) m - co model @xmath181 in the 8542 and 8662 lines , with amplitudes
again at the @xmath180 level for @xmath183 gauss , while the shape of @xmath3 in the 8498 line is very peculiar ( i.e. , similar to the typical observational signature of the transverse zeeman effect ) and with its maximum amplitude one order of magnitude smaller .
figure 7 shows the results of our rt calculations for the case of a magnetic field inclined by only @xmath184 .
note that now , @xmath185 . at present , detection of these weak linear polarization signals is challenging , though feasible .
but they will become straigtforward with future large - aperture solar telescopes such as the advanced technology solar telescope ( keil , rimmele & wagner 2009 ) or the european solar telescope ( collados 2008 ) .
in forward scattering geometry , symmetry imposes that linear polarization vanishes if the magnetic field is inclined but has a random azimuth distribution , unlike in the fixed inclination and azimuth case just considered .
however , for observations away from the solar disk center ( @xmath186 ) , scattering may produce linear polarization in the emergent spectral line radiation even for magnetic fields with a uniformly distributed azimuth within the spatio - temporal resolution element of the observations .
this is illustrated in fig .
8 , which shows the linear polarization amplitudes of the ca ii ir triplet in the fal - c model ( top panels ) and in the m - co model ( bottom panels ) vary with the magnetic strength ( @xmath19 ) and the inclination ( @xmath103 ) of the assumed random - azimuth field for a close - to - the - limb observation ( @xmath172 ) .
figure 8 shows how sensitive the @xmath3 line - center amplitude of the 8498 line is to the model s thermal structure , and its very peculiar behavior in the sub - gauss range .
the 8542 and 8662 lines are mostly sensitive to magnetic fields between 0.001 and 0.1 gauss , although the 8542 line is also weakly sensitive to fields between 1 and 10 gauss .
the 8542 line enters into the saturation regime of the upper - level hanle effect for fields stronger than @xmath15110 gauss , while just 0.1 gauss saturate the 8662 line . on the contrary
, the 8498 line is highly sensitive in the whole regime 0.00110 gauss . as seen in fig .
3 , in the absence of magnetic fields the linear polarization amplitude of the 8498 line for a los with @xmath172 is one order of magnitude smaller than those corresponding to the 8542 and 8662 lines .
however , the three ca ii lines may show similar @xmath3 amplitudes if there is a magnetic field of the order of only 0.1 gauss in the atmospheric region where the linear polarization of the 8498 line originates .
figure 9 shows the emergent @xmath3 amplitudes averaging the @xmath3 profiles of both atmospheric models with a weight of 0.55 for the fal - c profiles and 0.45 for the m - co profiles .
these weights are similar to those used by avrett ( 1995 ) to simultaneously reproduce the intensity profiles observed in the 4.6 @xmath187 lines of co and in the h & k lines of ca ii , and to those used by holzreuter et al .
( 2006 ) to fit the @xmath3 profile of the ca ii k - line observed in a quiet regions very close to the solar limb .
these type of fits obtained by mixing a cool and a hot component should not be considered as a realistic representation of the chromospheric conditions , but as further indication of the complexity of the ( time - dependent ) three - dimensional thermal structure of the solar chromosphere .
nevertheless , this two - component model shows that the @xmath3 observations of the ir triplet of ca ii reported by stenflo et al .
( 2000 ) are compatible with very weak magnetic fields in the region of formation " of the 8498 line ( i.e. , in the lower chromosphere ) and with fields stronger than 10 gauss ( and inclined by about @xmath184 ) in the regions of formation " of the 8542 and 8662 lines ( i.e. , in the upper chromosphere ) .
stokes @xmath2 is zero if the magnetic field has a random - azimuth distribution , and it is non - zero if the atmosphere is permeated by a magnetic field vector with a well - defined inclination and azimuth .
this is illustrated in the upper panels of fig .
10 , which corresponds to the fal - c model . in the absence of magnetic fields ,
only @xmath3 is non - zero ( see the dashed lines ) , but @xmath4 becomes significant even at fields strengths as low as 0.005 gauss . note that for this case ( b=0.005 gauss ) , characterized by a horizontal magnetic field pointing towards the observer ( triangles ) , perpendicular to the los ( squares ) and away from the los ( diamonds ) , the stokes @xmath3 profiles of the ca ii ir triplet lines are always positive , while @xmath4 changes its sign .
the dotted lines of fig .
10 show the case of a 100 gauss horizontal field pointing towards the observer ( i.e. , a strength for which the three ir lines of ca ii are in the saturation regime of the upper - level hanle effect ) .
the bottom panels of fig .
10 show the results of similar calculations for the cool m - co model . in this model
the @xmath3 profile of the 8498 line changes its sign and shape when the magnetic field azimuth changes from @xmath188 ( or @xmath189 ) to @xmath190 a hanle - effect diagram shows the line - center amplitudes of @xmath3 and @xmath4 varying with the strength and the azimuth of the magnetic field corresponding to a given line - of - sight and inclination of the magnetic field vector .
figure 11 shows restricted hanle - effect diagrams corresponding to the configurations considered in fig . 10 ( close - to - the - limb @xmath172 observation ) .
in particular , the thin solid lines show how linear polarization varies with the magnetic strength , for a horizontal magnetic field pointing towards the observer .
we recall that 50 gauss are needed to completely depolarize the 8498 line , while just 10 gauss are sufficient for the 8542 line and only 0.1 gauss for the 8662 line ( see fig . 8) .
thick solid lines show , for the case of a 0.005 gauss horizontal field , how the @xmath3 and @xmath4 line center amplitudes vary as the orientation of the magnetic field vector changes from pointing towads the observer ( triangles ) , to being perpendicular to the line of sight ( squares ) to pointing away from the observer ( diamonds ) . comparing the upper and lower panels we see the differences due to the different thermal structures of the fal - c and m - co models .
figures 12 and 13 show full hanle - effect diagrams obtained from the emergent stokes @xmath3 and @xmath4 profiles calculated for a close - to - the - limb ( @xmath172 ) line of sight in the hot " and cool " atmospheric models , respectively .
each panel corresponds to a fixed inclination of the magnetic field vector .
each of the four thick lines in any given panel correspond to a fixed strength of the magnetic field with azimuth ranging from @xmath188 to @xmath189 ( solid line ) , and from @xmath189 to @xmath191 ( dashed line ) .
thin lines correspond to magnetic fields ranging between 0 g and 1000 g for a fixed @xmath104 .
they show the pattern of depolarization ( @xmath3 decreases ) and rotation of the direction of linear polarization ( @xmath4 appears ) characteristic of the hanle effect in a close to the limb scattering geometry .
finally , figs .
14 and 15 show hanle - effect diagrams for a disc center observation ( @xmath179 ) .
note that in this forward scattering geometry the hanle effect of a magnetic field inclined with respect to the solar local vertical creates linear polarization .
we have investigated theoretically the sensitivity of the scattering polarization @xmath3 and @xmath4 profiles of the ca ii ir triplet to the presence of magnetic fields through the hanle effect . to this end , we have applied a multilevel radiative transfer code ( see appendix a ) for treating the transfer of spectral line polarization due to scattering in weakly magnetized stellar atmospheres .
we find that the ca ii 8542 and 8662 lines are mostly sensitive to fields between 0.001 and 0.1 g , the former being also weakly sensitive to fields between 1 and 10 g. otherwise the linear polarization signals are dependent only on the magnetic field geometry , but not to its intensity .
the 8498 line is highly sensitive in the whole regime 0.00110 g , but its scattering polarization amplitude is significantly lower in the absence of magnetic fields .
probably , the most interesting line of the ca ii ir triplet because of its diagnostic potential is the strongest one : @xmath192 .
its linear polarization is sensitive to the orientation of the magnetic field in the chromosphere of the quiet sun , though not too much to its strength unless @xmath193 g there .
although its scattering polarization amplitude depends on the thermal structure of the solar chromosphere , the emergent @xmath3 and @xmath4 profiles do not change sign between the ( hot " ) fal - c model and the ( cool " ) m - co model , unlike @xmath194 , which is very sensitive to the physical conditions of the lower chromosphere .
the thermal sensitivity of the @xmath195 line ( which is sensitive only to milligauss field strengths ) is similar to that of the 8542 line .
therefore , spatial fluctuations in the sign of the emergent @xmath4 profiles of the ca ii 8542 and 8662 lines may be ascribed more safely to spatial variations in the azimuth of the chromospheric magnetic field than is the case for @xmath194 . more generally , the emergent @xmath3 and @xmath4 profiles are produced by the joint action of atomic level polarization , and the hanle and transverse zeeman effects . in this paper we have considered only the action of atomic level polarization and the hanle effect .
these effects dominate the emergent linear polarization profiles for inclined magnetic fields with strengths weaker than @xmath196 , where the @xmath196 value depends on the scattering geometry . in the forward scattering geometry of a disk center observation ( @xmath179 ) the linear polarization of the ca ii ir triplet
is dominated by the hanle effect if the magnetic field is weaker than about 10 g. in fact , while the contribution of the transverse zeeman effect is negligible for @xmath197 g the hanle effect creates fractional linear polarization signals of the order of @xmath180 already for horizontal magnetic fields as weak as 0.1 g ( see figs . 6 and 7 ) . for magnetic strengths @xmath198 g the contribution of the transverse zeeman effect to the linear polarization observed at the solar disk center should not be neglected .
detection of @xmath3 and @xmath4 disk center signals caused either by the hanle effect alone ( if @xmath199 g ) or by the joint action of the hanle and transverse zeeman effects ( if @xmath198 g ) would require very high polarimetric sensitivity together with a spatial and temporal resolution sufficient to at least resolve partially the magnetic field azimuth .
given the weakness of the theoretical scattering polarization signals ( @xmath200 at @xmath179 ; @xmath201 at @xmath172 ) , the most favourable geometry for observing them is in quiet regions close to the limb . in this geometry the emergent linear polarization of the ca ii
ir triplet is non - zero even if the magnetic field has a random azimuth within the spatio - temporal resolution element of the observation ( see figs .
8 and 9 ) .
moreover , for magnetic strengths sensibly weaker than 100 g the contribution of the transverse zeeman effect to the emergent linear polarization is expected to be smaller than that caused by atomic level polarization . for
stronger fields ( e.g. , @xmath202 g ) the wings of the emergent @xmath3 and @xmath4 profiles may show sizable features produced by the transverse zeeman effect , but the line - center amplitudes would still be dominated by atomic level polarization . finally , we point out that information on the relative amplitudes and signs of the scattering polarization profiles of the ca ii ir triplet would be very useful to help constrain the thermal and magnetic structure of the quiet solar chromosphere . in principle , to determine the structure of the magnetic field via stokes inversion techniques is possible , although it would be easier if one is able to infer first the atmospheric thermal structure from the observed intensity spectrum .
the @xmath3 and @xmath4 signals we have studied here should also be exploited to investigate the reliability of three - dimensional models of the solar chromosphere resulting from holistic magnetohydrodynamic ( mhd ) simulations of the solar atmosphere . to this end
, it suffices with confronting spectropolarimetric observations of the ca ii ir triplet with synthetic stokes profiles obtained by solving the non - lte problem of the 2nd kind in 3d snaphot models resulting from such mhd simulations .
modeling the spectral line polarization produced by atomic level polarization and its modification by the hanle effect requires calculating , for multilevel systems , the excitation and ionization state of chemical species of given abundance that is consistent with both the intensity and polarization of the radiation field generated within the ( generally magnetized ) plasma under consideration .
this so - called non - lte problem of the 2nd kind ( cf .
, landi deglinnocenti & landolfi 2004 ) is a very involved non - local and non - linear radiative transfer problem which requires solving the statistical equilibrium equations for the elements of the atomic density matrix and the stokes - vector transfer equation for each of the allowed transitions in the multilevel model .
once such a self - consistent excitation state is known throughout the medium , it is then straightforward to solve the transfer equation for any desired line of sight in order to obtain the emergent stokes profiles to be compared with spectropolarimetric observations . to this end
, we developed a general computer program , which we will describe in greater detail in a future publication along with the efficient and accurate radiative transfer methods on which it is based ( for a first advance see the workshop contribution by manso sainz & trujillo bueno 2003 ; see also manso sainz 2002 ) . in this appendix
we present only a brief summary of the code with emphasis on its application to solve the problem of scattering polarization and the hanle effect in the ir triplet of ionized calcium . 1 .
the equations are formulated within the spherical tensors representation of the density matrix and radiation field tensor .
there are @xmath203 unknowns @xmath74 with @xmath42 even , for each level with integer angular momentum @xmath18 , and @xmath204 unknowns for each level with semi - integer angular momentum @xmath18 .
2 . the radiative transfer equations for the stokes parameters are integrated along short characteristics using the quasi - parabolic delo method ( delopar ; trujillo bueno 2003 ) .
the iterative corrections for the unknowns ( the statistical tensors @xmath74 ) are calculated applying a suitable generalization to the multilevel atom case of the fast iterative methods described in trujillo bueno & manso sainz ( 1999 ) .
this implies writing down the statistical equilibrium equations taking explicitly into account the contribution of the diagonal components of the @xmath205 operator and linearizing according to eqs .
( 53 ) and ( 54 ) of trujillo bueno ( 2003 ) .
the iterative scheme can speeded up further through polynomial ng - acceleration ( e.g. , auer 1987 ) .
the number density of the ion under consideration is computed in non - lte . at each point in the atmosphere
we solve the statistical equilibrium equations for the multipolar components of the atomic density matrix plus the conservation of particles equation to calculate the excitation state of the atomic system .
we calculate the radiation field at each point in the model atmosphere for each of the allowed radiative transitions in the model atom by formal integration of the radiative transfer equations for the stokes parameters applying the delopar method . then , radiation field tensors @xmath131 are calculated and used to work out the new excitation state of the atomic system and so on , iteratively .
figure 16 shows the convergence behavior of the above - mentioned iterative method for the numerical solution of the ( ca ii ) non - lte problem of the second kind . | the chromosphere of the quiet sun is a very important stellar atmospheric region whose thermal and magnetic structure we need to decipher for unlocking new discoveries in solar and stellar physics . to this end
, we need to identify and exploit observables sensitive to weak magnetic fields ( @xmath0 g ) and to the presence of cool and hot gas in the bulk of the solar chromosphere . here
we report on an investigation of the hanle effect in two semi - empirical models of the quiet solar atmosphere with different chromospheric thermal structures .
our study reveals that the linear polarization profiles produced by scattering in the ca ii ir triplet have thermal and magnetic sensitivities potentially of great diagnostic value .
the linear polarization in the 8498 line shows a strong sensitivity to inclined magnetic fields with strengths between 0.001 and 10 g , while the emergent linear polarization in the 8542 and 8662 lines is mainly sensitive to magnetic fields with strengths between 0.001 and 0.1 g. the reason for this is that the scattering polarization of the 8542 and 8662 lines , unlike the 8498 line , is controlled mainly by the hanle effect in their ( metastable ) lower levels .
therefore , in regions with magnetic strengths sensibly larger than 1 g , their stokes @xmath1 and @xmath2 profiles are sensitive only to the orientation of the magnetic field vector .
we also find that for given magnetic field configurations the sign of the @xmath3 and @xmath4 profiles of the 8542 and 8662 lines is the same in both atmospheric models , while the sign of the linear polarization profile of the 8498 line turns out to be very sensitive to the thermal structure of the lower chromosphere .
we suggest that spectropolarimetric observations providing information on the relative scattering polarization amplitudes of the ca ii ir triplet will be very useful to improve our empirical understanding of the thermal and magnetic structure of the quiet chromosphere . |
the possible existence and properties of a narrow dibaryon @xmath0 are discussed widely in the last years both from experimental [ 1,2,3 ] and theoretical [ 4,5 ] point of view .
being bosons such particles could condense in nuclear matter under proper conditions .
this phenomenon happens when the nucleon chemical potential grows up to half of the mass of the dibaryon .
dibaryons are weakly produced due to the reactions @xmath1 which put them in chemical equilibrium with protons and neutrons . at zero temperature the condensation of dibaryons in nuclear ( @xmath2 ) and neutron matter
was studied in refs.[6,7,8 ] . for the case of @xmath3
the condensation starts when the baryon number density @xmath4 grows up to approximately @xmath5 ( here @xmath6 is the normal nuclear density ) .
such conditions ( zero temperatures and high densities ) are hard to imagine in terrestrial experiments . in neutron
stars the densities of the central core can be even larger but it is difficult to see any clear signal [ 7 ] of the dibaryon condensation in neutron stars .
high densities of nuclear matter can in principle occur in heavy ion collisions ( hic ) .
but in that case not only the densities but also the temperatures are high .
so we should extend the consideration of the dibaryon condensation to finite temperatures .
secondly we should take into account the possible phase transition of nuclear matter into quark matter .
below we will describe our model of nuclear matter with dibaryons in section 2 , consider the thermodynamical quantities of nuclear matter in section 3 ( condensation curve ) , describe the model for quark matter in section 4 ( phase transition ) and speculate on possible signatures of dibaryons in hic in the conclusions .
we consider an extension of the walecka model [ 8 ] by including dibaryon fields to the lagrangian density @xmath7 here , @xmath8 is the nucleon field , @xmath9 and @xmath10 are the @xmath11- and @xmath10-meson fields , @xmath12 is the field strength tensor of the vector field ; @xmath13 is the dibaryon field , for which we assume that it is a isoscalar - pseudoscalar field .
this assumption includes the interesting case of the @xmath14-dibaryon .
the values @xmath15and @xmath16 are the @xmath11- and @xmath10-meson masses and @xmath17 , @xmath18 , @xmath19 , @xmath20 are the coupling constants of the @xmath11- and @xmath10-mesons with nucleons ( @xmath21 ) and dibaryons ( @xmath22 ) . the field operators can be expanded into @xmath23-number- and operator parts : @xmath24 the @xmath23-number parts of the fields @xmath25 @xmath10 , @xmath9 , @xmath13 , and @xmath26 are defined as expectation values @xmath27 over the ground state of the system .
the average values of the operator parts are zero by definition : @xmath28 . the @xmath10-meson mean field determines the effective nucleon and dibaryon masses in the medium @xmath29
@xmath30 the nucleon vector and scalar densities are defined by expectation values @xmath31 @xmath32 the vector and scalar density of the dibaryons are defined by @xmath33 @xmath34 neglecting the operator parts of the meson fields , we get the following expressions for the meson mean fields @xmath35 @xmath36 the nucleon vector and scalar densities are given by @xmath37 @xmath38 the statistical factor @xmath39 @xmath40 for nuclear ( neutron ) matter . here
@xmath41 is the nucleon chemical potential ( excluding electrostatic energy of nucleon in the vector field ) @xmath42 .
the dibaryon vector and scalar densities are given by @xmath43 @xmath44 here @xmath45 is the dibaryon statistical factor ( 1 for @xmath3 ) , @xmath46 is the dibaryon chemical potential ( excluding electrostatic energy of dibaryon in the vector field ) @xmath47 , @xmath48 , @xmath49 is the density of dibaryons in the condensate that is formed when @xmath50 .
the chemical equilibrium between nucleons and dibaryons means @xmath51 ( we consider the case when electrostatic energy is not changed in the process of transition of two nucleons to dibaryon ( @xmath52 ) ) .
the self - consistency equations have the form @xmath53 @xmath54 the total baryon number density equals @xmath55 we use here the following procedure of solving the self - consistency equations .
we start with fixing the effective neucleon mass @xmath56 .
the value of effective dibaryon mass then simply follows from eqs .
( [ 19],[20 ] ) @xmath57 further we assume that the condensation takes place and find the nucleon chemical potential : @xmath58 .
then we find the condensate dibaryon density @xmath49 from eq.([19 ] ) .
if it is positive the self - consistency equation is solved for a given point .
if it is negative we take @xmath59 and scan @xmath60 from @xmath61 to @xmath62 to make the eq.([19 ] ) true . after this procedure
all the parameters of the matter are defined and we are ready to calculate various thermodynamical quantities .
the energy density @xmath63 given by average value of the @xmath64 component has the form @xmath65 the last two terms are here the contributions of the classical @xmath11- and @xmath10-meson fields to the energy density . the hydrostatic pressure @xmath66 has the form @xmath67 because dibaryons in the condensate are at rest they do not contribute to the pressure .
now we are in a position to present the results for the thermodynamical quantities of nuclear matter . for our purpose
the most interesting thing is the condensation curve i.e. the curve on the temperature- baryon density plane where the condensation of dibaryons starts either at increasing the baryon density or at lowering the temperature .
the calculations are made for the standart values of rmf model parameters[9 ] .
the coupling constants of the dibaryon @xmath68 to @xmath69 and @xmath70 mesons were taken to be[8 ] @xmath71 the results are shown on the fig.1 ( condensation curve ) .
it means e.g. that at zero temperature ( @xmath72 ) one has to compress to @xmath73 ( @xmath74 of saturation density ) to obtain the transition to the dibaryon condensate if the dibaryon mass is @xmath75 .
also we show on fig.1 the so called critical curve .
it is the curve that shows the boundary of applicability of our model .
precisely it is the place where the effective mass of the nucleon goes to zero . on the right of the critical curve
our model can not be applied .
we hope that the more elaborated model will move this curve to the right in the unphysical region where the nuclear matter transforms to the quark matter for example . and
finally we show on fig.1 the phase transition curve i.e. the place where the nuclear matter ( with or without condensed dibaryons ) transforms to quark matter .
now we are coming to the description of this curve .
the nuclear matter we are considering is isotopically symmetric ( @xmath76 ) and has no net strangeness(@xmath77 ) . at the beginning of the phase transition to quark matter isospin and strangeness are conserved .
therefore we will compare the thermodynamical quantities of nuclear matter with those of quark matter consisting of equal number of @xmath78 and @xmath79 quarks and no net strange quarks .
the reliable temperatures in our consideration ( we are interested in nuclear matter with condensed dibaryons ) will be below @xmath80 .
for such temperatures the contribution of pions , kaons etc . to the thermodynamical quantities of nuclear matter and the contribution of strange quarks , gluons etc . to the thermodynamical quantities of quark matter can be neglected .
then the equation of state ( eos ) of nuclear matter @xmath81 is described by eq .
( [ 27 ] ) and the equation of state of quark matter @xmath82 is the following one[11 ] @xmath83 here @xmath84 is the constant of gluonic interaction of quarks . the first term in eq.([31 ] ) is the pressure of massless @xmath78 and @xmath79 quarks , the second term is the plasmon contribution to the pressure and the third term is the vaccuum pressure that is responsible for confinement ( mit bag model[11 ] ) .
the constants @xmath84 and @xmath85 were taken to be @xmath86 that ensures the stability of massive neutron stars against the transition to strange stars[12 ] .
the phase transition of nuclear matter to quark matter starts when the pressure of quark matter becomes equal to the pressure of nuclear matter at equal baryon chemical potentials and temperatures ( gibbs criterion , here we assume that the phase transition is of the first order ) @xmath87 the typical dependences of @xmath88 and @xmath89 on @xmath90 for fixed @xmath91 are shown on fig.2 .
the point of crossing of lines is the point of phase transition at given temperature .
the full phase transition curve is shown on fig.1 .
as it follows from fig.1 at temperatures above @xmath92 the dibaryon condensation does not occur : increasing the density for fixed temperatures above @xmath92 we come to the quark matter rather than to dibaryon condensation .
the region on the ( @xmath93 ) plot where dibaryon condensate is formed ( @xmath94 in fig.1 ) lies therefore below @xmath92 for temperatures and above @xmath95 for densities ( 2.7 of normal nuclear density ) . in conclusion
let us speculate on the possible signatures of dibaryon condensate formation in hic .
the accessible temperatures and densities of nuclear matter in hic are not very well known .
fig.3 shows our calculations of the temperature of compressed nuclear matter formed in hic as a function of the density of compressed matter ( different energies per nucleon of colliding ions ) .
we have used the eos of nuclear matter with dibaryons and a one- dimensional shock wave model[13 ] ( rankin - hugoniot - taub equation ) .
the compression curve lies above the region of condensation ( in @xmath96 or @xmath97 ) shown on fig.1 . on the first sight we should conclude that the dibaryon condensate never forms in hic .
but the compressed matter being formed evolves further and its evolution is governed by two effects : by the cooling due to emission of particles ( mainly pions ) and by the decompression ( that also results in the cooling ) . if the radiation cooling prevails over the cooling due to the decompression , the matter could evolve into the region of the formation of the dibaryon condensate ( @xmath94 in fig.1 ) .
if so , the significant part of nucleons will transform to dibaryons ( up to @xmath98 ) .
the dibaryons could survive in the process of decompression ( @xmath99 ) and then decay as free particles in one of the following ways @xmath100 the pions in these decays have the spectrum with the characteristic kinetic energy about @xmath101 , if the dibaryon has a mass about @xmath102 .
then these pions could form an excess in the temperature spectrum of all pions produced in hic . in order to these pions
could be observed they should not be thermalised in the nuclear matter like the other pions ( direct ones , pions from different other resonances not considered in this paper ) .
the principal possibility of this is due to the relatively long lifetime of @xmath3-dibaryon - it could decay after the decay of nuclear matter .
so , we conclude , there is a region of temperatures and densities where the dibaryon condensate forms in nuclear matter . if formed in hic , dibaryons could give a signature in form of @xmath101 kinetic energy pions -the products of their decays .
the authors acknowledge the discussions with drs . d.kosov and l.sehn .
two of us ( m.i.k and b.v.m . )
are grateful to the institute for theoretical physics of university of tuebingen for hospitality and financial support .
this work was supported by the deutsche forschungsgemeinschaft under contract @xmath103 fa67/20 - 1 .
99 r.bilger et al
. phys.lett . *
b269 * ( 1991 ) 247 .
r.bilger , h.a.clement and m.g.schepkin , phys.rev.lett .
* 71 * ( 1993 ) 42 .
h.clement , m.schepkin , g.j.wagner and o.zaboronsky , phys.lett .
* b337 * ( 1994 ) 43 .
l.ya.glozman , a.buchmann and a.faessler , j.phys . *
g20 * ( 1994 ) l49 .
g.wagner , l.ya.glozman , a.j.buchmann and amand faessler , nucl.phys .
* a594 * ( 1995 ) 263 .
a.m. baldin et.al .
, dokl.acad.sc .
279 * ( 1984 ) 602 ; + st .
mrowczynski , phys.lett.*b152 * ( 1985 ) 299 ; + a.v.chizov et.al .
, nucl.phys . *
a449 * ( 1986 ) 660 .
m.i.krivoruchenko , jetp letters * 46 * ( 1987 ) 3 ; + r.tamagaki , progr.theor.phys .
* 85 * ( 1991 ) 321 ; + a.olinto , p.haensel and j.frieman , preprint fermilab - pub-91 - 176-a , 1991 .
amand faessler , a.buchmann , m.i.krivoruchenko and b.v.martemyanov , phys.lett .
* b391 * ( 1997 ) 255 .
j.d.walecka , ann .
( n.y . ) * 83 * ( 1974 ) 491 .
b.a.freedman and l.d.mclerran , phys.rev .
* d16 * ( 1978 ) 1130 , 1147 , 1169 .
a.chodos , r.l.jaffe , k.johnson , c.b.thorn and v.f.weisskopf , phys.rev . *
d9 * ( 1974 ) 3471 .
m.i.krivoruchenko and b.v.martemyanov , astrophys .
j. * 378 * ( 1991 ) 628 ; nucl.phys .
* b24 * ( 1991 ) 134c .
h.stcker , w.greiner , w.scheid , z.phys . *
a286 * ( 1978 ) 121 . * figure captions * + * fig.1 * the condensation , phase transition and critical curves on the temperature(@xmath104 ) - baryon number density(@xmath105 ) plot
. the regions of nuclear matter(@xmath106 ) , nuclear matter with condensed dibaryons(@xmath107 ) and quark matter(@xmath108 ) are marked .
the critical curve shows the boundary of applicability of our model . + * fig.2 * the example of nuclear and quark matter equations of state .
the pressures of nuclear matter(@xmath88 ) and quark matter(@xmath89 ) are shown as the functions of baryon chemical potential(@xmath90 ) at the temperature @xmath109 .
the crossing point is the point of phase transition from nuclear matter to quark matter . + * fig.3 * the temperature of the shock wave(@xmath104 ) as a function of baryon density ( @xmath110 ) in ion collisions for mean field model of nuclear matter with dibaryons . | we consider the thermodynamics of the matter made of equal number of neutrons and protons and of scalar dibaryons .
they interact via the exchange of scalar and vector mesons .
the interaction is taken into account in the mean field approximation .
the condensation of dibaryons in this matter and the phase transition of matter to quark matter are considered .
possible signatures of dibaryons in heavy ion collisions are speculated on . + * pacs*:14.20.pt , 21.65.+f * keywords * : nuclear matter , dibaryon , condensation = -1.3truecm 16 truecm # 1#1 # 1#1 # 1 # 1 epsf |
graphene on hbn devices were fabricated using two different methods . in the first method , mechanically exfoliated graphene was transferred to high quality single crystals of hbn which were mechanically exfoliated on a sio@xmath40 substrate @xcite . in the second method , commercially available hbn ( momentive ac6004 ) was exfoliated on sio@xmath40 substrates and then cvd grown graphene was deposited over the hbn .
both types of devices gave similar results so we do not distinguish between the two types .
after depositing the graphene on hbn , cr / au electrodes were written using electron beam lithography .
the devices were annealed at 350@xmath30c for 2 hours in a mixture of argon and hydrogen and then at 300 @xmath30c for 1 hour in air before being transferred to the uhv lt - stm for topographic and spectroscopic measurements .
we have measured a total of 29 samples of which 7 had long enough moir patterns for the observation of a superlattice dirac point .
figure 1a shows a schematic diagram of the measurement setup used for imaging and spectroscopy of the graphene flakes .
all the measurements were performed in uhv at a temperature of 4.5 k. di / dv measurements were acquired by turning off the feedback circuit and adding a small ( 5 - 10 mv ) ac voltage at 563 hz to the sample voltage .
the current was measured by lock - in detection .
99 ashcroft , n. w. and mermin , n.d .
_ solid state physics _
( brooks cole , 1976 ) park , c .- h .
, yang , l. , son , y .- w . ,
cohen , m. l. , and louie , s. g. anisotropic behaviours of massless dirac fermions in graphene under periodic potentials .
_ nature phys . _ * 4 * , 213 - 217 ( 2008 ) ; park , c .- h .
, yang , l. , son , y .- w . , cohen , m. l. , and louie , s. g. new generation of massless dirac fermions in graphene under external periodic potentials .
lett . _ * 101 * , 126804 ( 2008 )
. xue , j. _ et al .
_ scanning tunnelling microscopy and spectroscopy of ultra - flat graphene on hexagonal boron nitride . _
nature mater . _
* 10 * , 282 - 285 ( 2011 ) .
decker , r. _ et al . _ local electronic properties of graphene on a bn substrate via scanning tunneling microscopy .
_ nano letters _ * 11 * , 2291 - 2295 ( 2011 ) .
castro neto , a. h. , guinea , f. , peres , n. m. r. , novoselov , k. s. , and geim , a. k. the electronic properties of graphene .
phys . _ * 81 * , 109 - 162 ( 2009 ) .
katsnelson , m. i. , novoselov , k. s. and geim , a. k. chiral tunnelling and the klein paradox in graphene .
_ nature phys . _
* 2 * , 620 - 625 ( 2006 ) .
klein , o. , die reflexion von elektronen an einem potentialsprung nach der relativistischen dynamik von dirac
_ z. phys . _
* 53 * , 157 - 165 ( 1929 ) .
stander , n. , huard , b. and goldhaber - gordon , d. evidence for klein tunneling in graphene _ p - n _ junctions .
_ phys . rev .
lett . _ * 102 * , 026807 ( 2009 ) .
young , a. f. and kim , p. quantum interference and klein tunnelling in graphene heterojunctions .
_ nature phys .
_ * 5 * , 222 - 226 ( 2009 ) .
barbier , m. , peeters , f. m. , vasilopoulos , p. , and pereira , j. m. dirac and klein - gordon particles in one - dimensional periodic potentials .
b _ * 77 * , 115446 ( 2008 ) .
brey , l. and fertig , h. a. emerging zero modes for graphene in a periodic potential .
_ * 103 * , 046809 ( 2009 ) ; sun , j. , fertig , h. a. , and brey , l. effective magnetic fields in graphene superlattices .
lett . _ * 105 * , 156801 ( 2010 ) .
burset , p. , levy yeyati , a. , brey , l. , and fertig , h. a. transport in superlattices on single - layer graphene .
b _ * 83 * , 195434 ( 2011 ) .
ortix , c. , yang , l. and van den brink , j. graphene on incommensurate substrates : trigonal warping and emerging dirac cone replicas with halved group velocity .
arxiv:1111.0399 , ( 2011 ) .
marchini , s. , gnther , s. , and wintterlin , j. scanning tunneling microscopy of graphene on ru(0001 ) .
b _ * 76 * , 075429 ( 2007 ) . vsquez de parga , a. l. _ et al . _
periodically rippled graphene : growth and spatially resolved electronic structure .
* 100 * , 056807 ( 2008 ) .
dean , c. r. _ et al . _
boron nitride substrates for high - quality graphene electronics .
_ nature nanotech . _ * 5 * , 722 - 726 ( 2010 ) .
dagotto , e. correlated electrons in high - temperature superconductors .
_ * 66 * , 763 - 840 ( 1994 ) .
the work at arizona was partially supported by the u. s. army research laboratory and the u. s. army research office under contract / grant number w911nf-09 - 1 - 0333 and the national science foundation career award dmr-0953784 , eecs-0925152 and dmr-0706319 .
j.s - y . and p.j - h .
were primarily supported by the us department of energy , office of basic energy sciences , division of materials sciences and engineering under award de - sc0001819 and partly by the 2009 us office of naval research multi university research initiative ( muri ) on graphene advanced terahertz engineering ( gate ) at mit , harvard and boston university . p.j .
acknowledges the support of the swiss center of excellence manep .
m.y . , j.x . ,
d.c . and b.j.l .
performed the stm experiments of the graphene on hbn .
m.y . and d.c .
fabricated the cvd graphene devices .
j. s .- y .
fabricated the devices on single crystal hbn .
k.w . and t.t .
provided the single crystal hbn .
p.j . performed the theoretical calculations .
p. j .- h . and b.j.l .
conceived and provided advice on the experiments .
all authors participated in the data discussion and writing of the manuscript .
the authors declare no competing financial interests .
scattering at the brillouin zone boundaries due to the periodic potential does not open a gap for chiral massless dirac fermions as long as sublattice symmetry is preserved .
this can be seen by performing a unitary transformation @xmath41 on the hamiltonian of eq .
( 3 ) , with @xmath42 $ ] . when @xmath43 the linear part of the potential can be gauged out of the hamiltonian for @xmath44 . for @xmath45 ,
the gauge transformation can no longer remove the linear part of the potential , and the potential creates new dirac points but still does not open a gap .
@xcite .
the magnitude of the dips in the local density of states ( ldos ) is determined by the strength and the wavelength of the moir periodic potential .
we see this perturbatively by projecting the hamiltonian of eq .
( 3 ) onto pairs of eigenstates @xmath46)^\dagger \exp[i { \bf k } { \bf r}]/\sqrt{2 \omega}\ ] ] of @xmath47 , with the lattice area @xmath48 and the angle @xmath49 .
the resulting @xmath50 hamiltonian reads @xmath51 with @xmath52 ) \sum_\alpha ( \delta_{{\bf k},{\bf g}_\alpha+{\bf k } ' } + \delta_{{\bf k},-{\bf g}_\alpha+{\bf k}'})$ ] .
we focus on momenta @xmath53 , @xmath54 . for this choice of pairs of eigenstates
, it is straightforward to see that the eigenvalues of @xmath55 are @xmath56 , with @xmath57 , and the component @xmath58 of @xmath59 that is perpendicular to @xmath28 . in particular
, there is no energy shift for @xmath60 and a maximal shift for @xmath61 .
this perturbative result is consistent with the numerical and experimental facts reported here that ( i ) no gap is opened , neither in the numerically obtained @xmath62 , nor in the experimentally obtained stm @xmath63 , ( ii ) a reduction of ( dip in ) the density of states is observed at an energy corresponding to @xmath64 , and ( iii ) both the spectroscopic moir pattern and the dips tend to disappear for periodic potentials with shorter wavelength ( hence larger g ) , when the relative rotation between the hbn substrate and the graphene layer is larger .
the reduction of the density of states occurs around three points determined by the three reciprocal lattice vectors . at this level
, there is no @xmath65-dependence of the strength of the dip , the latter is the same for @xmath66 ( @xmath67 ) as for @xmath68 ( @xmath69 ) .
an @xmath65-dependence emerges once off - diagonal terms @xmath70 are included in the hamiltonian of eq .
we consider a lattice hamiltonian for a graphene monolayer on a single layer of hbn . for each layer
, the hamiltonian reads @xmath71
-t_\alpha \sum_{\langle i , j \rangle } ( a^\dagger_i(\alpha ) b_j(\alpha ) + h.c . ) \ , , \end{aligned}\ ] ] where @xmath72 [ @xmath73 and @xmath74 [ @xmath75 are creation [ destruction ] operators on sublattice a and b , respectively , of the graphene ( @xmath76 ) or hbn ( @xmath77 ) honeycomb lattice , and @xmath78 indicates that the sum runs only over nearest neighbors . on the hbn lattice , we choose the boron atoms to be on the a sublattice and the nitrogen atoms to be on the b sublattice .
the on - site energies are @xmath79 , @xmath80 ev and @xmath81 ev , and the hopping integrals are @xmath82 ev and @xmath83 ev @xcite ( the precise value of the latter is of little importance ) .
we restrict the interlayer hopping potential to nearest - neighbor and next - nearest - neighbor hopping .
the interlayer hopping is given by @xmath84 \
, f_{ij } \ , , \end{aligned}\ ] ] with a characteristic function @xmath85 if site @xmath86 of sublattice @xmath87 on the graphene ( hbn ) sheet is nearest or next - nearest neighbor to site @xmath88 of sublattice @xmath89 on the hbn ( graphene ) sheet , and @xmath90 otherwise . the parameters @xmath91 ev and @xmath92 nm are calibrated to fit the interlayer couplings in bilayer graphene@xcite .
the two lattices are rotated with respect to one another by the angle @xmath6 , and for each site we determine its nearest and next nearest neighbor site on the other sheet , numerically evaluate the distance between the sites and finally the corresponding interlayer hopping . while @xmath93 should in principle depend on the sublattice index in the hbn layer , we neglect this dependence here .
the interlayer coupling hamiltonian is then given by @xmath94 + t_{ij } ' ( a , b ) [ a^\dagger_i(1 ) b_j(2 ) + h.c . ]
\nonumber \\ & & \left .
\;\;\;\;\ ; + t_{ij } ' ( b , a ) [ b^\dagger_i(1 ) a_j(2 ) + h.c . ] + t_{ij } ' ( b , b ) [ b^\dagger_i(1 ) b_j(2 ) + h.c.]\right ] \ , .\end{aligned}\ ] ] second - order perturbation theory maps @xmath95 onto a periodic potential of hexagonal symmetry , similar to the one in eq .
( 3 ) , with a modulation amplitude @xmath96 ev which we determined numerically via second order perturbation theory in the interlayer hopping .
this energy is smaller than @xmath97 with the moir superlattice vector @xmath1 , regardless of the rotation angle between graphene and hbn sheets .
the total hamiltonian reads @xmath98 .
we evaluate the ldos @xmath99 on the graphene sheet using the lanczos method , which allows to reach linear system sizes of @xmath100 or more sites @xcite .
the obtained @xmath14 depends on a smearing parameter @xmath101 which , as long as the density of states of the stm tip and the tunneling rate from the tip to the sample do not depend on energy can be related to the strength of the tip - graphene coupling .
figure [ fig : tipstrength](a ) illustrates how reducing this coupling allows finer and finer structures in the ldos to be explored . for this particular rotation of @xmath102 ,
corresponding to a moir pattern with @xmath103 nm , and the set of parameters we just discussed , we see that the ldos vanishes more or less linearly close to @xmath104 .
our numerical data suggest a complete , linear vanishing of the ldos there .
the linear vanishing of the ldos , gives an estimate for the fermi velocity close to the three new superlattice dirac points .
we estimate that the new velocity for the holes is @xmath105 . the new velocity for 1000 graphene lattice on a hbn sheet , with relative rotation @xmath102 .
the smearing parameter of the lanczos algorithm is @xmath106 ( black curve ) , 0.04 ( red ) , 0.018 ( green ) , 0.008 ( blue ) and 0.0036 ( violet ) .
the dashed lines give linear fits close to the dirac and superlattice dirac points .
( b ) ldos for a 1000@xmath1071000 graphene lattice on a hbn sheet , with relative rotation @xmath108 , for @xmath109 .
the black curve has nearest - neighbor interlayer coupling , the red curve has nearest and next nearest neighbor interlayer hopping.,width=453 ] electrons is smaller because of the increased slope and we estimate it to be @xmath110 .
the new reduced fermi velocities are in reasonable agreement with our experimentally determined values .
a significant energy asymmetry emerges in that the expected dip in the density of states is stronger in the valence @xmath69 than in the conduction @xmath67 band .
this asymmetry arises because of ( i ) the asymmetry in the on - site energies in the hbn layer , which effectively induces a second order potential that is stronger for negative than for positive energies and ( ii ) next - nearest neighbor interlayer hopping , which effectively induces a local periodic modulation of @xmath111 .
this is illustrated in fig .
[ fig : tipstrength](b ) .
we have also performed electrical transport measurements on the graphene device shown in fig .
the conductivity as a function of gate voltage is shown in fig .
[ fig : transport ] .
the gate voltage is plotted as the offset from the gate voltage at the dirac point .
we observe two locations of decreased conductivity which are located at approximately @xmath112 v from the dirac point .
this is the same separation in gate voltage as observed in the spectroscopy measurements , where the main dirac point was separated from the superlattice dirac point by 40 v. therefore , we conclude that the dips in conductivity are due to the presence of the superlattice dirac point .
this gives further indirect evidence of the superlattice dirac point .
we also see a second dip in the conductivity near -50 v. our stm topography measurements showed a region of the device with a second moir pattern of about 10 nm .
the superlattice dirac point due to this moir occurs at a higher energy and hence a larger gate voltage . | the schrdinger equation dictates that the propagation of nearly free electrons through a weak periodic potential results in the opening of band gaps near points of the reciprocal lattice known as brillouin zone boundaries @xcite .
however , in the case of massless dirac fermions , it has been predicted that the chirality of the charge carriers prevents the opening of a band gap and instead new dirac points appear in the electronic structure of the material @xcite .
graphene on hexagonal boron nitride ( hbn ) exhibits a rotation dependent moir pattern @xcite . in this letter
, we show experimentally and theoretically that this moir pattern acts as a weak periodic potential and thereby leads to the emergence of a new set of dirac points at an energy determined by its wavelength .
the new massless dirac fermions generated at these superlattice dirac points are characterized by a significantly reduced fermi velocity .
the local density of states near these dirac cones exhibits hexagonal modulations indicating an anisotropic fermi velocity .
due to its hexagonal lattice structure with a diatomic unit cell , graphene has low - energy electronic properties that are governed by the massless dirac equation @xcite .
this has a number of consequences , among them klein tunneling @xcite which prevents electrostatic confinement of charge carriers and inhibits the fabrication of standard semiconductor devices .
this has motivated a number of recent theoretical investigations of graphene in periodic potentials @xcite , which explored ways of controlling the propagation of charge carriers by means of various superlattice potentials . on the analytical side ,
one - dimensional potentials render particle propagation anisotropic @xcite and generate new dirac points , where the electron and hole bands meet , at energies @xmath0 given by the reciprocal superlattice vectors @xmath1 @xcite , and at zero energy for potentials with strengths exceeding @xmath2 @xcite .
numerical approaches have extended several of these results to the case of two - dimensional potentials @xcite . unlike for schrdinger fermions ,
the periodic potentials generally induce new dirac points but do not open gaps in graphene , owing to the chiral nature of the dirac fermions .
recent stm topography experiments have reported well developed moir patterns in graphene on crystalline substrates , which suggests that the latter generate effective periodic potentials @xcite .
of particular interest is hexagonal boron nitride ( hbn ) , because it is an insulator which only couples weakly to graphene .
furthermore , graphene on hbn exhibits the highest mobility ever reported for graphene on any substrate @xcite , and has strongly suppressed charge inhomogeneities @xcite .
hexagonal boron nitride is a layered material whose planes have the same atomic structure as graphene , with a 1.8% longer lattice constant .
the influence of the weak graphene - substrate interlayer coupling on the electronic transport and spectroscopic properties of graphene is not well understood .
in particular , there is to date no theory for local electronic properties such as those probed in stm experiments .
below we show that periodic interlayer couplings generate a new dirac point at an energy determined by the wavevector of the periodic potential .
the presence of this new dirac point is reflected in two dips in the density of states , symmetrically placed at @xmath3 around the @xmath4 graphene dirac point but generally of asymmetric strength .
there is also a periodic modulation of the local density of states , indicating an anisotropic fermi velocity , with the same period as the superlattice topographic moir pattern .
the fabrication procedure used for creating the graphene on hbn devices results in a random rotational orientation between the graphene and hbn lattices .
this rotation between the lattices and the longer lattice constant for hbn leads to topographic moir patterns .
given the lattice mismatch @xmath5 between hbn and graphene , the relative rotation angle @xmath6 between the two lattices uniquely determines the moir wavelength @xmath7 as @xmath8 where @xmath9 is the graphene lattice constant .
the relative rotation angle @xmath10 of the moir pattern with respect to the graphene lattice is given by @xmath11 figure [ fig : topography](b ) plots the wavelength of the moir pattern ( black ) and rotation angle ( red ) as a function of @xmath6 .
due to the lattice mismatch , there is a moir pattern for all orientations of graphene on hbn with a maximum possible length of about 14 nm .
figs .
[ fig : topography ] ( c)-(e ) show stm topography images of moir patterns for three different rotations of the graphene lattice .
( black ) and rotation ( red ) as a function of the angle between the graphene and hbn lattices .
( c)-(e ) stm topography images showing ( c ) 2.4 nm , ( d ) 6.0 nm and ( e ) 11.5 nm moir patterns .
typical imaging parameters were sample voltages between 0.3 v and 0.5 v and tunnel currents between 100 pa and 150 pa .
the scale bars in all images are 5 nm.,width=453 ] we explore how this moir structure influences the local density of states ( ldos ) @xmath12 in the graphene layer .
this quantity dominates di / dv stm measurements as long as the density of states of the stm tip and the tunneling rate from the tip to the sample are constant .
we take a lattice hamiltonian for a graphene monolayer deposited on top of a hbn monolayer , where the two layers are rotated with respect to one another , and the spatially dependent interlayer hopping @xmath13 is calculated from nearest neighbor and next nearest neighbor interlayer coupling @xcite .
we numerically calculate @xmath14 for this model using the lanczos method @xcite ( see supplementary information ) .
dips in the calculated @xmath14 are clearly seen in fig .
[ fig : di - dv](a ) .
the energy of these dips changes as a function of the rotation angle @xmath6 and hence the moir wavelength .
we have also observed the dips in the experimental di / dv curves as shown in fig .
[ fig : di - dv](b ) .
the black curve is for a 9.0 nm moir pattern and the energy of the dip is 0.28 ev from the dirac point .
the red curve is for a 13.4 nm moir pattern and the energy of the dips decreases to 0.22 ev from the dirac point .
both experimentally and theoretically , we found that the relative strength of the dips in the conduction and valence band are different , with the dip in the valence band being much deeper than the dip in the conduction band . in our numerical calculations
, we identified that most of this asymmetry arises due to next nearest neighbor interlayer coupling , which effectively induces modulated hopping between different graphene sublattices and breaks electron - hole symmetry ( see supplementary information ) .
figure [ fig : di - dv](c ) plots @xmath15d@xmath16i / dv@xmath17 for the 9.0 nm moir pattern as a function of gate voltage and sample voltage .
we clearly see the dirac point in this measurement crossing the fermi energy near zero gate voltage .
there is a second dip which moves parallel to it that is offset by -0.28 v in sample voltage .
this dip is due to the superlattice periodic potential induced by the hbn and indicates the emergence of a new superlattice dirac point .
we have observed these dips in the ldos for seven different moir wavelengths .
the energy of the dips from the dirac point is plotted ( red points ) as a function of wavelength in fig .
[ fig : di - dv](d ) . the solid black line plots the expected energy dependence @xmath18 assuming the linear band structure of graphene and @xmath19 m / s . for the necessary high resolution spectroscopy ,
our stm is limited to observing dips in an energy range of @xmath201 v which restricts the moir wavelengths to longer than 2 nm .
( 12.5 nm ) , blue is @xmath21 ( 10.0 nm ) and green is @xmath22 ( 6.3 nm ) .
the curves have been vertically offset for clarity .
( b ) experimental di / dv curves for two different moir wavelengths , 9.0 nm ( black ) and 13.4 nm ( red ) .
the dips in the di / dv curves are marked by arrows .
( c ) @xmath15d@xmath16i / dv@xmath17 as a function of gate and sample voltage for the 9.0 nm moir pattern showing the shift of the dirac point and one of the dips .
( d ) energy of the dips away from the dirac point as a function of moir wavelength .
the red points are the experimentally measured values and the black line is the expected theoretical dependence.,width=453 ] to better understand these dips , we focus on the low - energy regime and neglect intervalley scattering in graphene .
this is justified by the energy range in our stm experiments and the long wavelength of the moir potential .
the interlayer hopping term between the graphene and hbn layers reflects the same periodic structure as the moir pattern @xcite .
therefore we model the influence of the hbn by an effective periodic potential with the same symmetry as the observed moir pattern .
we accordingly consider the single - valley hamiltonian @xmath23 where @xmath24 , @xmath25 is a vector of pauli matrices and @xmath26 is the identity matrix .
the potential strength is estimated as @xmath27 ev from numerical second - order perturbation theory , and the @xmath28 are the reciprocal superlattice vectors corresponding to the periodic potential generated by the hbn substrate .
the reciprocal superlattice vector @xmath29 is determined by the relative rotation of the graphene and hbn lattices according to eqs .
( [ eq : lambda ] ) and ( [ eq : theta ] ) .
the other superlattice wavevectors are obtained by two rotations of 60@xmath30 .
larger superlattice vectors are not included in our model , since the corresponding couplings are smaller by more than one order of magnitude @xcite .
the dips in @xmath14 are due to @xmath31 processes induced by the periodic potential for values @xmath32 corresponding to one of the reciprocal superlattice vectors @xmath28 .
unlike for schrdinger particles , the chirality of the dirac fermions prevents such processes from opening a gap at the edges of the superlattice brillouin zone , as long as the potential does not break sublattice symmetry ( see supplementary information ) .
the presence of the new superlattice dirac points can be detected by examining the gate dependence of the ldos . as seen in fig .
[ fig : di - dv](c ) , the two dirac points move in parallel with gate voltage . figure [ fig : newdp](a ) plots di / dv for a 13.4 nm moir pattern over a larger range of gate voltage .
the white lines show the energy of the two dirac points as a function of gate voltage . when the superlattice dirac point approaches the fermi energy , the gate dependence of the dirac points changes .
this is plotted in fig .
[ fig : newdp](b ) , which tracks the energy of the main dirac point when the superlattice dirac point crosses the fermi energy at -53 v. both dirac points move more quickly with gate voltage , indicating a reduced density of states at this energy . the gate dependence for the main dirac point can be fit with the equation @xmath33 where @xmath34 is the gate voltage , @xmath35 is the offset voltage and @xmath36 is the coupling to the gate .
the factor of three comes from the fact that the periodic potential creates three superlattice dirac points corresponding to the three reciprocal lattice vectors @xmath28 . from the fit
, we extract the value of the fermi velocity at the dirac point to be @xmath37 m / s for both electrons and holes . at the new superlattice dirac points ,
the fermi velocity is reduced to @xmath38 m / s for the new electrons and @xmath39 m / s for the new holes .
the reduction agrees with recent theoretical predictions @xcite and our numerical calculations .
evidence for the presence of the superlattice dirac point can also be seen in the global conductivity as a function of gate voltage ( see supplementary information ) .
in addition to the formation of superlattice dirac points , the presence of the periodic potential also leads to a spatial variation in the ldos .
figure [ fig : moire_scan ] shows experimental and numerical images of the ldos as a function of energy .
near the dirac point , ( figs . [
fig : moire_scan ] ( b ) and ( e ) ) , we observe a nearly featureless density of states .
this is in agreement with previous stm measurements which show a strong suppression of charge fluctuations in graphene on hbn @xcite .
at higher energies , the presence of the moir potential manifests itself as a local variation in the density of states .
figures [ fig : moire_scan ] ( a ) and ( d ) are taken at a lower energy than the superlattice dirac point in the valence band . at this energy
the hexagonal pattern of the potential is clearly visible . on the conduction band side ,
figs .
[ fig : moire_scan ] ( c ) and ( f ) , the moir pattern is once again visible but its contrast is inverted . the centers of the hexagons now correspond to points of increased density of states .
these ldos modulations indicate that the fermi velocity is spatially dependent .
moreover , the inversion indicates that the relative strength of the potential locally determines whether electrons or holes have a higher fermi velocity .
we observe that the moir pattern becomes much more visible in the local density of states at energies above the superlattice dirac points .
previous measurements of the ldos in graphene on hbn did not observe any variations due to the short moir patterns and low energies probed @xcite .
graphene on hbn devices are becoming widely used due to their improved mobility and reduced charged impurities .
we have shown that lattice mismatch and relative rotation between the graphene and hbn leads to a periodic potential for graphene charge carriers .
this potential creates a new dirac point whose energy is determined by the wavelength of the potential .
this superlattice dirac point has the potential to control the transport properties of electrons in graphene as it induces spatially modulated velocities for the charge carriers .
future work is necessary to exploit this periodic potential for the creation of novel graphene devices . |
let @xmath3 be an oriented @xmath4-component link in @xmath5 with @xmath6 . for an oriented knot @xmath7
, we denote the orientation - reversed knot by @xmath8 . if @xmath9 is an orientation - reversing ( orientation - preserving , respectively ) homeomorphism of @xmath5 so that @xmath10 for all @xmath11 where @xmath12 or @xmath13 , and @xmath14 is a permutation of @xmath15 , then @xmath16 is said an _
@xmath17-amphicheiral link _ ( an _ @xmath17-invertible link _ , respectively ) .
a term amphicheiral link " is used as a general term for an @xmath17-amphicheiral link .
a link is said an _
interchangeable link _ if it is an @xmath17-invertible link such that @xmath14 is not the identity .
an @xmath17-invertible link is said an _ invertible link _ simply if there exists @xmath18 such that @xmath19 .
if @xmath14 is the identity , then an amphicheiral link is said a _ component - preservingly amphicheiral link _ , and @xmath14 may be omitted from the notation .
if every @xmath20 is identical for all @xmath11 ( including the case that @xmath14 is not the identity ) , then an @xmath17-amphicheiral link ( an @xmath17-invertible link , respectively ) is said an @xmath21-amphicheiral link ( an @xmath21-invertible link , respectively ) .
we use the notations @xmath22 and @xmath23 .
a link @xmath16 with at least @xmath0-component is said an _ algebraically split link _ if the linking number of every @xmath0-component sublink of @xmath16 is zero .
we note that a component - preservingly @xmath21-amphicheiral link is an algebraically split link .
necessary conditions for the alexander polynomials of amphicheiral knots are studied by r. hartley @xcite , r. hartley and a. kawauchi @xcite , and a. kawauchi @xcite ( cf .
lemma [ lm : hk ] ) . in @xcite , non - invertibility of @xmath24
is firstly proved by the conditions . on the other hand
, t. sakai @xcite proved that any one - variable laurent polynomial @xmath25 over @xmath26 such that @xmath27 and @xmath28 is realized by the alexander polynomial of a strongly invertible knot in @xmath5 .
b. jiang , x. lin , shicheng wang and y. wu @xcite showed that ( 1 ) a twisted whitehead doubled knot is amphicheiral if and only if it is the unknot or the figure eight knot , and ( 2 ) a prime link with at least @xmath0 components and up to @xmath29 crossings is component - preservingly @xmath30-amphicheiral if and only if it is the borromean rings .
they used s. kojima and m. yamasaki s @xmath31-function @xcite .
shida wang @xcite determined prime component - preservingly @xmath30-amphicheiral links with at least @xmath0 components and with up to @xmath1 crossings by the same method as @xcite .
there are four such links ( cf .
theorem [ th : table ] ( 3 ) ) . for geometric studies of symmetries of arborescent knots ,
see f. bonahon and l. c. siebenmann @xcite .
the author @xcite studied necessary conditions for the alexander polynomials of algebraically split component - preservingly amphicheiral links by computing the reidemeister torsions of surgered manifolds along the link ( cf .
lemma [ lm:2-comp ] ) .
the author and a. kawauchi @xcite obtained necessary conditions by invariants deduced from the quadratic forms of a link @xcite , and by using the conditions they showed that the alexander polynomial of an algebraically split component - preservingly @xmath21-amphicheiral link with even components is zero ( cf .
conjecture [ cj : kd2 ] ) and determined amphicheiral links with up to @xmath29 crossings ( cf .
lemma [ lm : sign ] ) .
we determine prime amphicheiral links with at least @xmath0 components and up to @xmath1 crossings . for a link with the crossing number up to @xmath29 , we use the notation of d. rolfsen s book @xcite , and for a link with the crossing number @xmath32 or @xmath1
, we use a slightly modified notation from m. thistlethwaite s table on d. bar - natan and s. morrison s website @xcite ( see section [ sec : table ] ) . in the present paper , we used information of the jones polynomial and the multi - variable alexander polynomial for the class on the website @xcite . for prime links with up to @xmath32 and @xmath1 crossings , firstly we checked a condition on the jones polynomial ( cf .
lemma [ lm : jones ] ) .
most of them are ruled out by the condition .
for the rest links , we applied the same conditions on the alexander polynomial as in @xcite ( cf .
lemma [ lm : amp ] , lemma [ lm : sub ] , lemma [ lm : even ] , lemma [ lm:2-comp ] and lemma [ lm : sign ] ) . @xmath33 and @xmath34 could not be detected not to be amphicheiral by the conditions and the homfly polynomial ( see figure 4 ) .
thus we made the following condition for a special case ( cf .
lemma [ lm : branch ] ) which is inspired by the method in @xcite , and the result of @xcite on the alexander polynomials of amphicheiral knots .
the conditions would be useful in determining the _ link - symmetric group _ of a link ( cf .
@xcite ) . for an @xmath4-component link @xmath35 ,
let @xmath36 be the @xmath4-variable alexander polynomial of @xmath16 which is an element of the @xmath4-variable laurent polynomial ring @xmath37 $ ] over @xmath26 where @xmath38 ( @xmath11 ) is a variable corresponding to a meridian of @xmath39 . for two elements @xmath40 and
@xmath41 in @xmath37 $ ] ( @xmath42 $ ] , respectively ) , we denote by @xmath43 ( @xmath44 , respectively ) if they are equal up to multiplications of trivial units .
a one variable laurent polynomial @xmath45 $ ] is _ of type @xmath46 _ if there are integers @xmath47 and @xmath48 with @xmath49 odd , and @xmath50 $ ] @xmath51 such that @xmath52 , @xmath53 , and for @xmath54 , @xmath55 where @xmath56 , and @xmath57 [ lm : branch ] let @xmath58 be an oriented @xmath0-component link in @xmath5 such that the linking number @xmath59 of @xmath16 is not zero , and @xmath60 is the trivial knot .
let @xmath61 be the lifted knot of @xmath62 in the @xmath63-fold branched covering over @xmath60 where @xmath64 is coprime to @xmath59 . 1 .
if @xmath16 is component - preservingly amphicheiral , then @xmath61 is an amphicheiral knot .
if @xmath16 is component - preservingly @xmath65-amphicheiral , then @xmath61 is a @xmath66-amphicheiral knot , and there is an element @xmath67 $ ] such that @xmath68 , @xmath69 , and @xmath70 where @xmath71 is a primitive @xmath63-th root of unity .
if @xmath16 is component - preservingly @xmath72-amphicheiral , then @xmath61 is a @xmath30-amphicheiral knot , and there are @xmath73 $ ] of type @xmath46 and an odd number @xmath74 @xmath75 such that @xmath76 where @xmath71 is a primitive @xmath63-th root of unity . in particular ,
if @xmath61 is hyperbolic , then we can take @xmath77 and @xmath78 .
[ co : branch ] under the same setting as lemma [ lm : branch ] , if @xmath16 is @xmath65-amphicheiral , then in the prime factorization of @xmath79 , the power of a prime which is congruent to @xmath80 modulo @xmath81 is even .
in particular , if @xmath59 is odd , then @xmath82 satisfies the condition .
let @xmath83 ( @xmath84 , respectively ) be the set of prime amphicheiral links ( component - preservingly amphicheiral links , respectively ) with at least @xmath0 components and up to @xmath85 crossings , and @xmath86 the subset of @xmath83 consisting of @xmath21-amphicheiral links ( @xmath87 the subset of @xmath84 consisting of component - preservingly @xmath21-amphicheiral links , respectively ) where @xmath88 or @xmath13 .
it is clear that @xmath89 , @xmath90 , @xmath91 and @xmath92 .
[ th : table ] under the setting above , we have the following : 1 .
@xmath93 @xmath94 2 .
3 . @xmath96 , @xmath97 .
4 . @xmath98 , + @xmath99 .
we remark that theorem [ th : table ] ( 3 ) corresponds to the theorem of shida wang @xcite
. the following would also be useful if we determine prime amphicheiral links with the crossing number greater than @xmath1 in the future ( see also conjecture [ cj : tait ] in section [ sec : remark ] ) .
[ lm : tait ] the minimal crossing number of an alternating amphicheiral link is even . in section [ sec :
proof ] , we prove lemma [ lm : branch ] , corollary [ co : branch ] and lemma [ lm : tait ] . in section [ sec :
table ] , we prove theorem [ th : table ] . in section [ sec : remark ] , we give some remarks related to our previous results @xcite .
to show lemma [ lm : branch ] , we need two results .
k. murasugi @xcite gave a formula of the alexander polynomial of a periodic knot in @xmath5 , m. sakuma @xcite extended it to the case of periodic links , and v. g. turaev @xcite extended them for more general settings .
in the present paper , we use only the case of periodic knots in @xmath5 .
[ lm : period ] ( murasugi @xcite ; sakuma @xcite ; turaev @xcite ) let @xmath100 be an @xmath101-component link in @xmath5 such that @xmath102 and @xmath103 is the trivial knot .
let @xmath104 be the lifted link in the @xmath63-fold branched covering over @xmath103 where @xmath64 and @xmath105 is the lifted link of @xmath39 @xmath106 .
then @xmath107 is a @xmath63-periodic link in @xmath5 , and we have @xmath108 where @xmath71 is a primitive @xmath63-th root of unity .
r. hartley @xcite , r. hartley and a. kawauchi @xcite , and a. kawauchi @xcite gave necessary conditions on the alexander polynomials of amphicheiral knots .
[ lm : hk ] ( hartley @xcite ; hartley and kawauchi @xcite ; kawauchi @xcite ) 1 .
let @xmath7 be a @xmath66-amphicheiral knot .
then there is an element @xmath67 $ ] such that @xmath68 , @xmath69 , and @xmath109 2 .
let @xmath7 be a @xmath30-amphicheiral knot .
then there are @xmath73 $ ] of type @xmath46 ( cf . section [ sec : intro ] ) and an odd number @xmath74 @xmath75 such that @xmath110 in particular , if @xmath7 is hyperbolic , then we can take @xmath77 and @xmath78 .
* proof of lemma [ lm : branch ] * let @xmath58 be an oriented @xmath0-component component - preservingly amphicheiral link in @xmath5 with the linking number @xmath111 .
since the exterior of @xmath16 is orienation - preserving homeomorphic to that of the mirror image @xmath112 of @xmath16 with preserving boundary components ( which may not preserve orientations ) , @xmath61 is amphicheiral for every @xmath63 which shows ( 1 ) .
suppose that @xmath16 is @xmath65-amphicheiral .
then it is easy to see that @xmath61 is a @xmath66-amphicheiral knot , and the condition of the alexander polynomial is obtained from lemma [ lm : period ] and lemma [ lm : hk ] ( 1 ) .
suppose that @xmath16 is @xmath72-amphicheiral .
then it is easy to see that @xmath61 is a @xmath30-amphicheiral knot , and the condition of the alexander polynomial is obtained from lemma [ lm : period ] and lemma [ lm : hk ] ( 2 ) .
* proof of corollary [ co : branch ] * by substituting @xmath113 in the equation in lemma [ lm : branch ] ( 2 ) ( or lemma [ lm : hk ] ( 1 ) ) , and a result from elementary number theory on primes in @xmath114 $ ] , we have the result . to show lemma [ lm : tait ] , we need two results .
let @xmath115 be an oriented link diagram for an @xmath4-component link @xmath35 , @xmath116 the _ writhe _ of @xmath117 which is the sum of the signs of the crossings , and @xmath118 the the _ crossing number _ of @xmath117 .
[ lm : writhe ] in the situation above , let @xmath119 be an oriented diagram obtained from @xmath117 by reversing the orientation of the @xmath120-th component @xmath121
. then we have @xmath122 where @xmath123 is the linking number of @xmath39 and @xmath124 , and @xmath125 does not depend on the orientation of @xmath117 .
w. menasco and m. thistlethwaite @xcite gave the affirmative answer for tait s flyping conjecture .
for an alternating link @xmath16 , a _ reduced diagram _ of @xmath16 is a diagram of @xmath16 which is an alternating diagram without nugatory crossings .
flyping _ is an operation on a link diagram .
[ lm : flype ] ( menasco and thistlethwaite @xcite ) let @xmath16 be an oriented prime alternating link .
let @xmath117 and @xmath119 be two reduced diagrams of @xmath16 .
then @xmath117 and @xmath119 are related by a finite sequence of flypings . as consequences , we have @xmath126 , and @xmath127 which is the minimal crossing number of @xmath16 .
* proof of lemma [ lm : tait ] * let @xmath16 be an oriented prime alternating link , and @xmath117 a reduced diagram of @xmath16 . by lemma [ lm : flype ] , @xmath116 is an invariant of @xmath16 . by lemma [ lm : writhe ]
, @xmath125 is an invariant of @xmath16 as an unoriented link .
let @xmath128 be the mirror image diagram of @xmath117 .
suppose that @xmath16 is amphicheiral .
then we have @xmath129 , and hence @xmath130 .
it implies that the crossing number of @xmath16 is even . since a reduced diagram for a non - prime alternating link is realized by connected sums of reduced diagrams of the prime factors ( see @xcite ) , we have the result .
in this section , we determine prime amphicheiral links with at least @xmath0 components and up to @xmath1 crossings . for a link with the crossing number up to @xmath29 , we use the notation of d. rolfsen s book @xcite , and for a link with the crossing number @xmath32 or @xmath1 , we use a slightly modified notation from m. thistlethwaite s table on d. bar - natan and s. morrison s website @xcite below . in rolfsen s table @xcite , an @xmath4-component link such that @xmath131 and the crossing number @xmath132 is denoted by @xmath133 where @xmath134 is the ordering of the link in the table . in thistlethwaite
s table @xcite , an @xmath4-component link such that @xmath131 and the crossing number @xmath132 is denoted by @xmath135 or @xmath136 where ` @xmath137 ' implies that the link is alternating , ` @xmath85 ' implies that the link is non - alternating , and @xmath134 is the ordering of the link in the table .
we modify the notations @xmath135 and @xmath136 into @xmath138 and @xmath139 , respectively .
we raise some conditions on the jones polynomial and the alexander polynomial of an amphicheiral link without proofs . for an oriented @xmath4-component link @xmath35 ,
let @xmath140 be the jones polynomial of @xmath16 with one variable @xmath141 , and @xmath142 the homfly polynomial of @xmath16 with two variables @xmath143 and @xmath144 .
then @xmath140 is an element of @xmath145 $ ] , and @xmath142 is an element of @xmath146 $ ] .
let @xmath147 be an oriented link obtained from @xmath16 by changing the oriented @xmath120-th component @xmath39 into @xmath148 ( @xmath11 ) where @xmath12 or @xmath13 , and @xmath149 .
let @xmath150 be the mirror image of @xmath16 with the induced orientation .
[ lm : jones ] under the settings above , we have the following : 1
. @xmath151 $ ] , @xmath152 , and @xmath153 .
2 . @xmath154 where @xmath155 if @xmath156 , and @xmath157 if @xmath158 .
if @xmath16 is @xmath17-amphicheiral , then we have @xmath159 where @xmath137 is the same as in ( 2 ) .
4 . if @xmath16 is amphicheiral , then @xmath160 is equal to @xmath140 up to multiplication of @xmath161 @xmath162 ( i.e. the coefficients of @xmath140 are symmetric ) .
@xmath163 $ ] , @xmath164 , and @xmath165 . 6
. if @xmath16 is @xmath17-amphicheiral , then we have @xmath166 7 .
@xmath167 .
lemma [ lm : jones ] ( 1 ) , ( 2 ) , ( 5 ) and ( 7 ) are basic properties .
it is easy to see that ( 3 ) and ( 4 ) are deduced by ( 1 ) and ( 2 ) , and ( 6 ) is deduced by ( 5 ) .
let @xmath168 , @xmath169 and @xmath170 be three oriented links such that they are identical except the local parts as in figure 1 . ,
@xmath169 and @xmath170,title="fig : " ] [ fig : skein ] we computed the homfly polynomial of a link by the following skein relation .
@xmath171 where @xmath172 implies the trivial knot .
[ lm : amp ] ( ( * ? ? ?
* lemma 2.5 ) ) let @xmath3 be an @xmath4-component @xmath17-amphicheiral link where @xmath12 or @xmath13 @xmath106 , and @xmath14 is a permutation of @xmath15
. then we have @xmath173 [ lm : sub ] ( ( * ? ? ?
* lemma 3.1 ) , ( * ? ? ?
* lemma 4.2 ) ) let @xmath3 be an oriented @xmath4-component link
if @xmath16 is an @xmath17-amphicheiral link , then a sublink @xmath174 @xmath175 is an @xmath176-amphicheiral link where @xmath14 is a permutation of @xmath15 , and @xmath177 is a permutation of @xmath178 induced by @xmath14 .
if @xmath179 is odd , and @xmath180 where @xmath181 is the linking number of @xmath182 and @xmath183 , then @xmath16 is not component - preservingly amphicheiral .
3 . if @xmath184 where @xmath181 is the linking number of @xmath182 and @xmath183 , then @xmath16 is not amphicheiral .
[ lm : even ] let @xmath58 be a @xmath0-component link with non - zero even linking number @xmath185 .
then we have the following : 1 .
( hartley @xcite , ( * ? ? ?
* lemma 3.2 ) ) @xmath16 is not component - preservingly amphicheiral .
* lemma 4.3 ) ) if @xmath186 , then @xmath16 is not @xmath187-amphicheiral where @xmath188 is the non - trivial permutation of @xmath189 .
[ lm:2-comp ] ( ( * ? ? ?
* corollary 1.4 ) ) if @xmath58 is an algebraically split component - preservingly amphicheiral link , then @xmath190 is divisible by @xmath191 .
[ lm : sign ] ( ( * ? ? ?
* corollary 1.2 ) ) let @xmath35 be an @xmath4-component amphicheiral link such that @xmath192 is even where @xmath193 is the total linking number
. then we have @xmath194 in particular , if @xmath16 is an @xmath21-amphicheiral link where @xmath88 or @xmath13 , and @xmath4 is even , then we have @xmath195 prime links with up to @xmath29 crossings have been determined in @xcite ( cf .
figure 2 ) . from now on , we restrict the case that a link is prime with the crossing number @xmath32 or @xmath1
. crossings , title="fig : " ] [ fig:9cross ] or @xmath1 crossings 1,title="fig : " ] [ fig:11cross1 ] or @xmath1 crossings 2,title="fig : " ] [ fig:11cross2 ] firstly , we determine which links are amphicheiral , and whether they are in @xmath196 , @xmath197 , @xmath198 , or @xmath199 . by figure 3 and figure 4 , @xmath200 , @xmath201 , @xmath202 , @xmath203 , @xmath204 , @xmath205 , @xmath206 , @xmath207 , @xmath208 , @xmath209 , @xmath210 , @xmath211 , @xmath212 , @xmath213 , @xmath214 , @xmath215 , @xmath216 , @xmath217 and @xmath218
are amphicheiral .
the reader can check by the figures whether they are in @xmath196 or @xmath198 .
suppose that @xmath208 is @xmath219-amphicheiral
. then milnor s @xmath220-invariant @xmath221 should be zero .
however since @xmath222 we have @xmath223 ( cf .
@xcite ) , and hence @xmath224 .
since @xmath217 has a @xmath80-component sublink which is equivalent to @xmath225 , it can not be component - preservingly @xmath66-amphicheiral by lemma [ lm : sub ] ( 1 ) .
however both @xmath208 and @xmath217 are in @xmath226 by suitable orientations .
the @xmath0-component sublinks of @xmath209 are the @xmath0-component trivial link , the positive whitehead link and the negative whitehead link . since the whitehead link is not amphicheiral , @xmath227 , and it is in @xmath228 and @xmath226 by suitable orientations .
let @xmath229 be @xmath210 or @xmath211 .
then the @xmath0-component sublinks of @xmath16 are the @xmath0-component trivial link @xmath230 , the positive hopf link @xmath231 and the negative hopf link @xmath232 by a suitable orientation .
suppose that @xmath16 is component - preservingly amphicheiral .
then @xmath233 should be satisfied by lemma [ lm : amp ] . since @xmath234 @xmath210 and @xmath211 are not component - preservingly amphicheiral .
suppose that @xmath216 is component - preservingly amphicheiral .
then @xmath235 should be satisfied by lemma [ lm : amp ] . since @xmath236 @xmath216 is not component - preservingly amphicheiral .
there is one essential torus @xmath237 in the exterior of @xmath218 .
the torus @xmath237 is trivial as a torus in @xmath5 , and we denote the core of the separated solid torus by @xmath238 @xmath239 where @xmath238 is in the same connected component of @xmath39 .
suppose that @xmath218 is component - preservingly amphicheiral .
then two links @xmath240 and @xmath241 are amphicheiral .
however since they are the positive hopf link and the negative hopf link , respectively , and they are not amphicheiral , @xmath218 is not component - preservingly amphicheiral . by applying lemma [ lm : jones ] ( 4 ) for the rest links
, we can see non - amphicheirality of them except @xmath33 , @xmath242 , @xmath243 , @xmath244 , @xmath245 , @xmath246 , @xmath34 , @xmath247 , @xmath248 , @xmath249 , @xmath250 , @xmath251 and @xmath252 . by lemma [ lm : amp ] , lemma [ lm : sub ] , lemma [ lm : even ] , lemma [ lm:2-comp ] and lemma [ lm : sign ] ( cf .
* example 4.5 and example 4.6 ) ) , we can see non - amphicheirality of the links above except @xmath33 and @xmath34 . and @xmath34,title="fig : " ] [ fig:1051 ]
let @xmath58 be an oriented @xmath33 or @xmath34 as in figure 5 , and @xmath253 .
the homfly polynomials of @xmath33 and @xmath34 are @xmath254 by lemma [ lm : jones ] ( 6 ) , we can not show they are not amphicheiral by the homfly polynomials .
the components of @xmath33 consist of two trivial knots , and the linking number of it is @xmath255 .
since the alexander polynomial of it is @xmath256 @xmath33 is component - preservingly amphicheiral by lemma [ lm : amp ] if it is amphicheiral .
since we have @xmath257 @xmath33 is not amphicheiral by lemma [ lm : branch ] ( 2 ) and ( 3 ) . the components of @xmath34 consist of one trivial knot and one figure eight knot , and the linking number of it is @xmath255 .
it is component - preservingly amphicheiral by lemma [ lm : sub ] ( 1 ) if it is amphicheiral .
since the jones polynomial and the homfly polynomial of @xmath61 which is the lifted knot of @xmath62 in the @xmath0-fold branched covering over @xmath60 ( cf .
figure 6 ) are @xmath258 @xmath34 is not amphicheiral by lemma [ lm : branch ] ( 1 ) , lemma [ lm : jones ] ( 3 ) and ( 6 ) . of @xmath62 in the @xmath0-fold branched covering over @xmath60,title="fig : " ] [ fig:1127 ] we remark that we can not complete the proof only by using the alexander polynomial . since the alexander polynomial of @xmath34 is @xmath259 we have @xmath260 and hence @xmath34 is not @xmath65-amphicheiral by corollary [ co : branch ] .
let @xmath71 be a primitive @xmath63-th root of unity for @xmath64 .
then there exists @xmath261 $ ] such that @xmath262 where @xmath263\ \mbox{(by ( \ref{eq:11n127}))}.\ ] ] hence we can not prove that @xmath34 is not @xmath72-amphicheiral by lemma [ lm : branch ] ( 2 ) .
( 1 ) in @xcite , the author raised a conjecture : we gave a partial affirmative answer in ( * ? ? ?
* theorem 1.3 ) for the case that @xmath16 is an algebraically split component - preservingly @xmath21-amphicheiral link with even components . in the prime links with up to @xmath1 crossings , only @xmath213 and @xmath217
are algebraically split component - preservingly amphicheiral links with even components .
they are also component - preservingly @xmath30-amphicheiral links ( i.e. @xmath265 ) .
we can confirm that the alexander polynomials of them are @xmath266 .
the condition component - preservingly " is needed .
@xmath209 , @xmath215 and @xmath218 are algebraically split amphicheiral links in @xmath197 and @xmath199 whose alexander polynomials are @xmath267 @xmath268 satisfies the condition @xmath269 in lemma [ lm : sign ] .
we can find examples of @xmath49-component algebraically split links in @xmath270 with @xmath271 even in the milnor links ( see figure 7 ) .
the alexander polynomials of them are @xmath266 ( cf .
* example 6.1 ( 1 ) ) )
. if there are counterexamples , then the knots are not alternating by lemma [ lm : tait ] .
a. stoimenow @xcite found prime amphicheiral knots with the odd minimal crossing number @xmath132 for every @xmath273 .
he points out that if @xmath274 , then conjecture [ cj : tait ] is affirmative .
we have already found counterexamples @xmath275 and @xmath218 for the case of links .
however they are not component - preservingly amphicheiral .
recently y. kobatake found an example of a @xmath0-component prime component - preservingly amphicheiral link with the minimal crossing number @xmath276 and with the linking number @xmath80 , whose components consist of the unknot and an amphicheiral knot @xmath277 in the table of @xcite .
d. bar - natan and s. morrison , _ knot atlas _ , http://katlas.math.toronto.edu/wiki/main@xmath278page f. bonahon and l. c. siebenmann , _ new geometric splittings of classical knots and the classification and symmetries of arborescent knots _ , http://www-bcf.usc.edu/ fbonahon / research / preprints
/ bonsieb.pdf t. d. cochran , _ concordance invariance of coefficients of conway s link polynomial _ , invent . math . * 82 * ( 1985 ) , 527541 .
r. hartley , _ invertible amphicheiral knots _ , math .
ann . , * 252 * ( 1980 ) , 103109 .
j. hoste , m. thistlethwaite and j. weeks , _ the first @xmath279 knots _ , math .
intelligencer , * 21 * ( 1998 ) , 3348 .
r. hartley and a. kawauchi , _ polynomials of amphicheiral knots _ , math .
* 243 * ( 1979 ) , 6370 .
j. hillman , _ symmetries of knots and links , and invariants of abelian coverings ( part i ) _ , kobe j. math . , * 3 * ( 1986 ) , 727 .
b. jiang , x. lin , shicheng wang and y. wu , _
achirality of knots and links _ , top .
* 119 * ( 2002 ) , 185208 .
t. kadokami , _ the link - symmetric groups of @xmath0-bridge links _ , to appear in jounal of knot theory and its ramifications .
t. kadokami , _ amphicheiral links with special properties , i _ , to appear in jounal of knot theory and its ramifications , arxiv math.gt/1107.0377 t. kadokami and a. kawauchi , _ amphicheirality of links and alexander invariants _ , to appear in science china mathematics .
a. kawauchi , _ on quadratic forms of @xmath80-manifolds _ , invent .
, * 43 * ( 1977 ) , 177198 .
a. kawauchi , _ the invertibility problem on amphicheiral excellent knots _ , proc .
japan acad .
, ser . a , math .
, * 55 * ( 1979 ) , 399402 .
a. kawauchi , _ the quadratic form of a link _ , in : proc .
of low dimensional topology , contemp .
, * 233 * ( 1999 ) , 97116 .
s. kojima and m. yamasaki , _ some new invariants of links _ , invent .
* 54 * ( 1979 ) , 213228 . w. menasco and m. thistlethwaite , _ the classification of alternating links _ , ann . of math . , second series , * 138 * ( 1993 ) , 113171 . j. w. milnor , _ link groups _ , ann . of math . ( 2 ) * 59 * ( 1954 ) , 177195 . k. murasugi , _ on periodic knots _ , comment . math .
helvetici , * 46 * ( 1971 ) , 162174 .
k. murasugi , _ jones polynomials and classical conjectures in knot theory _ , topology , * 26 * ( 1987 ) , 187194 .
d. rolfsen , _ knots and links _ , publish or perish , inc .
t. sakai , _ polynomials of invertible knots _ , math . ann .
, * 266 * ( 1983 ) , 229232 .
m. sakuma , _ on the polynomials of periodic links _ , math .
* 257 * ( 1981 ) , 487494 .
a. stoimenow , _
tait s conjectures and odd crossing number of amphicheiral knots _ , bull .
soc . , * 45 * ( 2008 ) , 285291 .
a. stoimenow , _ non - triviality of the jones polynomial and the crossing number of amphicheiral knots _ , arxiv math.gt/0606255 v. g. turaev , _
reidemeister torsion in knot theory _ , russian math .
surveys , * 41 - 1 * ( 1986 ) , 119182 .
shida wang , _ strict achirality of prime links up to @xmath1-crossings _ , acta math .
sinica , * 24 * ( 2008 ) , 9971004 . w. whitten , _ symmetries of links _
* 135 * ( 1969 ) , 213222 . | we determine prime amphicheiral links with at least @xmath0 components and up to @xmath1 crossings .
there are @xmath2 such links .
we check also special amphicheiralities .
most of prime links with up to @xmath1 crossings are detected not to be amphicheiral by a condition on the jones polynomial .
for the rest links , we applied conditions from the alexander polynomial .
we added new necessary conditions for a special case .
[ section ] [ df]lemma [ df]theorem [ df]remark [ df]proposition [ df]example [ df]corollary [ df]claim [ df]question [ df]problem [ df]conjecture addtoresetequationsection |
in this talk i would like to address the question about how to incorporate the heavy particle mass effects into the photon structure functions .
this talk is based on the works done in collaboration with yoshio kitadono , ryo sahara , ken sasaki , takahiro ueda and yutaka yoshida .
now let me make some remarks on why the photon structure is so interesting .
first of all , it provides a good probe to study the qcd dynamics in perturbation theory .
the unpolarized virtual photon structure function @xmath0 was studied up to the next - to - next - to - leading order ( nnlo ) and the polarized virtual photon structure function @xmath1 was investigated to next - to - leading order ( nlo ) .
the qcd sum rule for the first moment of @xmath1 has attracted much attention in the literature since it is related to the axial anomaly . in this
talk we investigate the heavy particle mass effects on the photon structure functions including the polarized structure function .
here we investigate two - photon processes ( figure 1 ) with the kinematical region where the mass squared of the probe photon ( @xmath2 ) is much larger than that of the target photon ( @xmath3 ) which is in turn much bigger than the @xmath4 , the qcd scale parameter squared .
the advantage for studying the virtual photon target is that we can calculate whole structure functions up to next - leading - order ( nlo ) , in contrast to the real photon target where there remain uncalculable non - perturbative pieces .
e@xmath5 collision experiments ]
the structure function @xmath6 for the real photon target ( @xmath7 ) was first studied in the parton model @xcite and then studied to the leading order ( lo ) qcd based on the operator product expansion ( ope ) in ref.@xcite .
the next - to - leading order qcd computation of @xmath6 was performed in the ope and renormalization group ( rg ) method by bardeen and buras @xcite and in the qcd - improved parton model by glck and reya @xcite and fontannaz and pilon @xcite and by other people quoted in ref.@xcite . for a virtual photon target ( @xmath8 )
the unpolarized photon structure function @xmath9 to lo in qcd was studied in ref.@xcite and to nlo in ref.@xcite .
the master formula for the @xmath10-th moment to nlo is given by @xmath11~ , \label{master}\end{aligned}\ ] ] where @xmath12 , @xmath13 , @xmath14 and @xmath15 are computed from the one- and two - loop anomalous dimensions together with one - loop coefficient functions . all of them are shown to be renormalization - scheme independent .
@xmath16 is the qcd running coupling constant , @xmath17 is the one - loop beta function , and @xmath18 denote the eigenvalues of one - loop anomalous dimensions @xmath19 .
we compute the deviation arising from heavy quark mass effects on the photon matrix elements of twist-2 quark and gluon operators and the corresponding coefficient functions to nlo qcd @xcite .
the moment of @xmath0 can be decomposed as @xmath20 where the additional moment due to mass effects reads @xmath21 \right . \nnb\\ & & \left.\hspace{-1 cm } + \sum_{i=\pm , ns } \delta \mathcal{b}_{i}^{n } \left [ 1 - \left ( \frac{\alpha_s(q^2 ) } { \alpha_s(p^2 ) } \right)^{d_{i}^n+1 } \right ] + \delta \mathcal{c}^{n } \hspace{0.2 cm } \right ] + \mathcal{o}(\alpha_s)~,\nnb \end{aligned}\ ] ] where @xmath22 ( @xmath23 ) . in the massive quark
limit : @xmath24 the deviations of the the coefficients turn out to be @xmath25 where @xmath26 is the heavy - quark charge and the average charge squared is defined as @xmath27 with @xmath28 being the number of active flavors and the deviation of operator matrix element @xmath29 is @xmath30.\nnb\\\end{aligned}\ ] ] the figure 2 shows the theoretical evaluation with charm quark mass effects compared to the pluto s experimental data @xcite for the effective structure function @xmath31 , where @xmath32 is the longitudinal structure function . the theoretical prediction ( red curve ) with the charm quark
mass effects shows the trend of reducing the massless qcd calculation ( purple curve ) and becomes consistent with the experimental data .
also shown is the theoretical curve ( blue curve ) with the resummation prescription given in @xcite . to nlo in qcd with charm quark mass effects compared to pluto data for @xmath33 , @xmath34gev@xmath35 , @xmath36gev@xmath35 , @xmath37gev @xcite ]
we now turn to the polarized photon structure function @xmath1 .
let us first consider massless quark contributions to the photon structure functions
. for a real photon target ( @xmath7 ) , bass , brodsky and schmidt have shown that the 1st moment of @xmath38 vanishes to all orders of @xmath16 in qcd @xcite : @xmath39 now the question is what about the @xmath40 moment of the virtual photon case . by replacing the @xmath41 with @xmath42 in ( [ master ] ) and taking @xmath43 limit ,
the first three terms vanish . denoting @xmath44 , the @xmath45-th quark charge and @xmath28 , the number of active flavors
, we have @xmath46 we can now go to @xmath47 corrections which turn out to be @xcite : @xmath48 + { \cal o}(\alpha\alpha_s^2).\label{sum - rule - nlo}\end{aligned}\ ] ] this result coincides with the one obtained by narison , shore and veneziano @xcite , apart from the overall sign for the definition of @xmath1 .
we then extended this nlo result to the next - to - next - to - leading order ( nnlo ) calculation , which is @xmath49 , where we need the 3-loop anomalous dimensions as well as 2-loop coefficient functions @xcite .
the same formalism can be applied to the heavy quark mass effects as in the case of unpolarized structure functions .
of course we have to replace the one- and two - loop anomalous dimensions as well as the one - loop coefficient functions by the polarized counterparts .
in addition we put the operator matrix elements with the photon states .
namely @xmath50.\end{aligned}\ ] ] let us first consider the heavy quark mass effects on the first moment of the virtual photon structure function @xmath51 . to the lowest order the first moment
is given by _
0 ^ 1 dx g_1^(x , q^2,p^2)= ^n=1+o ( _ s ) . for the massless case
we have ^n=1=12_0e^4_n_f ( b_g^n+_ng^)|_n=1 = -24_0n_fe^4_n_f , where @xmath52 . now in the presence of the heavy quark , the @xmath28-th flavor , with the charge @xmath53 : ^n=1=24_0e_h^4 .
hence we find that the heavy quark mass effects lead to & & _ 0 ^ 1 dx g_1^(x , q^2,p^2)= ( c^n=1+^n=1 ) + & & = -_i=1^n_f-1e_i^4+o ( _ s ) .
thus the heavy quark decouples from the sum rule .
this argument can be extended to the order @xmath47 and the sum rule becomes eq.([sum - rule - nlo ] ) where @xmath28 is replaced by @xmath54 .
we now investigate the heavy quark mass effects on the virtual photon structure function @xmath1 as a function of the bjorken variable @xmath55 @xcite . in figure 3
we have plotted the theoretical prediction for the structure function @xmath51 in the case of @xmath56gev@xmath35 and @xmath36gev@xmath35 , where @xmath33 and the heavy quark is the charm quark with mass squared @xmath57gev@xmath35 . in units of @xmath58 with and without the charm quark
mass effects for @xmath33 , @xmath56gev@xmath35 , @xmath36gev@xmath35 ] in units of @xmath58 with and without the bottom quark mass effects for @xmath60 , @xmath61gev@xmath35 , @xmath62gev@xmath35 ] note that the polarized structure function for the massive case ( red curve ) is suppressed in the large @xmath55 region compared to the massless case ( blue curve ) and becomes less negative for small @xmath55 region when the charm quark mass effect is taken into account . varying the values of @xmath2 and @xmath3 for the kinematical region where the charm quark can be regarded as the heavy quark we get the similar behaviors .
we have also shown in figure 4 , the structure function for the the case of @xmath61gev@xmath35 and @xmath62gev@xmath35 , where @xmath60 and the heavy quark is the bottom quark with mass squared @xmath63gev@xmath35 .
we note that the deviation from massless case is almost negligible , since the charge factor is much smaller compared to the charm case i.e. @xmath64 .
we have also applied our formalism for the photon structure functions in supersymmetric qcd ( sqcd ) where there exist squarks and gluinos in addition to the quarks and gluons .
the squark contribution in the parton model was studied in @xcite . for the sqcd
radiative corrections , we based our argument on the parton distribution functions ( pdf s ) which satisfy the dglap - type evolution equation .
the heavy mass effects can be incorporated by the boundary conditions where we require the pdf for each particle to vanish at @xmath65 ( @xmath45=gluino , squark , heavy quark ) . by solving the coupled boundary conditions we can determine the initial conditions and then we get the solution for the moment of the each pdf . by performing the inverse mellin transform
we finally get structure functions . in figure 5
we have plotted @xmath9 as a function of @xmath55 .
there we have also taken into account the threshold effects @xmath66 with @xmath67 . in supersymmetric qcd where heavy particle mass effects are taken into account @xcite ]
we note that at small @xmath55 there is no significant difference between massless and massive qcd , while there exists a large difference between massless and massive sqcd . at large @xmath55 ,
the significant mass - effects exist both for non - susy and susy qcd .
the sqcd case is seen to be much suppressed at large @xmath55 compared to the qcd .
in this talk we discussed the heavy particle mass effects for the virtual photon structure functions which are calculable by perturbative qcd ( sqcd ) . here
we are particularly interested in evaluation of unpolarized and polarized structure functions , @xmath0 and @xmath1 , with the heavy particle mass effects .
we first discussed our basic framework of studying heavy quark mass effects for @xmath0 .
we then present the analysis of the polarized photon structure function @xmath1 in two aspects ; qcd sum rule of the first moment which is related to the axial anomaly as well as a the @xmath55 dependence of the structure function .
we have shown that the heavy particle decouples from the qcd sum rule .
the analysis shows that the large @xmath55 region of @xmath51 is suppressed in the presence of the heavy quark effects .
we also investigated the photon structure functions in supersymmetric qcd .
experimental confrontation in the future is anticipated .
i would like to thank professor stephen narison for the wonderful organization and hospitality . | in the framework of the perturbative qcd we investigate heavy particle mass effects on the unpolarized and polarized photon structure functions , @xmath0 and @xmath1 , respectively .
we present our basic formalism to treat heavy particle mass effects to nlo in perturbative qcd .
we also study heavy quark effects on the qcd sum rule for the first moment of @xmath1 , which is related to axial anomaly .
the photon structure function in supersymmetric qcd is also briefly discussed .
qcd , photon structure function , heavy particle mass effects , sum rule , @xmath1 , axial anomaly |
let @xmath6 be a group .
we say that @xmath6 has _ spread r _ if for any set of distinct non - trivial elements @xmath7 there exists an element @xmath8 with the property that @xmath9 for every @xmath10 .
we say that @xmath11 is a _ mate _ to @xmath12 .
we say @xmath6 has _ exact spread _
@xmath13 if @xmath6 has spread @xmath14 but not @xmath15 .
this concept was first introduced by brenner and wiegold in @xcite , extending earlier work of binder in @xcite .
the concept of spread may be thought of as a generalization of the idea of 3/2-generation . since this note
is not intended for publication , we omit background and motivation , but the interested reader may wish to consult @xcite . the reader should at least note that there has been considerable interest in determining , or at least bounding , the exact spreads of the finite simple groups and in particular the exact spreads of the sporadic simple groups .
the precise value of the exact spread of a finite simple group is known in very few cases . here
we prove the following .
first note that from ( * ? ? ?
* table 1 ) ( itself a heavily corrected version of ( * ? ? ?
* table 1 ) ) we have that @xmath17 and so it is sufficient to show that any set of 8064 elements from m@xmath1 has a mate .
let @xmath18 m@xmath3 be a set of 8064 distinct elements .
now , from the maximal subgroups of m@xmath3 listed in @xcite we see that an element of order 23 is contained in only one maximal subgroup - a copy of the frobenius group 23:11 . since the normalizer in m@xmath3 of a cyclic subgroup of order 11 is a frobenius group 11:5 , each element of order 11
is contained in five disctinct copies of 23:11 .
it follows that the only way @xmath12 can avoid having a mate of order 23 is if @xmath12 consists of a well chosen configutation of elements of order 11 .
finally , note that of the maximal subgroups of m@xmath3 containing elements of order 11 , which are each isomorphic to one of m@xmath19 , m@xmath20 or 23:11 , none contain elements of order 14 .
it follows that even if @xmath12 only contains elements of order 11 , @xmath12 must have a mate and so @xmath21 . | we show that if @xmath0 m@xmath1 then there exists an element @xmath2 m@xmath3 such that @xmath4 m@xmath3 for every @xmath5 and that no larger set of distinct elements of m@xmath3 has this property . |
the sloan digital sky survey ( sdss ; york et al .
2000 ) saw first light a decade ago , with the goals of obtaining ccd imaging in five broad bands over 10,000 deg@xmath0 of high - latitude sky , and spectroscopy of a million galaxies and one hundred thousand quasars over this same region . with this , its seventh public data release , these goals have been realized .
the survey facilities have also been used to carry out a comprehensive imaging and spectroscopic survey to explore the structure , composition , and kinematics of the milky way galaxy ( sloan extension for galactic understanding and exploration ; segue ; yanny et al .
2009 ) , and a repeat imaging survey that has discovered almost 500 spectroscopically confirmed type ia supernovae with superb light curves ( frieman et al.2008 ; holtzman et al .
2008 ) .
the sdss uses a dedicated wide - field 2.5 m telescope ( gunn et al .
2006 ) located at apache point observatory ( apo ) near sacramento peak in southern new mexico .
the telescope uses two instruments .
the first is a wide - field imager ( gunn et al .
1998 ) with 24 @xmath3 ccds on the focal plane with @xmath4 pixels , that covers the sky in drift scan mode in five filters in the order @xmath5 ( fukugita et al . 1996 ) .
the imaging is done with the telescope tracking great circles at the sidereal rate ; the effective exposure time per filter is 54.1 seconds , and 18.75 deg@xmath0 are imaged per hour in each of the five filters .
the images are mostly taken under good seeing conditions ( the median is about @xmath6 in @xmath7 ) on moonless photometric nights ( hogg et al .
2001 ) ; the exceptions are a series of repeat scans of the celestial equator in the fall for a supernova search ( frieman et al .
2008 ) , as is described in more detail in
[ sec : stripe82 ] .
the 95% completeness limits of the images are @xmath8 , respectively ( abazajian et al .
2004 ) , although these values depend as expected on seeing and sky brightness .
the images are processed through a series of pipelines that determine an astrometric calibration ( pier et al .
2003 ) and detect and measure the brightnesses , positions and shapes of objects ( lupton et al . 2001
; stoughton et al .
the astrometry is good to 45 milli - arcseconds ( mas ) rms per coordinate at the bright end , as described in more detail in
[ sec : astrometry ] .
the photometry is calibrated to an ab system ( oke & gunn 1983 ) , and the zeropoints of the system are known to 12% ( abazajian et al .
the photometric calibration is done in two ways , by tying to photometric standard stars ( smith et al .
2002 ) measured by a separate 0.5-m telescope on the site ( tucker et al . 2006 ; ivezi et al .
2004 ) , and by using the overlap between adjacent imaging runs to tie the photometry of all the imaging observations together , in a process called ubercalibration ( padmanabhan et al .
results of both processes are made available ; with this data release , the ubercalibration results , which are uncertain at the @xmath9 level in @xmath10 and 2% in @xmath11 , are now the default photometry made available in the data release described in this paper .
the photometric catalogs of detected objects are used to identify objects for spectroscopy with the second of the instruments on the telescope : a 640-fiber - fed pair of multi - object double spectrographs , giving coverage from 3800 to 9200 at a resolution of @xmath12 .
the objects chosen for spectroscopic follow - up are selected based on photometry corrected for galactic extinction following schlegel , finkbeiner , & davis ( 1998 ; hereafter sfd ) , and include : * a sample of galaxies complete to a petrosian ( 1976 ) magnitude limit of @xmath13 ( strauss et al .
2002 ) ; * two deeper samples of luminous red ellipticals selected in color - magnitude space to @xmath14 and @xmath15 , respectively , which produce an approximately volume - limited sample to @xmath16 , and a flux - limited sample extending to @xmath17 , respectively ( eisenstein et al .
2001 ) ; * flux limited samples of quasar candidates , selected by their non - stellar colors or first ( becker et al . 1995 ) radio emission to @xmath18 in regions of color space characteristic of @xmath19 quasars , and to @xmath20 for quasars with @xmath21 ( richards et al .
2002 ) ; * a variety of ancillary samples , including optical counterparts to rosat - detected x - ray sources ( anderson et al . 2007 ) ; * stars for spectrophotometric calibration and telluric absorption correction , as well as regions of blank sky for accurate sky subtraction ; * a variety of categories of stellar targets with a series of color and magnitude cuts for measurements of radial velocity , metallicity , surface temperature , and galactic structure as part of segue ( yanny et al .
2009 ) .
these targets are arranged on _
tiles _ of radius @xmath22 , with centers chosen to maximize the number of targeted objects ( blanton et al .
each tile contains 640 objects , and forms the template for an aluminum spectroscopic plate , in which holes are drilled to hold optical fibers that feed the spectrographs .
spectroscopic exposures are 15 minutes long , and three or more are taken for a given plate to reach pre - defined requirements of signal - to - noise ratio ( s / n ) , namely @xmath23 per 1.5 pixel for stellar objects of fiber magnitude @xmath24 and @xmath25 . for the segue
faint plates , the exposures are considerably deeper , and typically consist of eight 15-minute exposures , giving @xmath26 at the same depth ( yanny et al .
2009 ) .
spectra are extracted and calibrated in wavelength and flux .
the typical s / n of a galaxy near the main sample flux limit is 10 per pixel .
the broad - band spectrophotometric calibration is accurate to 4% rms for point sources ( adelman - mccarthy et al .
2008 ) , and the wavelength calibration is good to 2 @xmath27 .
the spectra are classified and redshifts determined using a pair of pipelines ( stoughton et al . 2002 ; subbarao et al .
2002 ) , which give consistent results 98% of the time ; the discrepant objects tend to be of very low s / n , or very unusual objects , such as extreme bals , superposed sources , and so on .
the vast majority of the spectra of galaxies and quasars yield reliable redshifts ; the failure rate is of order 1% for galaxies , and slightly larger for quasars .
the stellar targets are further processed by a separate pipeline ( lee et al . 2008a , b ; allende prieto et al .
2008a ) which determines surface temperatures , metallicities , and gravities .
the resulting catalogs are stored and distributed via a database accessible on the web ( the catalog archive server , cas ; thakar et al .
2008 ) , and the images and flat files are available in bulk through the data archive server ( das ) .
the sdss saw first light in may 1998 , and started routine operations in april 2000 .
it was originally funded for five years of operations , but had not completed its core goals of imaging and spectroscopy of a large contiguous area of the northern galactic cap by 2005 .
the survey was extended for an additional three years , with the additional goals of the segue and the supernova surveys mentioned above .
the extended program is known as sdss - ii , and the component of sdss - ii that represents the completion of sdss - i is known as the legacy survey .
sdss - ii observations were completed in july 2008 .
the sdss data have been made public in a series of yearly data releases ( stoughton et al .
2002 ; abazajian et al . 2003 , 2004 , 2005 , 2006 ;
adelman - mccarthy et al . 2007 , 2008 ; hereafter the edr , dr1 , dr2 , dr3 , dr4 , dr5 , and dr6 papers , respectively ) .
the most recent of these papers described the sixth data release ( dr6 ) , which included data taken through july 2006 .
the present paper describes the seventh data release ( dr7 ) , including data taken through the end of sdss - ii in 2008 july , and thus represents two additional years of data .
the data releases are cumulative ; dr7 includes all data included in the previous releases as well . in
[ sec : sky_coverage ] , we describe the footprint of this survey ; most importantly , we have completed our goals of : * contiguous imaging and spectroscopy over 7500 deg@xmath0 of the northern galactic cap ( the legacy survey ) ; * imaging and spectroscopy of stellar sources over an additional 3500 deg@xmath0 at lower galactic latitudes to study the structure of the milky way , and * repeat imaging of @xmath28 deg@xmath0 on the celestial equator in the fall months to discover type ia supernovae with @xmath29 .
in [ sec : runsdb ] , we describe the repeat scans on the celestial equator , including a co - addition of the images to reach about two magnitudes deeper than the main survey . in
[ sec : imaging ] , we present improvements in the processing of the imaging data , including improved stellar photometry at low galactic latitudes , an astrometric recalibration , and improvements in our photometric redshift algorithms for galaxies .
the dr6 paper described a problem with the photometry of bright galaxies ; we explore this further in [ sec : skysub ] . in
[ sec : spectroscopy ] , we discuss improvements in the spectroscopic processing of the data .
the dr6 paper described improvements in the wavelength and spectrophotometric calibration ; we have implemented further refinements which are important in the determination of accurate stellar parameters from the spectra .
we conclude in [ sec : conclusions ] with a discussion of the future of the sdss project .
table [ table : dr7_contents ] summarizes the contents of dr7 , giving the imaging and spectroscopic sky coverage and number of objects .
the imaging footprint has increased by roughly 22% since dr6 ( most of it outside the contiguous area of the north galactic cap ) , and the number of spectra has increased by 29% . lr imaging area in cas & 11663 deg@xmath0 + imaging catalog in cas & 357 million unique objects + legacy footprint area & 8423 deg@xmath0 + & ( 7646 deg@xmath0 in n. galactic cap ) + legacy imaging catalog & 230 million unique objects + & 585 million entries ( including duplicates ) + segue footprint area , available in das & 3500 deg@xmath0 ( more than double dr6 ) + segue footprint area , available in cas & 3240 deg@xmath0 + segue imaging catalog & 127 million unique objects + m31 , perseus , sagittarius scan area & @xmath30 deg@xmath0 + southern equatorial stripe with @xmath31 repeat scans & @xmath32 deg@xmath0 + commissioning ( `` orion '' ) data & 832 deg@xmath0 + spectroscopic footprint area & 9380 deg@xmath0 + legacy & 8032 deg@xmath0 + segue & 1348 deg@xmath0 + total number of plate observations ( 640 fibers each ) & 2564 + legacy survey plates & 1802 + segue and special plates & 676 + repeat observations of plates & 86 + total number of spectra & 1,630,960 + galaxies & 929,555 + quasars & 121,363 + stars & 464,261 + sky & 97,398 + unclassifiable & 28,383 + spectra after removing skies and duplicates &
1,440,961 + the imaging for the legacy survey was substantially complete with dr6 . in dr7
, we include imaging of a few small gaps that were missed in the contiguous region of the north galactic cap , and repeat observations of a few regions of the sky which had particularly poor seeing in previous data releases .
the total footprint has increased by less than 10 deg@xmath0 in total .
the legacy imaging footprint is visible as the large contiguous gray area on the left side of the upper panel of figure [ fig : skydist ] , together with the three gray stripes visible on the right side .
the principal augmentation of the imaging data in dr7 are the stripes which are part of the segue survey .
they are indicated in red in the figure , and increase the sdss imaging footprint by roughly 2000 deg@xmath0 over dr6 .
note that many of these cross the galactic plane ( indicated by the sinuous line crossing the figure ) .
unlike dr6 , the union of the legacy and segue data are now available in a single database in cas in dr7 .
these data have been recalibrated using ubercalibration ( padmanabhan et al.2008 ) using the overlap between adjacent scans ; the resulting photometry is now the default photometry found in the cas .
we also make available the original photometry calibrated by the auxiliary photometric telescope ( tucker et al .
the ubercalibration solution was regenerated using all the imaging data , but the changes are tiny from the ubercalibration results published in dr6 : 0.001 mag rms in @xmath10 and 0.003 mag in @xmath11 .
the ubercalibrated photometry zeropoints are defined to be the same as that measured from the photometric telescope .
the green and blue patches indicate supplementary imaging stripes , which contain scans over m31 or in its halo , through the center of the perseus cluster of galaxies , over the low - latitude globular cluster m71 , near the south galactic pole , along the orbit of the sagittarius tidal stream , and through the star - forming regions of orion ( finkbeiner et al .
in addition , there are a number of scans at angles perpendicular , or at an oblique angle , to the regular legacy or segue imaging stripes .
these scans are used in the ubercalibration procedure to tie the zeropoints of the stripes together and to determine the flat - fields .
the lower panel in figure [ fig : skydist ] shows the coverage of spectroscopy in dr7 ; the light gray area shows the increment in the legacy survey over dr6 .
most importantly , the gap cutting the north galactic cap in two pieces in previous data releases has been closed ; we now have complete spectroscopy of our principal galaxy and quasar targets over a contiguous area of roughly 7500 deg@xmath0 .
an additional dozen plates were observed to fill holes in the nominally contiguous regions in dr6 .
adding in the three stripes in the southern galactic cap , the legacy spectroscopy footprint is 8032 deg@xmath0 , a 26% increment over dr6 .
in addition , spectroscopy was carried out using a series of target selection algorithms designed to find stars of a wide variety of types as part of the segue project ( dr6 paper ; yanny et al .
these targets were drawn from both the segue and legacy imaging , and are shown in red in the lower panel of figure [ fig : skydist ] . as some of these are lost in the density of legacy spectra , we show the distribution of segue and other non - legacy spectra in galactic coordinates in figure [ fig : skydist_segue ] . finally , as described in yanny et al .
( 2009 ) , we carried out spectroscopy of stars in 12 open and globular clusters to calibrate the measurements of stellar parameters in segue ( lee et al .
2008a , b ) .
many of these clusters are sufficiently close that the giant branches are brighter than the photometric saturation limit of sdss , so the targets for these plates were selected from the literature .
indeed , the spectrographs would saturate as well with our standard 15-minute exposures , so these observations had individual exposure times as short as one or two minutes . without proper flux calibrators or exposure of bright sky lines to set the zeropoint of the wavelength scale ,
the spectrophotometry and wavelength calibration of the spectra on these plates are often quite inferior to that of the main survey , and these plates are available only in the das , not the cas .
, width=453 ] as described in more detail below , the @xmath33 stripe centered on the celestial equator was imaged multiple times throughout sdss and sdss - ii .
each 2.5@xmath2 wide stripe is observed by a pair of offset _ strips _ to cover the full width ( york et al .
2000 ) ; the coverage of the two strips of stripe 82 is shown in figure [ fig : stripe82 ] .
the data are shown both for the subset of data included in a deep coaddition ( lower set of curves ; [ sec : coadd ] ) , and all scans , including those taken under non - ideal conditions for the supernova survey ( [ sec : stripe82 ] ; frieman et al .
the sdss imaging survey was primarily designed to give a single pass across the sky , thus in the cas , each photometric measurement is flagged either primary or secondary .
primary objects designate a unique set of detections ( i.e. , without duplicates ) using the geometric boundaries of survey stripes .
the set of secondary objects includes repeat observations of the same object in overlapping strips an stripes .
primary objects are associated with a run and field which is the primary source of imaging data at that position . in dr7 ,
the union of the legacy and segue footprints serves as the primary footprint ; a quantity inlegacy in the fieldqa table in cas indicates those objects which lie within the original legacy northern galactic cap survey ellipse , as defined in york et al .
legacy imaging can also be distinguished by the stripe number for each run ; stripes 9 - 44 , 76 , 82 and 86 are in the legacy survey , all others are segue stripes or other miscellaneous pieces of sky ( figure [ fig : skydist ] ) . while resolving the sky into a seamless primary region of unique detections of objects
is ideal for many science queries , it is sometimes convenient to query data by run without regard to the way the survey resolves overlaps and imposes the boundaries of the edge of the survey .
these boundaries are restricted to matched pairs of north and south strips in the main dr7 cas .
therefore in many runs , several fields at the beginning or end which do not have a match in the corresponding other strip are not included in the main cas .
thus we have now made available a separate runs database within the cas , which includes _ all _ fields in all runs , and which allows one to query objects by which run they are imaged in .
the runs database contains 530 complete runs from sdss - i and sdss - ii , where primary is set strictly based on geometric limits within each scan , regardless of overlapping runs or stripes .
the runs database also contains several scans outside the regular dr7 legacy or segue footprints .
for example , stripe 205 is covered by runs 4334 , 4516 , 6751 and 6794 , and follows the sagittarius stream , which is in three pieces , the first running from @xmath34 to @xmath35 , the second centered at @xmath36 , and the last ( overlapping several other runs ) which ends at @xmath37 .
the sdss stripe along the celestial equator in the southern galactic cap ( `` stripe 82 '' ) was imaged multiple times in the fall months .
this was first carried out to allow the data to be stacked to reach fainter magnitudes , and through fall 2004 , these data were taken only under optimal seeing , sky brightness , and photometric conditions ( i.e. , the conditions required for imaging in the legacy survey ; york et al .
there were 84 such runs made public in previous data releases . in fall 2005 , 2006 , and 2007 ,
219 additional imaging runs were taken on stripe 82 as part of the sdss supernova survey ( frieman et al .
2008 ) , often under less optimal conditions : poor seeing , bright moonlight , and/or non - photometric conditions .
these data have been photometrically calibrated following the prescription of bramich et al .
( 2008 ) , whereby the photometry of bright stars is tied to that of photometric data on a field - by - field basis ( see ivezi et al .
2007 for a similar approach ) .
bramich et al .
solved for photometric offsets both parallel and perpendicular to the scan direction in data from a given ccd ; we found that the term perpendicular to the scan direction added little , and we did not include it here .
as bramich et al .
( 2008 ) show , the resulting photometric calibration is good to 0.02 mag at the bright end in up to 1 mag of atmospheric extinction . of course ,
under non - optimal conditions , these data will not necessarily reach as deep as normal survey images .
sdss judges photometricity of a given night by monitoring fluctuations in the night sky measured by a wide area infrared camera ( the `` cloud camera '' ) sensitive at 10@xmath38 m , where clouds are emissive ( hogg et al .
if the sky fluctuations are small and constant , then the night is photometric .
clouds cause the fluctuations to increase .
plots of cloud cover and seeing for most nights on which stripe 82 was observed are available as part of the dr7 web documentation listing all stripe 82 scans .
in addition , for those runs which the cloud camera indicated as non - photometric , we examined the fluctuations in the zeropoint for each ccd in the camera as a function of time using the photometric calibration procedure of bramich et al .
these zeropoint values are available in the cas ; rms variations of more than 0.1 mag are an indication of considerable variable cloud cover , and a value of more than 1 magnitude suggests that the approximate calibration procedure of bramich et al .
( 2008 ) breaks down , and the resulting photometry should be regarded with caution .
all 303 runs covering stripe 82 are made available as part of the stripe82 database , which is structured like the runs database .
we have carried out a coaddition of the repeat imaging scans on stripe 82 taken through fall 2005 under the best conditions ( see below ) .
the coaddition includes a total of 122 runs , covering any given piece of the @xmath28 deg@xmath0 area between 20 and 40 times ( figure [ fig : stripe82 ] ) , and the results are made available in the stripe82 database as well .
the coaddition runs are designated 100006 ( south strip ) and 200006 ( north strip ) respectively in the das , and 106 and 206 in the cas .
the coaddition is described in detail in annis et al .
( 2009 ) ; see also jiang et al .
( 2008 ) . from the list of runs on stripe 82 taken through the fall 2005 season , all fields with seeing in the @xmath7 band worse than @xmath39 fwhm ,
@xmath7-band sky brightness brighter than 19.5 magnitudes in one square arcsecond , or whose photometric correction la bramich et al .
( 2008 ; see above ) was greater than 0.2 mags were excised ; this rejected 32% of the available data .
the individual runs were remapped onto a uniform astrometric coordinate system . interpolated pixels in each individual run ( e.g. ,
for bad columns , bleed trails , and cosmic rays ) were masked in the coaddition process .
the sky was subtracted from each frame , and the images coadded with weights for each frame proportional to the transparency and inversely proportional to the square of sky noise and seeing on each frame .
strongly discrepant pixels were clipped in the coaddition .
the effective seeing fwhm is @xmath40 ( for the southern strip of the stripe ) and @xmath41 ( for the northern strip ) .
the resulting coadded images were run through the sdss photometric pipeline , yielding the catalogue made available in the stripe82 database . rather than deriving the point spread function ( psf ) from scratch , we synthesized the psf at each point in the sky by taking the suitably weighted sum of the psfs output by the sdss photometric pipeline from each of the individual runs .
color - color diagrams of stars and counts of stars and galaxies as a function of magnitude demonstrates that the photometry reaches roughly two magnitudes fainter than single sdss scans , similar to what is expected given the number of runs in the coadd .
we have found that star - galaxy separation is improved over that in the single scans , in that the cut can be made closer to the stellar locus . in the main survey ,
objects with @xmath42 are flagged as galaxies in a given band .
however , the stellar peak in the psf model magnitude difference distribution in the coadd is much narrower , allowing objects with @xmath43 in @xmath7 to be flagged as galaxies .
the coaddition does not properly propagate information on saturated pixels in individual runs , and therefore the photometry of objects brighter than roughly @xmath44 is suspect .
unfortunately , there is no processing flag that one can use to identify such data ; we recommend a simple magnitude cut .
the sdss photometry is quoted in terms of asinh magnitudes , as described by lupton , gunn , & szalay ( 1999 ) , whereby the logarithmic magnitude scale transitions to a linear scale in flux density @xmath45 at low signal - to - noise ratio : @xmath46.\ ] ] the magnitude at which this transition occurs is set by the quantity @xmath47 , which is roughly the fractional noise in the sky in a psf aperture in @xmath48 seeing ( edr paper ) . here
@xmath49 jy , the zeropoint of the ab flux scale .
the quantity @xmath47 for the coaddition is given in table [ table : asinh ] , along with the asinh magnitude associated with a zero flux object .
compare with the equivalent numbers for the main survey , given in table 21 of the edr paper .
table [ table : asinh ] also lists the flux corresponding to @xmath50 , above which the asinh magnitude and the traditional logarithmic magnitude differ by less than 1% in flux .
cccc @xmath11 & @xmath51 & 27.50 & 24.99 + @xmath52 & @xmath53 & 28.42 & 25.91 + @xmath7 & @xmath54 & 27.72 & 25.22 + @xmath55 & @xmath56 & 27.13 & 24.62 + @xmath57 & @xmath58 & 26.08 & 23.57 as with the main survey , it is important to use the various processing flags output by the photometric pipeline ( e.g. , as recommended by richards et al .
2002 ) to reject spurious objects , and to select objects with reliable photometry .
as was noted in the dr6 paper , the sdss imaging pipeline ( photo ) was designed to analyze data at high galactic latitudes , and is not optimized to handle very crowded fields .
the legacy survey is restricted to high latitudes , and photo performs adequately throughout the legacy footprint . however , at lower latitudes , when the density of stars brighter than @xmath59 grows above 5000 deg@xmath60 , the pipeline is known to fail , as it is unable to find sufficiently isolated stars to measure an accurate psf , and the deblender does poorly with overly crowded images .
many of the segue scans probe these low latitudes ( figure [ fig : skydist ] ) , and we therefore adapted an alternative stellar photometry code called ps . developed by the pan - starrs team ( kaiser et al .
2002 ; magnier 2006 ) to be used for these runs . in brief
, we first run this code , and then run photo using the list of objects detected by ps . as input to help photo s object finder in crowded regions .
this approach thus provides two sets of photometry at low latitudes . like e.g. , daophot ( stetson 1987 ) , ps
. begins with the assumption that every object is unresolved , and therefore does a better job than photo in crowded stellar regions .
it uses an analytical model based on gaussians to describe the basic psf shape , with parameters which may vary across the field of the image to follow the psf variations .
it also uses a pixel - based representation of the residuals between the psf objects and the analytical model , which is also allowed to vary across each field .
candidate psf stars are selected from the collection of bright objects in the frame by searching for a tight clump in the distribution of second moments . after rejecting outliers ,
the psf fit parameters are used to constrain the spatial variations in the psf model . unlike photo , ps .
processes each frame separately ( without any requirement of continuity of psf estimation across frame boundaries ) , and each filter separately ( without any requirement that the list of objects between the separate filters agree ) . the pipeline outputs positions and psf magnitudes ( and errors ) for each detected object ; the results are found in the psobjall table in the cas .
the resulting photometry is then matched between filters using a @xmath48 matching radius . while the estimated psf errors output by photo include a term from the uncertainty in the psf fitting , this component is not included in the errors reported by ps . .
we then run photo , asking it to carry out photometry at the position of each object detected by ps .
, in addition to the positions of objects photo itself detects .
this allows photo to do a much better job of distinguishing individual objects in crowded regions .
in addition , the pipeline is fine - tuned to less aggressively look for overlap between adjacent objects , and not to give up as soon as it does at high latitude when faced with deblending large numbers of objects .
we describe below how the photometry directly out of ps . , and that from photo , compare .
the sdss psf photometry had an offset applied to it to make it agree with aperture photometry of bright stars within a radius of @xmath61 ; this large aperture photometry was in fact what was used by ubercalibration to tie all the photometry together ( padmanabhan et al .
in crowded regions , finding sufficiently isolated stars to measure aperture photometry becomes difficult .
photometry was forced to agree with these large - aperture magnitudes for bright stars ; this was done in practice by determining , for each ccd in the imaging camera for each run , the average aperture correction needed to put the two on the same system , using stars at galactic latitude @xmath62 , where crowding effects are less severe .
if any part of a segue imaging run extended to @xmath63 , the entire run was processed through the photo and ps . code .
this sample includes most ( but not all ) of the segue imaging runs .
+ photo processed runs , designated with rerun=648 in the dr7 cas and das , are declared the best reduction of these segue runs .
there is also an inferior target version of these segue runs which was used to design segue spectroscopic plates ; it is based on photo alone , as the ps .
code was unavailable at the time the plates were designed .
the target reductions have rerun=40 , and are segregated to the seguetargdr7 database .
this processing revealed a problem with photo .
in crowded regions , one can not find sufficiently isolated stars to measure counts through such a large aperture , and in practice , the code corrected psf magnitudes to an aperture photometry radius of @xmath64 instead , whenever any part of a given run dipped below @xmath65 .
thus the aperture correction was underestimated , typically by 0.030.06 mag , depending on the seeing .
this was not a problem for any of the legacy imaging scans , but is very much an issue for the segue runs .
fortunately , there is a strong correlation , in a given detector , between the aperture correction from a @xmath64 aperture to a @xmath61 aperture ( as measured on high - latitude fields ) , and the seeing .
we therefore applied this correction after the fact to the photo psf , de vaucouleurs , exponential , and model magnitudes for all segue runs affected by this problem .
this was carried out before ubercalibration , so these runs are photometrically calibrated in a consistent way .
the quality of the photometry produced by ps . and by photo with the ps .
-detected objects as input , was evaluated by comparing the magnitudes computed by the two methods . within each field , we calculated the median of the difference of psf magnitudes for stars with @xmath66 .
this median difference had an rms of 0.014 mag .
fields with a difference greater than 0.08 mag are suspect , and further investigation is needed to determine which of the two pipelines might be at fault .
we followed mcgehee et al .
( 2005 ) to measure reddening - free colors of the same stars that track the stellar locus : @xmath67 where @xmath68 and @xmath69 .
these are normalized to equal zero at high galactic latitude ( note that these colors do _ not _ include the @xmath11 band ) .
median @xmath70 and @xmath71 colors in each field were computed for objects identified as stars in each field , and satisfying magnitude and color cuts as follows : @xmath72 , @xmath73 , @xmath74 , @xmath75 , and @xmath76 .
the @xmath77-parameters were found to be lower by up to 0.1 mag at low galactic latitudes ; to remove this effect , we fit a model of a constant plus lorentzian to the median @xmath77 values as a function of galactic latitude , and subtracted it .
the distributions of the @xmath70 and @xmath71 values for both photo and ps . are compared as density plots in figure [ fig : qparams ] . from equation [ eq : redfree ] ,
photometric errors in a single filter manifest themselves differently : @xmath78 as a shift in @xmath70 , @xmath79 as a line with slope @xmath80 , @xmath81 as a line with slope @xmath82 , and @xmath83 as a shift in @xmath71 .
the photo data in a given field was flagged as bad when either @xmath84 or @xmath85 mag ( @xmath86 ) as measured from photo magnitudes , and similarly for the ps .
outputs . of course
, a field could be flagged as bad in both sets of outputs . by this criterion , about 2% of the fields processed with ps .
were flagged bad based on the photo outputs , and 3.6% were bad based on ps .
the vast majority of the flagged fields are within @xmath87 of the galactic plane , and essentially all the fields in which the median difference between photo and ps .
photometry was greater than 0.08 mag in a given band were flagged as bad by the @xmath77 criteria .
this flag and the @xmath70 and @xmath71quantities themselves can be found in the fieldqa table in the cas . although more fields are flagged based on the ps .
outputs , the ps .
scatter in figure [ fig : qparams ] is tighter at both high and low galactic latitudes than for photo .
stellar photometry is therefore preferred for studies of the stellar locus ( we have not fully assessed its robustness to outliers ) , but comes with the caveat that fields flagged bad should be identified in the fieldqa table and be culled . an alternative check of sdss photometry in dense stellar fields was carried out by an et al .
( 2008 ) , who reduced the sdss imaging data for crowded open and globular cluster fields using the daophot / allframe suite of programs ( stetson 1987 , 1994 ) . at a stellar density of @xmath88 stars
deg@xmath60 with @xmath89 , they found @xmath90 rms variations in the difference between photo and daophot magnitudes in the scanning direction in all five bandpasses ( see their fig . 3 ) .
the systematic structures are likely due to imperfect modeling of the psfs in photo , given that daophot magnitudes exhibit no such large variations with respect to aperture photometry .
in other words , the psf variations were too rapid for the photo pipeline to follow over a time scale covered approximately by one field ( @xmath91 or @xmath92 sec in time ) .
an et al .
( 2008 ) further examined the accuracy of photo magnitudes in semi - crowded fields using three open clusters in their sample .
stellar densities in these fields were as much as @xmath93 times higher than those in the high galactic latitude fields , but photo recovered @xmath94% of stellar objects in the daophot / allframe catalog .
an et al .
( 2008 ) found that these fields have only marginally stronger spatial variations in photo magnitudes than those at lower stellar densities . and @xmath71 parameters measuring the position of the stellar locus within each field for the photo ( left ) and ps .
( right ) photometric pipelines ; zero values are indicative of uniform photometry . within the galactic plane ( lower panels ) , the ps .
values are more concentrated , but contain a higher number of systematic departures from the main locus .
code in fact gives a tighter locus at high latitudes as well ( upper panels ) .
histogram equalization of the gray - scale was used to emphasize low density regions .
[ fig : qparams],width=453 ] section 4.6 of the edr paper describes a series of flags available in the database to assess the quality of each field in the imaging data ; this includes information on whether the data in a given field meet survey requirements on seeing and sky brightness .
we have added additional criteria to assess the quality of each field .
the cas table called fieldqa includes a flag called problemchar associated with each field , which is set when : * the median of the telescope focus over three frames moved more than 60@xmath38 m , indicating a problem with the automated focus of the telescope ( gunn et al .
2006 ) ( problemchar = f ) ; * the rotator angle moved more than @xmath95 between adjacent fields ( corresponding to a @xmath96 image shift at the edge of the camera ) ( problemchar=r ) ; * the astrometric solution shifted by more than 4 pixels ( @xmath97 ) from a smooth interpolation between adjacent fields ( problemchar=a ) . * miscellaneous other problems , including voltage problems in the camera , and lights left on in the telescope ; this was triggered in only two imaging runs ( problemchar=s ) .
we flag all fields with these problems in any of the five bandpasses . because the imaging observations are done in driftscan mode ( gunn et al .
1998 ) , different areas of the sky are observed simultaneously in each bandpass and referenced to the field number of the @xmath7-band observation .
thus in the case of focus problems , we mark the 11 fields preceding , and the three fields following the field in question in all camera columns in the run as bad . for the rotator and astrometric shift problems
, we similarly mark the nine preceding and the one following field as bad .
only 0.3% of all fields in the cas are marked with one of these problems ( the majority of which are due to focus problems ) ; these flags should be consulted when examining the reliability of the photometry in a given area of sky .
early sdss imaging runs were astrometrically calibrated against tycho-2 ( hg et al .
2000 ) , which yielded statistical errors per coordinate for bright stars ( @xmath98 ) of approximately 75 mas and systematic errors of 20 30 mas .
later runs were calibrated against preliminary versions of the usno ccd astrograph catalog ( ucac , zacharias et al . 2000 ) , which yielded improved statistical errors per coordinate of approximately 45 mas , with systematic errors of 20 30 mas ( pier et al .
proper motions were not available for the preliminary versions of ucac .
since the typical epoch difference between the sdss and ucac observations is a few years and the typical proper motion of ucac stars is a few mas year@xmath99 , this introduces an additional roughly 10 mas of systematic error in the positions due to the uncorrected proper motions of the calibrating stars .
all of dr7 has been recalibrated astrometrically against the second data release of ucac ( ucac2 ; zacharias et al .
while the systematic errors for ucac2 are not yet well characterized , they are thought to be less than 20 mas ( n. zacharias , private communication ) .
ucac2 also includes proper motions for stars with @xmath100 . for stars at higher declination , proper motions from the sdss+usno - b catalog ( munn et al .
2004 ) have been merged with the ucac2 positions . with these improvements ,
all dr7 astrometry has statistical errors per coordinate for bright stars of approximately 45 mas , with systematic errors of less than 20 mas .
the mean differences per run between the old and new calibrations is a function of position on the sky , with typical absolute mean differences of 0 to 40 mas .
the rms differences are of order 10 to 40 mas for runs previously reduced against ucac , and 40 to 80 mas for runs previously reduced against tycho-2 , consistent with what we would expect given the errors in the reductions .
note that the formal sdss names of objects in the cas are of the form [email protected]@xmath102ddmmss.s . because of the subtle changes in the astrometry , these names will be slightly different for many objects between dr6 and dr7
. the user should be aware of this in comparing objects between dr6 and dr7 .
the cas includes proper motions for objects derived by combining sdss astrometry with usno - b positions , recalibrated against sdss ( munn et al .
these are given in the propermotions table in the cas .
an error was discovered in the proper motion code in data releases 3 through 6 , which causes smoothly varying systematic errors , in the proper motion in right ascension only , of typically 12 mas year@xmath99 ( see munn et al .
2007 for a full description of the problem and its effects ) .
this error has been corrected in dr7 , thus any use of proper motions should use the dr7 cas .
several of the segue target selection algorithms evolved during the course of sdss - ii .
the most significant changes occurred to the k - giant algorithm , as it was realized that good color - based luminosity separation could be done only for the very reddest ( @xmath103 ) giant candidates by their deviation from the main sequence locus in the @xmath104 color diagram ; this of course requires accurate @xmath11 band photometry .
early k giant target selection included stars with @xmath105 ( where the subscript refers to values after correcting for sfd galactic extinction ) as blue as 0.35 .
the final selection chose stars with @xmath105 between 0.5 and 1.2 , and was restricted to @xmath106 ; this gives much cleaner samples of k giants ( yanny et al .
2009 ) . in order to allow users to analyze completeness and efficiency of segue stellar target selection samples , the latest ( v4.6 ) version of the algorithms ( yanny et al .
2009 ) was applied to _ all _ stellar objects in the imaging catalog which had @xmath107 or @xmath108 , over the entire sky .
the appropriate bits were propagated into the segueprimtarget and seguesectarget fields of the photoobjall table of the dr7 cas .
a description of the bits and the target selection algorithms is available in yanny et al .
( 2009 ) .
as described in the dr5 paper , the sdss makes available the results of two different photometric redshift determinations for galaxies , one based on neural nets and the other based on a template - fitting approach . with dr7 , we include improvements to both , as we now describe . the neural net solutions for photometric redshifts and their errors ( listed as photoz2 in the cas , and described in oyaizu et al .
2008 ) have not changed since dr6 , and do not use the ubercalibrated magnitudes .
however , we now provide a value - added catalog containing the redshift probability distribution for each galaxy , @xmath109 , calculated using the weights method presented in cunha et al .
the @xmath109 for each galaxy in the catalog is the weighted distribution of the spectroscopic redshifts of the @xmath110 nearest training - set galaxies in the space of dereddened model colors and @xmath7 magnitude . for the @xmath109 calculation we also added the zcosmos ( lilly et al . 2007 ) and deep2-egs ( davis et al .
2007 ) galaxies to the spectroscopic training set used for the photoz2 solution .
cunha et al . ( 2008 ) showed that summing the @xmath109 for a sample of galaxies yields a better estimation of their true redshift distribution than that of the individual photometric redshifts .
mandelbaum et al . ( 2008 ) found that this gives significantly smaller photometric lensing calibration bias than the use of a single photometric redshift estimate for each galaxy .
with dr7 , we have made substantial improvements in the other photometric redshift code ( photoz ) , using a hybrid method that combines the template fitting approach of csabai et al .
( 2003 ; i.e. , the approach used in dr5 and dr6 ) and an empirical calibration using objects with both observed colors and spectroscopic redshifts .
we summarize the method briefly here , with details to follow in a paper in preparation .
the spectroscopic sample of sdss contains over 900,000 spectroscopically confirmed galaxies , and the combination of the main sample ( strauss et al .
2001 ) , the lrg sample ( eisenstein et al .
2001 ) and special plates targeted at fainter blue galaxies ( dr4 paper ) more or less cover the whole color region in which galaxies lie to the depths of sdss .
thus we use the dr7 spectroscopic set as a reference set for redshift estimation without any additional data from synthetic spectra .
the estimation method uses a k - d tree ( following csabai et al.2007 ) to search in the ubercalibrated @xmath111 , @xmath112 , @xmath113 , @xmath114 color space for the 100 nearest neighbors of every object in the estimation set ( i.e. the galaxies for which we want to estimate redshift ) and then estimates redshift by fitting a local hyperplane to these points , after rejecting outliers . if an object lies outside the bounding box of the 100 nearest neighbors in color space , the photometric redshift is less reliable , and the object is flagged accordingly .
we use template fitting to estimate the k - correction , distance modulus , absolute magnitudes , rest frame colors , and spectral type .
we search for the best match of the measured colors and the synthetic colors calculated from repaired ( budavari et al .
2000 ) empirical template spectra at the redshift given by the local nearest neighbor fit .
we have found that the mean deviations of the redshifts from the best - fit hyperplane is a good estimate of the error .
that , together with the flag indicating whether an object lies outside the bounding box of its neighbors , and the difference between the estimated photometric redshift and the average redshift of its neighbors , can be used to select objects with reliable photometric redshift values .
the rms error of the redshift estimation for the reference set decreases from 0.044 in dr6 to 0.025 in dr7 with this improved algorithm ( figure [ fig : photoz ] ) .
iteratively removing the outliers beyond @xmath115 gives rms errors of 0.028 and 0.020 for the old and new methods , respectively .
in addition , the reliability of the quoted errors is much higher . .
[ fig : photoz],width=453 ] the response functions of the sdss imager as a function of wavelength have been monitored throughout the survey .
the @xmath10 responses were stable over time , although very small seasonal ( i.e. , temperature ) variations were observed , at a level well below our typical photometric errors .
however , we have found a relatively large change in both the amplitude and shape of the @xmath11-band response , which is likely due to a degradation of the uv enhanced coating of the @xmath11-band ccd .
this change in instrumental zero - point is effectively corrected by the photometric calibration for objects near the mean color of the standard stars , and , in fact , the repeat photometry of stars in stripe 82 is stable with time for stars with @xmath116 ( bramich et al . 2008 ; ivezi et al .
however , the observed response changes involve a roughly 30 redward shift in the effective wavelengths of the @xmath11 filters over the lifetime of the survey , so one would expect significant changes in the measured colors of objects of extreme color over the period , and this is being investigated .
doi et al .
( in preparation ) will summarize the filter characteristics in full , including column - to - column variations within the camera and the changes with time .
as described in the dr6 paper and mandelbaum et al .
( 2005 ) , systematic errors in the estimation of the sky near bright ( @xmath117 ) galaxies causes their fluxes and scale sizes to be underestimated and the number of neighboring objects to be suppressed . indeed , a number of authors ( lauer et al .
2007 , bernardi et al .
2007 , lisker et al .
2007 ) have pointed out systematic errors in sdss galaxy photometry at the bright end . in the dr6 paper , this effect was quantified by adding simulated galaxies to the sdss images using a code described in masjedi et al .
these simulations found that the @xmath7 band brightness of galaxies was underestimated by as much as 0.8 magnitudes for a 12th magnitude galaxy with srsic index , @xmath118 ( an `` exponential '' , or disk galaxy ) . for @xmath119 galaxies ( `` de vaucouleurs '' , or elliptical galaxies ) ,
the effect was less pronounced , with a brightness underestimate of less than 0.6 magnitudes . however , the simulations shown in the dr6 paper used an incorrect relation between galaxy size and magnitude , in the sense that they overestimated the extent of the problem for the typical galaxy . using instead the relationships between apparent magnitude and half - light radius measured for sdss bulge and disk galaxies ( blanton et al .
2003 ) , we repeated the exercise : we simulated pure @xmath118 and @xmath119 galaxies with axis ratios @xmath120 of 0.5 and 1 , and added them to real @xmath7-band sdss images .
we ran the results through photo and compared their measured model magnitudes to their true magnitudes ; the bias in the measurement is shown as a function of true magnitudes in figure [ fig : delta_bright ] .
there is appreciable scatter at a given magnitude , due both to the changing background and the different axis ratios . on average , however , the flux is underestimated by approximately 0.2 magnitudes at @xmath121 and @xmath122 magnitudes at @xmath123 for simulated galaxies with a srsic index of 1 . for a srsic index of 4 ,
the flux is underestimated by as much as 0.3 magnitudes at @xmath121 .
the effect is more severe for simulated objects with an axis ratio of 1 than for an axis ratio of 0.5 ( see figure [ fig : delta_bright ] ) .
the scale sizes of galaxies are similarly underestimated by as much as 20% for simulated galaxies with srsic index of 1 , and 30% for an index of 4 .
of course , the most massive elliptical or cd galaxies will have more extended envelopes , producing a larger effect than we have found here ( lauer et al . 2007 ) .
-band magnitudes of a series of simulated galaxies with srsic index of 1 ( disk galaxies ; upper panel ) and 4 ( elliptical galaxies ; lower panel ) .
these galaxies followed the magnitude - effective radius relation observed in the sdss value - added galaxy catalog ( blanton et al .
2005 ) , and were either circularly symmetric ( circles ) or had an axis ratio of 0.5 ( diamonds ) .
they were added to random areas of real high - latitude fields , and run through photo .
the simulated elliptical galaxies show a systematic offset even at the faint end ; this is due to the fact that the photo model magnitude code assumes a truncation beyond 7 scale - lengths , while the `` true '' magnitude has no such truncation .
this is a 0.05 mag effect .
[ fig : delta_bright ] ]
spectroscopic flatfields for the blue camera in the first spectrograph contain an interference pattern produced by the dichroic .
the thickness of the dichroic coating is believed to be sensitive to the ambient humidity , and moisture which enters the system during plate changes affects the instrument response , shifting the interference pattern in wavelength in unpredictable ways on timescales comparable to the 900s exposure time .
the flats applied in processing were exposed several minutes prior to , or after , the science frames and therefore were not always representative of the true instrument response at the time of exposure .
the interference pattern is most pronounced in the 3800 - 4100 region of the spectrum .
if it shifts during an exposure , it will not be properly corrected by the flatfield , causing significant distortion of blue absorption lines in stellar spectra , and systematically affecting estimates of metallicities and surface temperatures .
flats obtained under different conditions were used to identify and model the stable and unstable ( shifting ) components of the flat , as shown in figure [ fig : superflat ] . with this model in hand , we searched for shifts in the interference pattern over the typically 45 minute time a given plate was observed by comparing the results of the individual 15-minute exposures for each object .
thus we took ratios of the extracted spectra from the separate exposures , and computed the median over all objects on a plate , giving results like those on the left - hand side of figure [ fig : ratios ] .
we fit this ratio to the results expected from a shifting interference pattern ( essentially a derivative of the shifting component in figure [ fig : superflat ] ) , with the only free parameter being the amount of shift , and divided out this remaining component in each spectrum .
the right - hand panel of figure [ fig : ratios ] shows that this technique removes the majority of the effects of the shifting interference .
an example is shown in figure [ fig : wiggle_spectrum ] , the spectrum of an a star observed on a plate where the interference term was particularly bad .
the shapes of the absorption lines , especially h@xmath124 at 3970 , is much more regular in the new reductions . ,
width=453 ] ) after allowing for an arbitrary wavelength shift .
[ fig : ratios],width=453 ] j172637.26 + 264127.6 , an a0 star observed as part of segue .
the strong broad lines are due to balmer absorption .
the red dashed spectrum is that available in dr6 , while the black solid spectrum is from dr7 , with its improved flat - field .
[ fig : wiggle_spectrum],width=453 ] the spectroscopic wavelength calibration is done quite accurately in sdss , with typical errors of 2 @xmath27 or better .
as the dr6 paper describes , however , detailed analyses of stellar spectra revealed occasional errors that were substantially larger than this , especially in the blue end of the spectrum .
the algorithms for fitting arc and sky lines were made more robust for dr6 , and this improved the situation considerably .
we have implemented two further improvements for dr7 : * spectroscopy is often done on nights with a moderate amount of moon .
the bluest sky line used for wavelength calibration is a hg line at 4046 , which is very close to a strong fe i absorption line in the solar spectrum .
thus when there is substantial moonlight in the sky spectrum , a fit to what is assumed to be an isolated emission line can be significantly biased , systematically skewing the wavelength solution at the blue end by as much as 20 @xmath27 . in dr7
, we now fit this line to a linear combination of a gaussian plus a stellar template including the absorption line , giving an unbiased estimate of the wavelength of the line . in practice , bright moon affected 10 plates ( listed in yanny et al .
2009 ) out of a total of 410 segue plates . *
the sky and arc lines for each fiber are fit to a wavelength solution ; this is done independently for each fiber .
this works well for the vast majority of plates .
however , for a small fraction of plates , the arcs are weak ( perhaps because the arc lamps themselves were faulty at that time , or because the petals which reflect the arc lamp light were not properly deployed ) , and the wavelength solution is poorly constrained .
we therefore required that second- and higher - order terms in the wavelength solution be continuous functions of fiber number , to constrain the solution .
we found that this produces much more robust wavelength solutions for those plates with weak arc observations , and has no substantial effect on the remaining plates .
the stellar spectral template library which gives the best radial velocity estimates is based on the elodie library ( prugniel & soubiran 2001 ) .
we have removed one elodie template that gave velocities with a consistent offset from the rest of the library , as measured using the sample of @xmath125 stars with duplicate observations on each segue plate pair . in order to provide more complete coverage in effective temperature , surface gravity and metallicity for hot stars
, we generated a grid of synthetic spectra using the models from castelli & kurucz ( 2003 ) over the same wavelength range and at the same resolving power as the spectra in the elodie library .
this blue grid spans 60009500k in 500k increments , @xmath126 > -2.5
$ ] in increments of 0.5 dex , and @xmath127 of 2 and 4 . we also added a grid of synthetic carbon enhanced spectra ( plez , private communication , using the stellar atmospheric code described by gustafsson et al .
2008 ) at values of [ fe / h ] between @xmath128 and @xmath129 , [ c / fe ] between 1 and 4 , @xmath127 values between 2 and 4 , and @xmath130 in the range 4000k6000k . with these improvements ,
the radial velocity scatter in repeat observations for objects that match the carbon star templates is now the same as for the full sample .
the dr6 paper describes a 7 @xmath27 systematic error in the radial velocities of stars ( in the sense that the pipeline - reported velocities are too small ) .
this is still with us in dr7 ; a correction is put into the outputs of the segue stellar parameter pipeline ( lee et al .
2008a ) but not elsewhere in the cas or das . beyond this problem ,
the plate - to - plate velocities of segue stars have systematic errors of about 2 @xmath27 in the mean .
the rms velocity error of any given segue star observation is about 5.5 @xmath27 at @xmath131 , degrading to 12 @xmath27 at @xmath132 .
the spectroscopic pipeline combines observations of a given object on the red and blue spectrographs , and between the separate 15-minute exposures on the sky , by fitting a tightly - constrained spline to the data , allowing discrepant points such as cosmic rays to be rejected .
this spline requires as input the effective resolution of the spectra . as described in the dr6 paper
, it did not do a perfect job ; occasionally , very strong and sharp emission lines were erroneously rejected by this algorithm .
this turned out to be due to the fact that the spline code did not adequately track the changing resolution of the spectra as a function of wavelength and fiber number . including this effect significantly improved the behavior of this algorithm .
figure [ fig : clipped_lines ] shows an example spectrum of an object affected by this problem in dr6 , and its improved counterpart in dr7 , as is apparent by the correct 3:1 ratio of the 5007 and 4959 lines of [ oiii ]
. emission line at 5020 was mistaken for a cosmic ray and clipped away completely in dr6 , while the weaker line at 4970 was slightly affected .
with the improved algorithm in dr7 , the lines are not clipped .
[ fig : clipped_lines],width=453 ] there is another problem , unfortunately not fixed in dr7 , which has a similar effect .
if the line is so bright that it is saturated in the individual 15-minute exposures of the spectrograph , it will also appear clipped .
the flux value corresponding to saturation is a function of wavelength , but ranges from 2000 to 10,000 times @xmath133 ( the units in which spectral flux density in reported in the sdss outputs ) .
unfortunately , such saturated pixels are not flagged as such , although usually they are recognizable as having an inverse variance equal to zero .
luckily , objects with such strong emission lines are very rare , but the user should be aware of the possibility of objects with extremely strong emission lines and unphysical or unusual line ratios
. there have been several improvements made in the segue stellar parameter pipeline ( sspp ; lee et al . 2008a ; 2008b
; allende prieto et al .
2008a ) since the release of dr6 .
in particular , in dr6 , the sspp under - estimated metallicities
( by about 0.3 dex ) for stars approaching solar metallicity .
this was fixed in dr7 by adding synthetic spectra with super - solar metallicities to two of the synthetic grid matching techniques ( ngs1 and ngs2 ) , and by recalibrating the caiik2 , acf , caiit , and annrr methods .
see table 5 in lee et al .
( 2008a ) for the naming convention for each technique .
two new techniques ( annrr and caiik3 ) were also added to the sspp metallicity estimation schemes , and contributed to the high - metallicity performance improvement .
two methods , acf and caiit , have been recalibrated to the native " @xmath112 system , instead of making use of calibration on @xmath134 , which required application of an uncertain transformation in color space .
the annrr approach , which also tended to under - estimate metallicity for near - solar metallicity stars , has been re - trained on the sdss / segue spectra with improved stellar parameters , resulting in a better determination of the metallicity for metal - rich stars .
moreover , a neural network approach , based solely on noise - added synthetic spectra , has also been introduced .
there remains a tendency for the sspp to assign slightly higher metallicities for stars with [ fe / h ] @xmath135 .
this offset is presently being calibrated out , and will be corrected in segue-2 ; see below . for more detailed descriptions of individual methods of the sspp
, we refer the interested reader to lee et al .
( 2008a ) .
additionally , the pipeline now identifies cool main sequence stars of low metallicity ( late - k and m subdwarfs ) .
the stars are assigned metallicity classes and spectral subtypes following the classification system of lpine et al .
cool and ultra - cool subdwarfs are classified as subdwarfs ( sdk , sdm ) , extreme subdwarfs ( esdk , esdm ) , and ultrasubdwarfs ( usdk , usdm ) in order of decreasing metal content .
the classification is based on the absolute and relative values of the tio and cah molecular bandstrengths , and derived from fits to k - m dwarf and k - m subdwarf spectral templates .
a number of open and globular clusters have been observed photometrically and spectroscopically with the sdss instruments to evaluate the performance of the sspp ( lee et al .
in addition , high - resolution spectra have been obtained for about 100 field stars included in the sdss , and used to expand the sspp checks over a wider parameter space ( allende prieto et al .
this paper marks the release of the final data of sdss - ii .
the original sdss science goals ( york et al .
2000 ) included five - band imaging over @xmath136 deg@xmath0 with 2% rms errors or better in photometric calibration , and spectroscopy of @xmath137 galaxies and @xmath138 quasars .
we have met these goals , and have in addition carried out extensive stellar spectroscopy of close to half a million stars , and repeat imaging over 250 deg@xmath0 to search for supernovae . over 2200 refereed papers have been published to date using sdss data or results , on subjects ranging from the large - scale distribution of galaxies to distant quasars to substructure in the galactic halo to surveys of white dwarfs to the color distribution of main belt asteroids . *
segue-2 extends the science goals of segue with the same instrumentation and data processing pipelines , but targets fainter stars to study the distant halo .
it will increase the number of distant halo stars by a factor of 2.5 with respect to the results of sdss and sdss - ii . *
the baryon oscillation spectroscopic survey ( boss ) will perform spectroscopy of 1.5 million luminous red galaxies to @xmath139 and 160,000 quasars with @xmath140 to measure the scale of the baryon oscillation signal in the correlation function as a function of redshift ( schlegel et al . 2007 ) . *
the multi - object apo radial velocity exoplanet large - area survey ( marvels ) will monitor the radial velocities of 11,000 bright stars to search for the signature of planets with periods ranging from several hours to two years ( ge et al .
* the apo galactic evolution experiment ( apogee ) will perform @xmath141 @xmath142-band spectroscopy of @xmath138 giant stars to @xmath143 for detailed radial velocity and chemical studies of the milky way ( majewski et al .
2008 ; allende prieto et al .
2008b ) .
this paper represents the end of sdss - ii , the culmination of a project taking two decades and involving an enormous number of scientists from all over the world .
we would like to dedicate this paper to colleagues who made essential contributions to the sdss but are no longer with us : john n. bahcall , don baldwin , norm cole , arthur davidsen , jim gray , bohdan paczyski , and david n. schramm .
the successful completion of this project is in large part a reflection of the hard work and intellectual capital they put into it .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is { \tt http://www.sdss.org/}[\{\tt http://www.sdss.org/ } ] .
the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , university of cambridge , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington .
lupton , r.h . ,
gunn , j.e .
, ivezi , . , knapp , g.r .
, kent , s. , & yasuda , n. 2001 , in _ astronomical data analysis software and systems x _ , edited by f. r. harnden jr .
, f. a. primini , and h. e. payne , asp conference proceedings , 238 , 269 | this paper describes the seventh data release of the sloan digital sky survey ( sdss ) , marking the completion of the original goals of the sdss and the end of the phase known as sdss - ii .
it includes 11663 deg@xmath0 of imaging data , with most of the @xmath1 deg@xmath0 increment over the previous data release lying in regions of low galactic latitude .
the catalog contains five - band photometry for 357 million distinct objects .
the survey also includes repeat photometry on a 120@xmath2 long , 2.5@xmath2 wide stripe along the celestial equator in the southern galactic cap , with some regions covered by as many as 90 individual imaging runs .
we include a coaddition of the best of these data , going roughly two magnitudes fainter than the main survey over 250 deg@xmath0 .
the survey has completed spectroscopy over 9380 deg@xmath0 ; the spectroscopy is now complete over a large contiguous area of the northern galactic cap , closing the gap that was present in previous data releases .
there are over 1.6 million spectra in total , including 930,000 galaxies , 120,000 quasars , and 460,000 stars .
the data release includes improved stellar photometry at low galactic latitude .
the astrometry has all been recalibrated with the second version of the usno ccd astrograph catalog ( ucac-2 ) , reducing the rms statistical errors at the bright end to 45 milli - arcseconds per coordinate .
we further quantify a systematic error in bright galaxy photometry due to poor sky determination ; this problem is less severe than previously reported for the majority of galaxies .
finally , we describe a series of improvements to the spectroscopic reductions , including better flat - fielding and improved wavelength calibration at the blue end , better processing of objects with extremely strong narrow emission lines , and an improved determination of stellar metallicities . |
in khler geometry , a basic question is to find on a khler manifold a canonical metric in each khler class , such as a khler
einstein ( k e ) metric , a constant scalar curvature khler ( csck ) metric , or even an extremal metric . if @xmath3 is a compact khler manifold with the definite first chern class , the question has been solved thoroughly and there are lots of references on this topic . among these
, the fundamental one @xcite is on the calabi conjecture solved by yau . in the non compact case , tian and yau proved in @xcite that there exists a complete ricci
flat metric on @xmath4 , where @xmath0 is a compact khler manifold and @xmath5 is a neat and almost ample smooth divisor on @xmath0 ; or @xmath0 is a compact khler orbifold and @xmath5 is a neat , almost ample and admissible divisor on @xmath0 .
several years ago , the second named author presented the following question : * problem 1*. assume that @xmath0 is a compact khler manifold and @xmath1 is its higher co - dimensional subvariety .
let @xmath6 . how to find a complete canonical metric on such a non - compact khler manifold @xmath7 ? certainly , this problem is equivalent to finding a canonical metric on @xmath8 , where @xmath9 is a compact khler manifold and @xmath5 is a divisor on @xmath9 .
more precisely , blowing up of @xmath0 along @xmath1 , one obtains a new compact khler manifold @xmath10 .
then @xmath7 is bi - holomorphic to @xmath8 where @xmath5 is the exceptional divisor of this blow - up .
hence our problem is transferred to finding a complete canonical metric on @xmath8 . however , this blowing up process can not make problem 1 easier since it does not alter the geometric properties of @xmath7 . for example ,
although @xmath11 is bi holomorphic to @xmath12 , we can not use tian
yau s results mentioned above to get a complete k e metrics on @xmath11 since the exceptional divisor @xmath5 is not ample .
the basic strategy to solve problem 1 is to perturb one family of approximate metrics on @xmath7 .
this method has been carried out successfully in @xcite to construct csck or extremal metrics on blow - up of a khler manifold at some points .
the key point is that the csck metric of burns - simanca @xcite on @xmath13 is asymptotic locally euclidean ( ale ) at infinity . motivated by this ,
if one want to solve problem 1 on @xmath14 , one should first construct a canonical metric on @xmath15 which is also ale at infinity .
fortunately , the metrics in the following theorem admit this asymptotic property .
let @xmath16 be the euclidean norm squared function on @xmath17 .
[ thm1 ] there exist on @xmath15 a family of complete zero scalar curvature khler metrics @xmath18 ( @xmath19 ) with the following asymptotic properties : as @xmath20 , @xmath21 and as @xmath22 , @xmath23 here @xmath24 is a smooth function whose @xmath25th partial derivatives for all @xmath26 are bounded by a constant times @xmath27 . for the cases of constant scalar curvature @xmath28 , we have the following theorem .
denote @xmath29 as the unit disc of @xmath17 .
[ thm2 ] 1 . for any @xmath30
, there exist on @xmath31 a family of complete khler metrics @xmath32 ( @xmath19 ) with constant scalar curvature @xmath33 . as @xmath34 ,
these metrics are asymptotic to the poincar metric @xmath35 \2 . for @xmath36 and @xmath19 with @xmath37
, there exists on @xmath15 a khler metric @xmath32 with constant scalar curvature @xmath33 which are not complete at infinity and asymptotic to @xmath38 for two constants @xmath39 and @xmath40 . in both cases ,
the metrics have the following asymptotic property : as @xmath20 , @xmath41 naturally one would ask whether there are any complete k
e metrics on @xmath15 or on @xmath42 . using the method in @xcite
, we can prove the following theorem which turns out that in some sense the choice of csck metrics is optimal .
[ thm3 ] there do not exist any complete khler - einstein metrics on @xmath15 or on @xmath31 .
in fact , we can prove @xmath43 can not admit any complete khler - einstein metrics in our further paper @xcite .
theorems [ thm1 ] and [ thm2 ] remind us to recall the mok
yau metric in @xcite . in 1980s ,
mok and yau introduced on @xmath42 the metric with bounded ricci curvature @xmath44 they used this metric to characterize domains of holomorphy by holomorphic sectional conditions . comparing the mok
yau metric with the metrics in theorems [ thm1 ] and [ thm2 ] leads to the following definition .
[ def ] let @xmath0 be a compact khler manifold with a khler metric @xmath45 and let @xmath1 be a higher co dimensional subvariety .
a khler metric @xmath46 on @xmath2 has the poincar
yau ( pmy ) asymptotic property if near the subvariety @xmath1 @xmath47 where @xmath48 is some distance function to @xmath1 , and @xmath49 and @xmath50 are two positive constants . in the second part of this paper ,
we generalize theorems [ thm1 ] and [ thm2 ] to the cases of holomorphic vector bundles .
we will use the moment construction to find complete csck metrics on the complement of the zero section in ( the total space of ) a holomorphic vector bundle or a projective bundle ( i.e. a ruled manifold ) .
there are many references such as @xcite which use the method of moment construction to look for canonical metrics on khler manifolds .
one can consult @xcite for construction of csck metrics on vector bundles and @xcite for extremal metrics on ruled manifolds .
let @xmath51 be a compact @xmath52dimensional khler manifold with a csck metric @xmath53 .
let @xmath54 be a holomorphic line bundle over @xmath51 with a hermitian metric @xmath55 , which is given by local positive functions @xmath56 defined on the open sets which locally trivialize @xmath57 .
for the technical reason , assume that there exists a constant @xmath58 such that @xmath59 let @xmath60 be the direct sum of @xmath61 copies of @xmath57 with associate hermitian metric @xmath55 .
let @xmath62 be the logarithm of the fibre norm squared function defined by @xmath55 and consider calabi s ansate @xmath63 denote the zero section of @xmath64 simply by @xmath51
. also denote @xmath65 as the set of points @xmath66 in @xmath64 such that @xmath67 .
we first concern about csck metrics with _
pmy asymptotic property _ on @xmath68 or on @xmath69 .
[ thmlag0 ] let @xmath51 be a compact khler manifold and @xmath53 a kler metric with constant scalar curvature @xmath70 .
let @xmath57 be a holomorphic line bundle over @xmath51 with a hermitian @xmath55 .
assume that @xmath55 and @xmath53 satisfy ( [ lo ] ) .
let @xmath64 be the direct sum of @xmath61 copies of @xmath57 .
if @xmath71 , there exists a constant @xmath72 such that for any @xmath73 , there exists on @xmath69 or on @xmath68 a complete khler metric with constant scalar curvature @xmath33 .
such metrics admit the poincar mok
yau asymptotic property except the case that the metrics are defined on @xmath69 with @xmath74 and @xmath75 .
if @xmath76 and @xmath77 , there exists on @xmath68 a complete positive constant scalar curvature khler metric with poincar mok
yau asymptotic property .
if @xmath78 , this theorem generalizes theorem [ thm1 ] and the case @xmath30 of theorem [ thm2 ] .
we then consider problem 1 on a projective bundle .
denote @xmath79 as the structure sheaf of @xmath51 .
the projective bundle @xmath80 over @xmath51 has a globally defined section @xmath81 : for @xmath82 , @xmath83 is a point corresponding to the line @xmath84 .
the following theorem gives some special solutions to problem 1 .
[ thmpm ] under the assumptions of theorem [ thmlag0 ] , if @xmath76 and @xmath85 or if @xmath86 and @xmath87 , there exists on @xmath88 a complete constant scalar curvature khler metric with poincar mok yau asymptotic property
. * acknowledgement .
* fu is supported in part by nsfc grant 11421061 .
yau is supported in part by nsf grant dms-0804454 .
zhou is supported by china postdoctoral science foundation funded project grant no.2015m571479 .
in this section , we construct complete csck metrics on @xmath89 or @xmath90 . here denote @xmath91 and @xmath92 . we first follow calabi s method @xcite to get an ode on the khler potential .
then we determine the constants of integration appeared in the ode by discussing completeness of the metric near the punctured point .
afterwards we analyze the asymptotic properties of the khler potential .
thus , theorems [ thm1 ] and [ thm2 ] are proven . in the last subsection , we give some remarks and a simple proof of theorem [ thm3 ] .
let @xmath93 be the coordinates of @xmath17 .
assume that the khler metric we are seeking for is rotationally symmetric .
that is , if we let @xmath94 then the khler potential is a function @xmath95 . by a direct calculation , @xmath96 hence , @xmath97 and @xmath98 is a khler metric if and only if @xmath99 for simplicity , let @xmath100 the components of the ricci tensor of @xmath46 are @xmath101 and then the scalar curvature is @xmath102 here @xmath103 denotes the inverse matrix of @xmath104 .
explicitly , @xmath105 assume that the scalar curvature of @xmath46 is a constant @xmath33 . integrating ( [ c(t ) ] ) with the integrating factor @xmath106 , we obtain the first order differential relation between @xmath95 and @xmath107 @xmath108 with an arbitrary constant @xmath109 . substituting ( [ v ] ) to the above equation and multiplying both sides with @xmath110 , we get the equation @xmath111integrating the above equation , we obtain @xmath112 with another arbitrary constant @xmath113 . if we denote @xmath114 , then the above equation can be written as the first order differential equation @xmath115 with @xmath116 or rewritten as @xmath117 it follows that @xmath95 is a khler potential if and only if @xmath118 assume that @xmath119 is a solution to ode ( [ phia ] ) or ( [ dt ] ) in an interval @xmath120 , where @xmath121 can be equal to @xmath122 , and assume that it determines a khler potential @xmath123 in the punctured disc @xmath124 with radius @xmath125 . in this subsection , for the sake of the completeness of @xmath126 near the punctured point , the constants @xmath109 and @xmath113 in ode ( [ phia ] ) can be determined .
the key point is the following observation .
[ lemc ] under the above assumption , the metric @xmath46 determined by the khler potential @xmath95 is complete near the punctured point if and only if @xmath127 has a factor @xmath128 with @xmath19 and @xmath129 .
hence @xmath130 since the metric @xmath98 is rotationally symmetric , for any point @xmath131 , the ray @xmath132 , @xmath133 $ ] , is a geodesic .
the tangent vector of this curve at the point @xmath134 is @xmath66 and its square norm under the metric @xmath46 is , if we assume that @xmath135 , @xmath136 then the length of @xmath137 is @xmath138 under the assumption that @xmath139 and @xmath140 , there is a nonnegative constant @xmath49 such that @xmath141 by equation ( [ dt ] ) , we have @xmath142 the completeness requires @xmath143 , which is equivalent to the fact that @xmath127 has a factor @xmath128 . hence , we can determine @xmath109 and @xmath113 as in ( [ 629 ] ) .
we claim @xmath19 . if @xmath144 , @xmath145 .
hence , @xmath146 , which leads to a contradiction . because of completeness , in this subsection assume that the constants @xmath109 and @xmath113 have been chosen as in ( [ 629 ] ) .
hence , @xmath147 .
since @xmath148 in case @xmath149 , @xmath150 and so @xmath151 on domain @xmath152 . in case @xmath36 , if assume that the constants @xmath49 and @xmath33 satisfy @xmath153 then on domain @xmath154 , @xmath150 and so @xmath151 .
hence , @xmath151 on domain @xmath155 for some constant @xmath50 or @xmath156 .
thus , we obtain the solution of equation ( [ dt ] ) , up to a constant : @xmath157 for a given @xmath158 . since @xmath127 has the factor @xmath128 , @xmath159 . in the following
we will discuss more details of the solutions ( [ sol1 ] ) for different signs of @xmath33 and finish the proofs of theorems [ thm1 ] and [ thm2 ] .
case @xmath160 .
* in this case , @xmath161 since the only root of @xmath162 is @xmath49 , @xmath127 obtains its minimum at the point @xmath49 and @xmath151 for all @xmath163 . as @xmath164 , @xmath165 and solution ( [ sol1 ] ) has the property that @xmath166 .
hence @xmath119 is defined on the entire punctured space @xmath89 .
therefore there exist a family ( depending on @xmath19 ) of zero csck metrics with khler potential @xmath95 such that @xmath167 . since @xmath168 as @xmath169
, @xmath170 it turns out that from @xmath171 , @xmath172 moreover , by lhsital s rule we get the more accurate expression of @xmath173 : @xmath174 on the other hand , we divide the case into @xmath175 and @xmath176 to discuss the approximation of the solution as @xmath177 . for @xmath175 , @xmath178 and it follows that @xmath179 and @xmath180 . obviously , the derived metric is complete at entire @xmath181 .
moreover , lhsital s rule can be used to get more accurate estimate @xmath182 for @xmath176 , as @xmath22 @xmath183 and then @xmath184 which implies @xmath185 replacing @xmath186 by @xmath187 in the right hand side of the above equality , we have @xmath188 or @xmath189 thus we have finished the proof of theorem [ thm1 ] .
we give the picture of @xmath119 with @xmath175 and @xmath190 as figure [ p1 ] .
recall that in this situation the function @xmath119 is defined in equation ( [ 70901 ] ) .
note that we also have @xmath191 we give a rotational picture of the function @xmath192 as figure [ p2 ] which shows the ale and pmy properties of the metric . with @xmath160 and @xmath190.,scaledwidth=50.0% ]
case @xmath30 . * in this case , we have seen that @xmath151 when @xmath193 . from ( [ sol1 ] ) , we also see that when @xmath194 , the upper bound of @xmath195 exists since the degree of @xmath196 is @xmath197 . for simplicity , we take this upper bound to be zero since the solution ( [ sol1 ] ) is unique up to be a constant . then @xmath173 is defined on the punctured unit disc @xmath90 .
the analysis of the boundary behavior is as follows . as @xmath198 , @xmath199 which implies @xmath200 or @xmath201 where the right hand side is the khler potential of the standard poincar metric on @xmath29 .
hence , the metric we constructed is also complete near the boundary of @xmath29 . for the asymptotic behavior of @xmath119 at the origin ,
it is the same as for the case @xmath160 : as @xmath20 , @xmath202 then we have finished the proof of theorem [ thm2 ] for the case @xmath30 .
we give the picture of @xmath119 as @xmath175 , @xmath190 and @xmath203 as figure [ p3 ] .
note in this situation , @xmath204 with @xmath203.,scaledwidth=50.0% ] * 3 .
case @xmath36 .
* we have seen that if the constants @xmath33 and @xmath49 satisfy the relation @xmath205 , then on domain @xmath206 , @xmath151 . obviously , when @xmath186 is big enough , @xmath207 .
hence we can let @xmath50 be the first number in @xmath152 such that @xmath208 .
it follows that there is a polynomial @xmath209 such that we can write @xmath210 we first claim @xmath211 . if @xmath212 , then @xmath213 .
together with @xmath147 , there are at least two different positive roots for the equation @xmath214 . however , equation @xmath215 has only one positive root @xmath216 , which leads to a contradiction .
hence , as @xmath217 , @xmath218 where @xmath219 this implies that when @xmath220 , @xmath221 and then the khler potential @xmath173 is defined on entire @xmath181 .
the approximation of @xmath173 near the zero is the same as for the case @xmath160 : @xmath202 whereas when @xmath222 , from ( [ ffd ] ) we can derive @xmath223 the metric is not complete as @xmath22 .
in fact as in the proof of lemma [ lemc ] , the length of the geodesic ray @xmath132 on domain @xmath224 is @xmath225 thus we have finished the proof of theorem [ thm2 ] for the case @xmath36 .
we give the picture of function @xmath119 in case @xmath175 , @xmath190 and @xmath226 as figure [ p4 ] .
in this situation , @xmath227 with @xmath226.,scaledwidth=50.0% ] \1 . for @xmath228 , @xmath229 and @xmath230 is the standard poincar metric on @xmath231 with gauss curvature @xmath232 .
one can also construct on @xmath233 a complete metric with zero gauss curvature @xmath234 \2 .
it is mentioned in introduction that the mok
yau metric defined on @xmath235 has good properties .
one can see that @xmath236 is also a khler metric on @xmath235 . however , its scalar curvature is infinity as @xmath20 .
in fact , the term @xmath237 in the mok
yau metric results in the boundedness of the scalar curvature near the punctured point .
hence , the asymptotic property appeared in definition [ def ] is named as the pmy asymptotic property .
write @xmath238 and viewed zero as a point @xmath66 in @xmath239 .
one will ask whether the metric on @xmath89 constructed above with @xmath36 can be extended across @xmath240 .
this is impossible .
in fact , it can be seen form lemma [ lem4 ] in section [ spec ] below that the metric @xmath241 can be extended across @xmath240 if and only if @xmath242 replacing @xmath186 by @xmath243 and @xmath33 by @xmath244 , we can assume that @xmath190 in @xmath127 . from @xmath245 and @xmath208 , we get the relation @xmath246 then @xmath247 implies @xmath248 , which is a contradiction to ( [ kap ] ) .
so in this way we can not get a complete csck metric on @xmath249 . in our another paper @xcite
, it has been proved that there also do not exist any complete csck metrics on @xmath250 with pmy asymptotic property .
nevertheless , a family of complete extremal metrics on @xmath250 have been constructed in @xcite . \4 . at last , we give a simple proof of theorem [ thm3 ] which states that there are not any complete k
e metrics on @xmath181 or on @xmath251 .
since @xmath89 and @xmath90 are not compact , by myers theorem , we only need to consider the cases @xmath149 . in @xcite ,
mok and yau proved that if a bounded domain @xmath252 admits a complete hermitian metric such that @xmath253 , then @xmath254 is a domain of holomorphy .
since @xmath251 is not a holomorphic domain , it does not admit a complete e
k metric with @xmath149 . for the nonexistence of negative k
e metrics on @xmath181 , we use generalized yau s schwarz lemma .
@xcite let @xmath255 be a complete hermitian manifold with scalar curvature bounded below by @xmath256 and let @xmath257 be a hermitian manifold of the same dimension with ricci curvature @xmath258 for some @xmath259 . if @xmath260 is a holomorphic map and the jacobian is nonvanishing at one point , then @xmath261 and @xmath262 now take @xmath263 and take the metric @xmath264 on @xmath51 as in theorem [ thm1 ] .
then the above lemma leads to the nonexistence of negative k
e metric on @xmath265 .
as for the ricci
flat case , if we let @xmath266 be a complete ricci
flat metric on @xmath89 , then the function @xmath267 is pluriharmonic on it .
since the de rham cohomology group @xmath268 vanishes , @xmath267 is the real part of a holomorphic function . by hartogs extension theorem for holomorphic functions
, @xmath267 is pluriharmonic on the entire space @xmath17 .
hence , @xmath269 for some positive constant @xmath270 . then following the discussions on page 49 of @xcite
, one can get a contradiction to the completeness of the metric near the origin of @xmath17 .
this section is devoted to prove theorem [ thmlag0 ] .
the first subsection almost follows the paper @xcite .
that is we first use the calabi s ansatz to derive an ode and then use the moment profile to simplify it . in the second subsection completeness of metrics near zero section and at infinity
are used to get constraint conditions . in the third subsection
, we then consider the existence of metrics and their asymptotic property .
let @xmath51 be a compact khler manifold with a khler metric @xmath53 .
let @xmath271 be a holomorphic line bundle with a hermitian metric @xmath55 .
for any point @xmath82 , there is a holomorphic coordinate system @xmath272 of @xmath273 with @xmath274 such that @xmath275 is holomorphically trivial . under this trivialization
, the hermitian metric @xmath55 can be given by a positive function @xmath56 .
assume that @xmath53 and @xmath55 satisfy the condition @xmath276 let @xmath64 be the direct sum of @xmath61 copies of @xmath57 , i.e. @xmath277 , with an associated hermitian metric still denoted by @xmath55 .
we also denote @xmath278 .
we have a local trivialization of @xmath64 induced from one of @xmath57 and denote the fiber coordinates by @xmath279 . in the following , we denote @xmath280 and @xmath281 as the lower index of the components of @xmath282 and @xmath283 and @xmath284 as the lower index of the components of @xmath285 . there is a fibrewise norm squared function @xmath16 on the total space of @xmath64 defined by @xmath55 @xmath286 if @xmath51 is viewed as the zero section of @xmath64 , i.e. the set defined by @xmath287 , we want to construct complete csck metrics on @xmath288 under condition ( [ 70101 ] ) . in this section denote @xmath68 simply by @xmath289 .
let @xmath290 consider calabi s ansatz @xmath291 using condition ( [ 70101 ] ) , we have @xmath292 and @xmath293 the reason one can derived the above equality is @xmath294 . in practise , when computing at any point @xmath295 , one can let @xmath296 for @xmath297
. then @xmath46 is ( strictly ) positive if and only if @xmath298 the above constructed metric @xmath46 is call a bundle adapted metric .
next we compute the ricci curvature and the scalar curvature of the bundle adapted metric @xmath46 . from ( [ 70102 ] ) , we have @xmath299 let @xmath300 for any @xmath82 , we can assume that the local coordinates @xmath301 around @xmath273 also satisfy @xmath302 .
then under assumption ( [ 70101 ] ) , the ricci form of @xmath46 at a point @xmath303 is @xmath304 where @xmath305 is the ricci form of @xmath53 on @xmath51 .
the matrix composed by components of metric @xmath46 at @xmath66 is @xmath306 where @xmath307 is the coefficients matrix of metric @xmath53 .
its inverse matrix is @xmath308 if @xmath70 denotes the scalar curvature of @xmath53 , the scalar curvature of @xmath46 at the point @xmath66 is @xmath309 the above formula of scalar curvature is obviously globally defined .
usually it is more suitable to use the legendre transform to solve scalar curvature equation ( [ cnu ] ) . from the positivity ( [ pos2 ] ) of @xmath46 , @xmath310 must be strictly convex
. then one can take the legendre transform @xmath311 of @xmath310 .
the legendre transform @xmath311 is defined in term of the variable @xmath312 by the formula @xmath313 let @xmath314 be the image of @xmath315 .
the _ momentum profile _ @xmath316 of the metric is defined to be @xmath317 , @xmath318 then the following relations can be verified : @xmath319 also , @xmath62 can be viewed as a function of @xmath320 , up to a constant , @xmath321 the advantage of the legendre transform is that the scalar curvature of @xmath46 can be described in terms of @xmath316 and the domain @xmath322 of @xmath320 .
especially , the boundary completeness ( see the next subsection ) and the extendability properties ( see the next section ) can also be read off from the behaviour of @xmath323 near the end points of @xmath322 . using these transformations ,
we have @xmath324 let @xmath325 then @xmath326 . by direct computation ,
@xmath327 inserting the above equalities into ( [ cnu ] ) , and replacing @xmath315 by @xmath320 , then simplifying it , at last we obtain @xmath328 now we assume that @xmath70 is a constant .
we want to look for @xmath329 such that @xmath33 is a constant . by integrations
, we solve equation ( [ 70201 ] ) as @xmath330 where @xmath49 is the left endpoint of @xmath322 . here
@xmath331 and @xmath332 are constants .
@xmath333 is a polynomial and hence @xmath316 is a real rational function .
assume that @xmath334 , @xmath50 may be infinity , is the maximum interval where @xmath316 determined by ( [ 70401 ] ) is defined and positive .
further assume that @xmath335 and @xmath336 then from the above subsection , the bundle adapt metric @xmath46 is well - defined on @xmath337 . here
@xmath338 . if @xmath339 is infinity , then @xmath340 , and if @xmath339 is a constant , we can take an integration constant in ( [ ttau ] ) such that @xmath341 and hence @xmath342 is @xmath343 as defined in introduction .
we first establish the following lemma .
[ 070501 ] under the above assumptions , the bundle adapt metric @xmath46 is complete near the zero ( punctured ) section if and only if @xmath19 and @xmath344 .
thus , @xmath345 .
moreover , if @xmath50 is finite , then @xmath46 is defined on @xmath289 and @xmath46 is complete if and only if @xmath329 also satisfies @xmath346 ; whereas if @xmath50 is infinity , then @xmath46 is defined on @xmath289 or on @xmath347 and is automatically complete . for any point @xmath82 , @xmath348 , we consider the ray starting from @xmath273 on the fiber @xmath349 : @xmath350\ \
\text{or}\ \ s\in [ 1,s_0)\ ] ] for @xmath351 .
such an @xmath352 can be derived from the following calculation by ( [ 70405 ] ) : @xmath353 since @xmath354 is a totally geodesic submanifold of @xmath355 and the induced metric on @xmath356 is @xmath357-invariant , @xmath137 is a geodesic on @xmath356 , and hence is also a geodesic on @xmath342 . also since @xmath51 is compact , the metric @xmath46 is complete if and only of the lengths of the rays @xmath137 defined in ( [ 70406 ] ) are infinity . as done in lemma [ lemc ] ,
the length of @xmath137 on domain @xmath358 $ ] is @xmath359 the completeness near the zero section requires @xmath360 , which is equivalent to that @xmath316 has a factor @xmath361 , i.e. @xmath344 .
we claim @xmath19 .
if @xmath144 , the lowest degree term of polynomial @xmath333 defined in ( [ 70401 ] ) would be determined as @xmath362 hence we can write @xmath333 as @xmath363 for some polynomial @xmath364 and thus get @xmath365 in this way we find @xmath366 , which is a contradiction .
next , we should consider the endpoint @xmath50 .
the length @xmath367 of @xmath137 for @xmath368 is @xmath369 if @xmath50 is finite , then @xmath333 has a factor @xmath370 . by ( [ ttau ] ) , @xmath371 .
hence , @xmath46 is well - defined on @xmath289 . if @xmath46 is complete , the above @xmath367 is also infinity , which is equivalent to say that @xmath346 .
if @xmath50 is infinity , then @xmath316 is defined on @xmath152 . since @xmath372 or 1 , from ( [ ttau ] ) if @xmath373 , @xmath46 is defined on @xmath347 and if @xmath374 , @xmath46 is defined on @xmath289 . also since @xmath373 or @xmath375
, there exists a constant @xmath270 big enough such that @xmath376 as @xmath377 .
hence , in this situation , @xmath367 is infinity and @xmath46 is automatically complete .
we discuss the solutions in this subsection divided into three cases : @xmath378 , @xmath78 and @xmath76 . *
1 . case @xmath378 . * given constants @xmath70 , @xmath378 , and @xmath19 , define the set @xmath379 to be of allowable scalar curvatures " as @xmath380 @xmath381 is not empty since @xmath316 is positive if @xmath382 .
@xmath383 has a supermum .
in fact , if @xmath36 , @xmath333 and hence @xmath316 will be negative when @xmath320 is big enough .
hence , the supermum @xmath72 of @xmath379 is nonpositive .
we can easily get the conclusions : if @xmath384 , @xmath385 ; if @xmath386 , two possibilities occur : one is @xmath387 , and the other is @xmath388 , which means that there exists a constant @xmath50 such that @xmath389 and @xmath316 is positive on @xmath155 .
hence we should consider the existence of khler metrics with constant scalar curvature @xmath33 as the following four cases : @xmath390 [ prop3 ] given constants @xmath70 , @xmath378 and @xmath19 , there exists a constant @xmath391 such that : \1 .
for any @xmath73 in cases ( i ) and ( iii ) , there exists a complete csck metric @xmath46 on @xmath347 with constant scalar curvature @xmath33 ; and \2 . for @xmath392 in cases ( ii ) and ( iv ) , there exists a complete csck metric @xmath46 on @xmath289 with constant scalar curvature @xmath33 . for cases ( i ) and
( iii ) , since the degrees of polynomials @xmath333 and @xmath393 are @xmath394 and @xmath395 respectively , the limit of @xmath396 defined by ( [ ttau ] ) is finite as @xmath397 .
set this constant to be zero .
then the metric @xmath46 is defined on @xmath347 . according to lemma [ 070501 ]
, @xmath46 is complete . for case ( ii ) , @xmath398 .
the limit of @xmath396 is infinity as @xmath397 .
hence the metric is defined on @xmath289 and is complete by lemma [ 070501 ] . for case ( iv ) ,
@xmath399 for @xmath400 . in this case
there exists a constant @xmath50 such that @xmath401 in @xmath402 and @xmath389 .
hence @xmath403 . according to lemma [ 070501 ] ,
@xmath46 is defined on @xmath289 and is complete .
we give two examples .
[ ee ] consider the case @xmath387 and @xmath404 .
take @xmath405 we have @xmath406 if @xmath386 , @xmath75 and @xmath19 satisfy @xmath407 we find that @xmath408 when @xmath400 .
it is easy to check that @xmath409 [ bb=0 ] then consider the case @xmath388 .
let @xmath410 , @xmath175 , @xmath411 , @xmath412 and @xmath190 .
it follows that @xmath413 we can solve the inequality @xmath399 to get @xmath414 . here
@xmath415 hence @xmath346 if and only if @xmath416 .
the pictures of @xmath417 and @xmath316 are given as figure [ c ] .
we find that @xmath417 achieves its maximum at @xmath418 with the maximum @xmath419 .
so @xmath420 and @xmath421 .
thus , @xmath316 gives a complete csck metric on @xmath289 with scalar curvature @xmath72 .
we consider the asymptotic property .
let @xmath422 since @xmath344 , @xmath423 .
[ pop ] for cases ( i ) , ( ii ) and ( iv ) the csck metrics in proposition [ prop3 ] have the pmy asymptotic property , and for case ( iii ) , the metrics have the asymptotic property : as @xmath20 , @xmath424 or @xmath425 for cases ( i ) , ( ii ) and ( iv ) , we claim that @xmath426 .
if the claim holds , then as @xmath20 , @xmath427 from which we can get @xmath428 which means that the metric is with pmy asymptotic property .
we prove the claim .
for case ( i ) , since @xmath429 , @xmath430 .
for case ( ii ) , if @xmath431 , then @xmath432 and @xmath316 has the taylor expansion at @xmath433 : @xmath434 the positivity of @xmath316 when @xmath435 implies @xmath436 . on the other hand , if @xmath431 , i.e. , @xmath437 as @xmath160 , then @xmath438 which is a contradiction to @xmath439 .
hence @xmath440 and @xmath426 is deduced from the positivity of @xmath316 .
we then consider case ( iv ) .
in this case , @xmath441 and @xmath442 these equalities guarantee that there are already two roots in ( a , b ) for @xmath443 .
hence @xmath444 .
the positivity of @xmath329 then implies @xmath426 . for case ( iii )
, we first prove that @xmath431 which means that the metrics for this case is not pmy .
since @xmath445 , there exist constants @xmath446 and @xmath447 such that @xmath448 in @xmath449 . if @xmath426 , there would exist constants @xmath450 and @xmath451 such that @xmath452 in @xmath453 and @xmath454 in @xmath455 $ ]
. then we could choose a positive constant @xmath456 such that @xmath316 is still positive by replacing @xmath392 with @xmath457 .
hence @xmath458 which contradicts to that @xmath72 is the supermum of @xmath379 . in this case , there are two subcases which should be considered : @xmath459 and @xmath460 : if @xmath459 , @xmath461 if @xmath460 , @xmath462 and @xmath463 the conclusion then follows .
[ poop ] the metrics in proposition [ prop3 ] with constant scalar curvature @xmath33 have the asymptotic property : \1 . for cases ( i ) and ( iii ) ( hence defined on @xmath347 ) , as @xmath34 , @xmath464 \2 . for case ( ii ) ( hence defined on @xmath465 ) , as @xmath177 , @xmath466 with @xmath467 and @xmath468 ; \3 . for case ( iv )
( hence defined on @xmath289 ) , as @xmath22 , @xmath469 we omit the proof here .
it is a calculus exercise . *
2 . case @xmath78 . * in this case , @xmath470 .
hence @xmath471 .
[ prpmy ] given constants @xmath70 , @xmath78 , and @xmath19 , there exists a complete csck metrics on @xmath347 with @xmath472 and on @xmath289 with @xmath473 .
all these metrics admit pmy asymptotic property .
the proof is the same as propositions [ prop3 ] and [ pop ] . *
3 . case @xmath76 .
* the method of this case is quite different from the cases @xmath474 .
[ ppppp ] for any @xmath76 and @xmath77 , there exists on @xmath289 a complete positive csck metric with pmy asymptotic property .
we need to prove that there exist two constants @xmath49 and @xmath50 with @xmath475 such that the function @xmath316 is positive on domain @xmath155 and @xmath346 . on interval @xmath476 ,
the polynomial @xmath393 is positive .
hence there is a number @xmath477 such that @xmath346 if and only if @xmath478 , and when @xmath402 , @xmath316 is positive if and only if @xmath333 is positive . from @xmath478 , we can solve @xmath70 and @xmath33 as @xmath479 where we have defined @xmath480 we first note that when @xmath76 and @xmath475 , the functions @xmath481 for @xmath482 are always positive .
the proofs for the first and second functions are direct by the hlder inequality .
the proof for the third one is also direct by using the common techniques in calculus .
thus the constant @xmath70 and @xmath33 defined in ( [ cc ] ) are indeed positive .
we need the following .
* claim : * for any given positive @xmath70 , there exist two constants @xmath49 and @xmath50 with @xmath475 such that the first equality in ( [ cc ] ) holds .
define a function @xmath483 by continuity , if we can prove that as @xmath484 , @xmath485 , and as @xmath486 , @xmath487 , then the claim holds .
but as @xmath484 and ( hence ) @xmath488 , one can easily estimate to get @xmath489 and @xmath490 , and hence @xmath491 . on the other hand , as @xmath486 , @xmath492 , and @xmath493 and hence @xmath494 .
according to the claim , we have @xmath478 .
the condition @xmath408 for @xmath402 is automatically satisfied . for if there exists a point @xmath495 such that @xmath496 , equation @xmath497 has three roots in @xmath155 .
this is impossible . from the proof of proposition [ pop ]
, we see that if @xmath426 , then the metric has the pmy asymptotic property .
since @xmath498 , @xmath431 is equivalent to @xmath499 , and hence @xmath500 has three roots in @xmath155 .
it is impossible .
it follows from propositions [ prop3 ] , [ pop ] , [ poop ] , [ prpmy ] and [ ppppp ] .
recall that @xmath502 can be viewed as a compactification of @xmath64 : @xmath64 can be imbedded into @xmath503 .
in fact , let @xmath272 be a local holomorphic chart of @xmath51 such that @xmath504 is ( holomorphically ) isomorphic to @xmath505 .
if we denote the coordinates of @xmath17 as @xmath279 , the imbedding map can be defined as follows : for any @xmath506 , @xmath507 , @xmath508).\ ] ] this map is clearly well - defined on @xmath64 .
it defines a section @xmath81 of @xmath80 : @xmath509)\ ] ] which is just the zero section of @xmath64 .
hence we still denote @xmath510 simply by @xmath51 .
set @xmath511 .
@xmath512 is a divisor on @xmath80 and is called the infinity divisor . by these notations
, @xmath68 is bi - holomorphic to @xmath513 .
now the question is when the metric @xmath46 defined on @xmath68 as the above section can be extended across @xmath512
. first note that if @xmath46 can be extended across @xmath512 , @xmath46 must be defined on @xmath68 and is not complete at infinity . hence according to the proof of lemma [ 070501 ]
, the endpoint @xmath50 of @xmath441 is finite .
[ lem4 ] let @xmath46 be the bundle adapted metric with momentum profile @xmath514 in ( [ 70401 ] ) .
assuming that there is a constant @xmath50 such that @xmath316 is positive on @xmath155 and @xmath389 .
then @xmath46 defined on @xmath68 can be extended across @xmath512 if and only if @xmath515 .
first recall that @xmath518 $ ] is bi - holomorphic to the line bundle @xmath519 over @xmath240 .
let @xmath520 $ ] be the homogeneous coordinates of @xmath521 . here
the open set @xmath17 is @xmath522 and the hyperplane @xmath240 is @xmath523 .
hence @xmath524 is the @xmath525th coordinate of @xmath17 and @xmath526 $ ] is the homogeneous coordinates of @xmath240 . let @xmath527\ |\ v_\alpha\not=0\}$ ] and define @xmath528 with @xmath529 .
let @xmath530 be the coordinate of the trivialization of @xmath531 .
then its transition function defined on @xmath532 is @xmath533 the bi - holomorphic map @xmath534\to \mathcal o(1)$ ] is @xmath535\mapsto [ w^\alpha_1,\cdots , w^\alpha_{\beta-1},w^\alpha_{\beta+1},\cdots , w^\alpha_n,\frac{v_0}{v_\alpha}],\qquad \text{for}\ v_\alpha\not=0.\ ] ] define on @xmath519 the function @xmath536 we have @xmath537 on @xmath15 . by direct computation
, we have @xmath538 define the functions @xmath539 and @xmath540 since as @xmath541 , @xmath542 and @xmath543 , the function @xmath544 is continuous at @xmath545 . as to @xmath546 , we shall prove that if we take a suitable constant in ( [ ttau ] ) , then it is also continues at @xmath545 .
in fact @xmath547 is smooth . since @xmath389 , @xmath515 , and @xmath329 is rational , we can write @xmath548 for some smooth function @xmath549
. then @xmath550 here @xmath551 is a smooth function with @xmath552 .
hence @xmath553 and @xmath554 is a smooth function of @xmath320 .
moreover , by the implicit function theorem , we can solve ( [ hh ] ) to get a smooth function @xmath555 .
hence @xmath547 is a smooth function of @xmath556 .
now we can also see that @xmath557 is smooth at @xmath545 since @xmath558 however , condition ( ii ) is satisfied automatically if condition ( i ) holds . for one can show that under condition ( i ) @xmath49 is the unique solution of @xmath565 for @xmath566 .
in fact , we have the following result .
since @xmath568 has at most two roots and we have already @xmath569 , @xmath570 has at most one root @xmath571 in @xmath567 or in @xmath155 . also since @xmath515 and @xmath389
, @xmath333 is positive as @xmath542 .
hence if @xmath495 , there are two cases should be considered .
one is @xmath572 on @xmath573 , the other is @xmath574 on @xmath575 .
the former is impossible as @xmath576 which would lead to @xmath577 has at least three roots .
the latter is also impossible as from this one can derive @xmath578 which would also lead the contradiction .
thus , @xmath579 . hence ,
in the following we only need to find a constant @xmath50 such that condition ( i ) is satisfied
. we can solve constants @xmath70 and @xmath33 from @xmath389 and @xmath515 as : @xmath581 and @xmath582 with the definitions ( [ hhh ] ) and of @xmath583 so we should determine the range of @xmath70 such that @xmath316 satisfies ( i ) .
we need to prove @xmath426 . by lemma [ lem0 ]
, @xmath49 is a minimum of @xmath608 . from the taylor expansion of @xmath608 at @xmath609 , if @xmath431 , @xmath610 .
so if @xmath611 , it is impossible .
thus for @xmath75 ( hence @xmath77 ) , or @xmath76 and @xmath612 , the metrics are pmy . for the case @xmath76 and @xmath77 , if @xmath613 , then @xmath614 would have one root in @xmath615 when @xmath616 .
but it is impossible when @xmath77 .
hence @xmath617 and then @xmath426 .
the metrics are also pmy .
let @xmath410 , @xmath175 , @xmath411 and @xmath190 .
then @xmath618 we can solve @xmath346 to get @xmath619 for example , let @xmath620 .
then @xmath621 which implies @xmath622 we give a picture of @xmath316 as figure [ p5 ] .
let @xmath410 , @xmath175 , @xmath626 and @xmath627 .
we choose @xmath628 , then the graph as figure [ p6 ] shows @xmath629 and @xmath630 , or @xmath631 and @xmath632 .
we give a picture of @xmath316 with @xmath633 as figure [ p7 ] .
99 v. apostolov , d. calderbank , p. gauduchon and c. tnnesen - friedman ; _ extremal khler metrics on ruled manifolds and stability . _ astrisque * 322 * ( 2008 ) , 93150 . c. arezzo and f. pacard ; _ blowing up and desingularizing constant scalar curvature khler manifolds
. _ acta . math .
* 196 * ( 2006 ) , 179228 .
c. arezzo and f. pacard ; _ blowing up khler manifolds with constant scalar curvature .
ii . _ ann . of math . ( 2 )
* 170 * ( 2009 ) , 685738 . c. arezzo , f. pacard and m. singer , _ extremal metrics on blow ups . _
duke math .
j. * 157 * ( 2011 ) , 151 .
e. calabi ; _ mtriques , khleriennes et fibrs holomorphes . _
annales scientifiques de lcole normale suprieure * 12 * ( 1979 ) , 268294 .
e. calabi ; _ extremal khler metrics , in seminar on differential geometry .
_ ann . of math
. stud . * 102 * , 259290 , princeton univ . press , princeton , 1982 .
j. fu , s.t .
yau and w. zhou ; _ complete csck metrics on the local models of the conifold transition .
* 335 * ( 2015 ) , 12151233 .
j. fu , p. gao and w. zhou ; nonexistence for complete canonical metrics and the calabi
futaki invariant for khler metrics of poincar - mok - yau asymptotic property on complete khler manifolds .
a. d. hwang and m. a. singer ; _ a momentum construction for circle invariant khler metrics .
* 354 * ( 2002 ) 22852325 .
n. koiso and y. sakane ; _ non - homogeneous khler
einstein metrics on compact complex manifolds in curvature and topology of riemannian manifolds .
_ springer lnm * v. 1201 * , 1986 , 165179 . n. mok and s.t .
yau ; _ completeness of the khler
einstein metric on bound domains and the characterization of domain of holomorphy by curvature condition .
_ proceddings of symposia in pure mathematics , * v.39 * ( 1983 ) , part 1 , 4159 .
h. pedersen and y.s .
poon ; _ hamiltonian constructions of khler einstein metrics and khler merics of constant scalar curvature .
phys . * 136 * ( 1991 ) , 309326 .
s. simanca ; _ khler metrics of constant scalar curvature on bundles over @xmath240 .
ann . * 291 * ( 1991 ) , 239 - 246 .
g. szkelyhidi ; _ blowing up extremal khler manifolds .
_ duke math .
j. * 161 * ( 2012 ) , 14111453 .
g. szkelyhidi ; _ blowing up extremal khler manifolds ii .
_ invent . math . * 200 * ( 2015 ) , 925977 g. tian and s.t .
yau ; _ complete khler manifolds with zero ricci curvature .
soc . * 3 * ( 1990 ) , 579609 .
g. tian and s.t .
yau ; _ complete khler manifolds with zero ricci curvature .
invent . math . * 106 * ( 1991 ) , 2760 .
yau ; _ on the ricci curvature of a compact khler manifold and the complex monge ampre equation .
i. _ comm .
appl . math .
* 31 * ( 1978 ) , 339411 . s.t .
yau , _ a general schwarz lemma for khler manifolds .
_ amer . j. math * 100 * ( 1978 ) , 197203 . | let @xmath0 be a compact khler manifold and @xmath1 a subvariety of @xmath0 with higher co - dimension .
the aim is to study complete constant scalar curvature khler metrics on non - compact khler manifold @xmath2 with poincar
mok
yau asymptotic property ( see definition [ def ] ) . in this paper ,
the methods of calabi s ansatz and the moment construction are used to provide some special examples of such metrics . |
in a number of correlated electron systems , ordering of charge degrees of freedom of electrons , or charge ordering ( co ) , is a key feature to understanding their physical properties because it is intimately related to the electronic transport and magnetic phenomena .
@xcite the concept of co was originally introduced to interpret an array order of fe@xmath0 and fe@xmath1 ions in magnetite fe@xmath2o@xmath3 , which is known as the verwey transition .
@xcite for several years , the importance of co has been increasingly recognized in conjunction with spatial localization of charge carriers observed in manganese oxide la@xmath4ca@xmath5mno@xmath2 ( @xmath6 ) , @xcite copper oxide la@xmath7nd@xmath8sr@xmath5cuo@xmath3 ( @xmath9 ) , @xcite etc .
especially , in the field of low - dimensional organic materials , the co itself has been one of the main issues for discussion .
@xcite quite recently , peculiar charge fluctuations have been found in the co state of quasi - two - dimensional ( 2d ) organic conductors @xmath10-(bedt - ttf)@xmath11x ( hereafter , bedt - ttf is abbreviated as et ) and they have attracted much attention due to their intriguing relation to the superconducting ( sc ) state .
the crystal structure of @xmath10-(et)@xmath11x are built with an alternating stack of et conducting and anion ( x ) insulating layers . in the conducting layer ,
the et molecules form an anisotropic triangular - lattice structure ( see fig .
[ lattice ] ) and the band is @xmath12-filled in terms of electrons or @xmath13-filled in terms of holes . in experiments , the transfer integrals ( @xmath14 ) are controllable by altering a dihedral angle ( @xmath10 ) with the substitution of anion and/or adjustment of pressure .
@xcite a decrease in the dihedral angle corresponds to an increase in @xmath15 and therefore a decrease in @xmath16 . at low temperatures ( @xmath17 k ) ,
the compounds exhibit a variety of phases , such as co insulator , paramagnetic metal , and sc , as a function of the dihedral angle .
overall , the metal - insulator transition temperature ( @xmath18 ) decreases with decreasing the dihedral angle . only @xmath10-(et)@xmath11i@xmath2 that has the smallest dihedral angle @xmath19 in the @xmath10-et family shows sc in place of co below @xmath20 k. @xcite thus far , the co pattern in each compound has been investigated by x - ray diffraction and @xmath21c - nmr measurements . in the series of x@xmath22(scn)@xmath3 with @xmath23co and zn , @xmath24tl and rb , a stripe - type co which has a twofold periodicity along the @xmath25 axis
is observed below @xmath18 .
@xcite the compounds with @xmath24tl and rb is known to have an average dihedral angle of @xmath26 to @xmath27 . using the estimated ratio of the transfer integrals @xmath28 , the stripe - type co was successfully reproduced by the mean - field theory .
@xcite but then , the x - ray diffuse scattering experiments @xcite demonstrated the presence of peculiar charge fluctuations in @xmath10-(et)cs@xmath29(scn)@xmath3 with @xmath24co and zn ; two diffuse peaks with the wave numbers @xmath30 and @xmath31 were found .
the wave number @xmath32 is the same as that of the stripe - type co observed in the compounds with @xmath24tl and rb , but the other wave number @xmath33 corresponds to a three - fold periodicity .
the compounds with @xmath24cs have relatively small dihedral angle @xmath34 and are located in the vicinity of the quantum critical point @xmath35 k. accordingly , the ratio of the transfer integrals is expected to be rather small , i.e. , @xmath36 , which is different from the case of @xmath24tl and rb .
thus , a very absorbing subject , i.e. , the coexistence of charge fluctuations with different wave numbers , has been provided .
the appearance of the spatially inhomogeneous co has also been suggested by the dielectric response , ac resistivity , @xcite @xmath37h - nmr , epr , and static magnetic susceptibility @xcite measurements .
furthermore , a power - law behavior of the current - voltage characteristics over a wide range of currents @xcite and a current - induced melting of insulating co domains @xcite have been observed .
in addition , the optical conductivity spectra show a transition from short - ranged to long - range co around @xmath38 , and stronger insulating features appear in @xmath39cszn than in @xmath39rbzn .
@xcite a number of theoretical studies have then been carried out in order to clarify the nature of the co and charge fluctuations .
@xcite motivated by such developments in the field , we consider in this paper the extended hubbard model at quarter filling defined on the anisotropic triangular lattices .
we employ the density - matrix renormalization group ( dmrg ) and lanczos exact - diagonalization methods to investigate the electronic states of the model .
first , we calculate the hole density and double occupancy to determine the ground - state phase diagram .
next , the dynamical density - density correlation functions are calculated to study the low - energy charge excitations .
we also calculate the single - particle excitation spectrum to elaborate the anomalous metallic states seen in the low - energy charge excitations of the three - fold co phase .
we finally obtain the optical conductivity for each phase and clarify the behavior of the charge degrees of freedom of the model by focussing on the three - fold charge fluctuations .
this paper is organized as follows . in sec .
ii , the 2d extended hubbard model on the anisotropic triangular lattices is introduced .
we also define some physical quantities and explain the applied methods for the calculations . in sec .
iii , we present the ground - state phase diagram of the extended hubbard model and discuss the physical properties related to the co in each phase .
section iv contains conclusions and discussions including comparison with the experimental results .
the hamiltonian of the extended hubbard model defined on the anisotropic triangular lattice is given by @xmath40 where @xmath41 ( @xmath42 ) is the creation ( annihilation ) operator of a hole with spin @xmath43 at site @xmath44 , @xmath45 is the number operator , and @xmath46 .
the sum @xmath47 runs over nearest - neighbor pairs .
@xmath48 and @xmath49 are hopping integral and intersite coulomb repulsion between sites @xmath44 and @xmath50 ; here we retain two kinds of nearest - neighbor hopping integrals ( repulsions ) @xmath15 ( @xmath51 ) and @xmath52 ( @xmath53 ) as shown in fig . [ lattice ] .
@xmath54 is the onsite coulomb repulsion .
we restrict ourselves to the case at quarter filling ; i.e. , @xmath55 where @xmath56 denotes the ground - state expectation value . in this paper
, we consider the case of @xmath57 and @xmath58 as a typical set of parameter for x = csco(scn)@xmath3 .
@xcite hereafter , we take @xmath59 as the unit of energy .
we employ two kinds of numerical methods , i.e. , the dmrg and lanczos methods .
either of these methods is chosen for the calculation of each physical quantity .
the details are explained in the following .
we are interested in the co phenomena in the subject materials , so that it is useful , first of all , to make the ground - state phase diagram associated with the charge degrees of freedom . to this end , we calculate the hole density @xmath60 and double occupancy @xmath61 ( @xmath62 ) for all sites @xmath44 to investigate the charge distribution .
a large size system is required for reproducing all possible co patterns and for an accurate estimation of these quantities .
we thus study a finite - size cluster of @xmath63 and @xmath64 using the dmrg method .
the periodic boundary conditions ( pbc ) are applied for the @xmath25-direction , whereas the open - end boundary conditions ( obc ) are applied for the @xmath65-direction .
this choice of the boundary conditions enables us to detect the co state clearly ( see sec.iii a ) .
we keep up to @xmath66 density - matrix eigenstates in the dmrg procedure and thus the maximum truncation error , i.e. , the discarded weight , is less than @xmath67 . in this way
, the maximum error in the ground - state energy is estimated to be @xmath68 for the most inaccurate case @xmath69 .
based on the results of the hole density and double occupancy , we determine the ground - state phase diagram on the parameter space @xmath70 .
next , we consider the dynamical density - density correlation function defined as @xmath71 in order to study the low - energy charge excitations . here ,
@xmath72 and @xmath73 are the @xmath74-th eingenstate and eigenenergy of the system with @xmath75 holes ( @xmath76 corresponds to the ground state ) .
since the exact definition of the momentum - dependent operators with the obc is quite difficult , we choose the pbc for a quantitative estimation of the spectrum ; we therefore use the pbc for both the @xmath77- and @xmath78-directions ( or the @xmath65- and @xmath25-directions ) . with this pbc ,
the density operators @xmath79 can be precisely defined by @xmath80 where @xmath81 is the number of lattice sites @xmath82 ( or @xmath83 ) and @xmath84 is the position of site @xmath44 . however , in this case , it is quite difficult to carry out sufficiently accurate calculations with the dmrg method , so that we here use the lanczos method on small clusters with 16 ( @xmath85 , @xmath86 ) and 18 ( @xmath87 , @xmath88 ) sites .
consequently , the system size is restricted but the result is numerically exact . of interest
are also the evolution of the fundamental excitations in the systems with strongly frustrated correlations . to see this
, we calculate the single - particle excitation spectrum , which is obtained as @xmath89 with the photoemission ( pes ) spectrum @xmath90 and inverse photoemission ( ipes ) spectrum @xmath91 where @xmath92 ( @xmath93 ) is the fourier transform of the creation ( annihilation ) operator @xmath94 ( @xmath95 ) . for the same reason as the calculation of the dynamical density - density correlation function , we employ the lanczos method with the pbc where the operator is defined by @xmath96 the calculation
is carried out with a 18-site ( @xmath87 , @xmath88 ) cluster .
finally , for more elaborate study on the behavior of the charge degrees of freedom and direct comparison with experiments , we calculate the optical conductivity defined by @xmath97 with a component of the dipole operator @xmath98 where @xmath99 is a unit vector parallel to the @xmath100-direction .
it is known that the obc is more feasible for calculation of the optical conductivity spectra because finite - size effects have much smaller influence on the results in comparison with a case using the pbc .
@xcite hence , we apply the obc for both the @xmath65- and @xmath25-directions and study a finite - size cluster @xmath63 , @xmath64 with the dynamical dmrg ( ddmrg ) method .
the ddmrg method is an extension of the standard dmrg method for the calculation of dynamical properties . in the ddmrg calculation
, a required cpu time increases rapidly with the number of the density - matrix eigenstates @xmath101 , so that we try to keep it as few as possible .
because the dmrg approach is based on a variational principle , we have to prepare a ` good trial function ' of the ground state with the density - matrix eigenstates as much as possible .
we therefore keep @xmath102 to obtain the true ground state in the first five ddmrg sweeps and keep @xmath103 to calculate the spectrum of the system . as a result , the maximum truncation error ,
i.e. , the discarded weight , is about @xmath104 , while the maximum error in the ground - state is about @xmath105 .
so far , it has been suggested that the system defined by eq.([hamiltonian ] ) has three types of co phases , depending on the values of the intersite coulomb repulsions .
the co patterns for each phase are shown in the inset of fig .
[ n_d ] : ( a ) vertical stripe for @xmath106 , ( b ) diagonal stripe for @xmath107 , and ( c ) three - fold alignment for @xmath108 .
other likely patterns are also shown : ( d ) horizontal and ( e ) randomly aligned stripes , which are not realized in the ground state .
a random alignment co pattern is formed with any combination of alternately - occupied chains parallel to the @xmath25-axis .
hence , the diagonal co pattern is a special case of the random alignment ones .
first of all , we note that the co is observed as a state with a broken translational symmetry in our calculations : actually , there are two or more degenerate ground states and one of the degenerate states is picked out as the ground state by an initial condition of the dmrg calculation as we here apply the obc along the @xmath77-direction .
thus , a co state includes more than two kinds of sites with different hole densities , i.e. , hole - rich and hole - poor sites . in fig .
[ n_d ] , we show dmrg results of the hole density @xmath109 and double occupancy @xmath110 at the hole - rich sites @xmath111 for ( a ) @xmath112 , ( b ) @xmath113 , and ( c ) @xmath114 .
when both @xmath51 and @xmath53 are small , the analysis shows that the hole density is uniform over the system , i.e. , @xmath115 for all sites @xmath44 , and the system is in the metallic state .
as @xmath51 and/or @xmath53 increase , the charge fluctuations toward the co states are enhanced and we find that @xmath116 starts to increase at the co phase boundary . in the co phase ,
the charge fluctuations are weakened again as the co state is stabilized , where @xmath116 approaches a value in the atomic limit @xmath117 .
let us now evaluate the phase boundaries between the metallic and co phases .
when we increase @xmath51 from zero with fixing @xmath112 [ see fig .
[ n_d](a ) ] , @xmath116 begins to increase around @xmath118 ; accordingly , a critical strength of the vertical co state is obtained as @xmath119 . at @xmath120 ,
@xmath116 increases rapidly to unity but the double occupancy remains almost zero .
this means that ( nearly ) half - filled chains are alternated with ( nearly ) empty chains and the charge fluctuations are totally suppressed in the entire region of vertical co phase . in analogy with the case of the vertical co state , we can estimate a critical strength of the diagonal co state as @xmath121 [ see fig .
[ n_d](b ) ] .
we should however note that @xmath122 seems to approach unity more slowly than that in the vertical co phase .
this implies that the charge fluctuations in the diagonal co state are rather stronger than those in the vertical co state .
the reason of this strong charge fluctuation in the diagonal co phase is the existence of nearly energetically degenerate states , i.e. , the horizontal and an infinite number of the randomly aligned co patterns , which are shown in fig .
[ n_d](d ) and ( e ) , respectively .
this situation has also been confirmed by the mean - field approximation , @xcite variational monte carlo ( vmc ) method , @xcite and strong - coupling study of the spinless model .
@xcite the energies of those co patterns including the diagonal one are exactly equal to each other in the atomic limit @xmath123 . as for the three - fold co phase ,
the situation is somewhat different from the other co phases .
if a relation @xmath124 is kept , the three - fold co instability begins to appear around a critical point @xmath125 [ see fig .
[ n_d](c ) ] . above the critical point
, we find the rapid increase of not only the hole density but also the double occupancy at the hole - rich sites .
this implies that the competition between the effects of @xmath54 and @xmath126 s is essential for the presence of the three - fold co phase .
we also note that the three - fold co state is metallic in the sense that the drude weight is nonzero .
@xcite the presence of metallic co phase has previously been suggested for the extended square - lattice hubbard models ; simply , the pocket - like small fermi surface appears by doping the long - range co phases with holes or electrons without destroying the long - range co. @xcite note that the ground state has both the onsite hole - pairing and co , which may be essentially the same as the coexistence of the _ s_-wave superconducting and co states in the two - dimensional negative-@xmath54 hubbard model .
@xcite in fig .
[ pd ] , we determine the ground - state phase diagram from the results of the hole density , where the phase boundaries are drawn with dashed curves
. we can make it certain that the three substantial co phases are distinguished from a metallic phase with uniform charge distribution .
this phase diagram is basically consistent with that from the vmc calculation .
@xcite the main difference is that a metallic region is sandwiched between the vertical and three - fold co phases in our results . on the other hand ,
the diagonal and three - fold co phases are contiguous , in agreement with the vmc result .
the appearance of the metallic region in - between may be explained as follows .
the diagonal and three - fold co phases may be contiguous because the transition can be derived by the flow of charges only via the nearest - neighbor hopping integrals @xmath15 ( see fig .
[ flow ] ) , whereas a drastic charge redistribution is required in the transition from the vertical co pattern to the three - fold co pattern and therefore the ground state needs to go through the metallic one .
we may also suggest that this ` charge - flowing ' transition between the diagonal and three - fold co phases occurs in the presence of the three - fold ( diagonal ) charge fluctuations in the diagonal ( three - fold ) co phase as is evident in the calculated dynamical density - density correlation functions ( see the following subsection ) .
we also note that the three - fold co phase is shrunk with increasing @xmath54 .
the dynamical density - density correlation functions calculated by the lanczos method for several sets of @xmath51 and @xmath53 values are shown in fig .
each parameter set corresponds to a different phase as follows : @xmath127 to the diagonal co phase , @xmath128 to the vertical co phase , @xmath129 to the three - fold co phase , and @xmath130 to the uniform metallic phase .
in addition , we plot the results for @xmath131 , where the system is in the non - co metallic phase but with strong three - fold co fluctuations .
when @xmath132 , the model ( [ hamiltonian ] ) is equivalent to the square - lattice hubbard model at quarter filling and no particular modes of charge fluctuations are developed ; we therefore find the broad spectral features for all the momenta , which come basically from the particle - hole transition in the corresponding noninteracting system . with increasing the intersite interactions , however , the broad features change to the low - energy sharp peaks , indicating that the excitations turn to be collective - mode like .
generally , the enhancement of low - lying peak at a particular momentum must be expected for the co instability .
when the system is unstable to the vertical [ diagonal ] co phase , the particular momentum is located at @xmath133 [ @xmath134 . looking at the spectra for @xmath135 , a sharp peak appears around @xmath136 at @xmath133 and there are nearly no peaks at the other momenta .
this means that the system is unstable to the vertical co state with ( almost ) complete charge disproportionation .
this result is consistent with the results of the hole density given in sec .
iii a. on the other hand , in the spectra at @xmath127 , we see not only a low - energy peak at @xmath137 but also relatively high - energy peaks at the other momenta ; most of the high - energy peaks concentrate at momenta with @xmath138 and few peaks appear at momenta with @xmath139 . by the axes rotation , we find the charge fluctuations mostly occur along the @xmath25-axis , which are associated with the nearly degenerate states , i.e. , horizontal and randomly aligned co phases .
it is moreover striking that the intensity around @xmath140 is significantly large , which means that the three - fold charge fluctuations are rather strong although the system is in the diagonal co phase .
let us now discuss the spectral features for the three - fold co metallic state , of which the particular momentum is located at @xmath140 .
the charge excitation spectra in the co metallic state are currently not well understood , so that a good opportunity for studying them may be offered here .
we thus investigate a process of gradual change in the spectra with increasing the intersite coulomb interactions @xmath126 s .
since the three - fold co phase lies around @xmath141 , we increase @xmath51 ( @xmath53 ) from @xmath142 to @xmath143 with keeping a condition @xmath114 .
the spectra for @xmath132 have the broad spectral features for all the momenta as mentioned above and these features basically remain unchanged as far as @xmath51 and @xmath53 are less than @xmath144 . only around @xmath131
, we can recognize the changes clearly in the calculated spectra : the low - energy spectrum at @xmath140 is particularly enhanced as anticipated , but those at @xmath145 and @xmath133 are also enlarged , while the spectral weights for the other momenta are strongly suppressed .
we understand that these changes suggest that the three types of co instability are simultaneously developed and that , considering that the energies of the lowest - lying excitations for the three momenta are nearly equal , the charge fluctuations for the three co patterns are strongly competing . after increasing the intersite coulomb interactions to @xmath146 ,
even greater changes are seen in the spectral features . at @xmath140 , we find the enhancement of a sharp peak around @xmath136 , which indicates the strong charge fluctuations of the three - fold periodicity and is consistent with the results of the hole density .
it is particularly worth noting that the low - energy spectral intensities at @xmath137 is continuously enhanced [ and those at @xmath147 are suppressed ] .
let us discuss some experimental situations here .
we may point out first that the wave vectors @xmath32 and @xmath33 of the two diffuse peaks observed experimentally @xcite in the co phase of @xmath10-(et)@xmath148(scn)@xmath3 [ @xmath24cs , @xmath24co or zn ] should be equivalent to our momenta @xmath137 and @xmath149 , respectively .
we may therefore suggest that the appearance of the two enhanced peaks in our calculations correspond to the coexistence of the two different charge fluctuations .
note that these enhanced peaks are already obtained for more realistic parameter values @xmath131 in our calculations .
it is also striking that the low - energy peaks appear in the entire brillouin zone and the charge fluctuations are induced in all the momentum transfers .
this situation is quite different from those for the diagonal and vertical co stripes where the low - energy peak appears only at the particular momenta . in the diagonal and vertical co phases
, we just find the insulating band structures reflecting the co states in our calculations ; namely , the pes and ipes peaks are separated by the charge gap and the fermi level lies between the two spectra .
we thus focus on the spectral features in the three - fold co phases here . in fig .
[ aqwspec ] , we show the single - particle excitation spectra calculated with the lanczos method for @xmath132 , @xmath150 , and @xmath129 . at @xmath132 , we can clearly see the spectral weight separated into two bands due to the onsite coulomb interaction , i.e. , lower hubbard band ( lhb ) and upper hubbard band ( uhb ) with the distance @xmath151 ( @xmath152 ) .
the lhb , which corresponds to the conduction band , can be approximately expressed by the dispersion of noninteracting holes , @xmath153 on the other hand , the uhb has relatively small spectral weight and narrow band width , which are general features in the strong - coupling regime . at @xmath131 ,
the shape of conduction band remains almost unchanged from that at @xmath132 .
however , we can see that an overlap between the lhb and uhb occurs due to the competition between the onsite and nearest - neighbor coulomb interactions . as a result
, the band structure is seen to be reduced to a single dispersion line .
the interaction between holes is effectively weakened ( or vanished ) by the frustration of the coulomb interactions , so that the dispersion relation looks as if it were for the single - orbital model or the noninteracting case .
surprisingly , the spectral features are drastically changed at @xmath146 .
the spectral weight appears to be separated to form several bands again .
it is particularly worth noting that the width of conduction band is much narrower than that at @xmath131 . in the three - fold co phase
, the holes can conduct only through the hole - rich sites but the hole - rich sites are not connected by direct hopping integrals ; in other words , the carriers form a small fermi surface .
@xcite thus , the conduction band is centered at @xmath154 and the band width should be very narrow .
it may be related to the appearance of low - energy peaks in the entire brillouin zone observed in the dynamical density - density correlation function as discussed above .
let us then consider the peaks in the higher - energy range , @xmath155 .
these peaks are essentially formed in connection with the empty ( hole - poor ) sites .
an empty site is surrounded by three hole - rich sites and a hole - rich site is occupied by one or two holes . when a hole is added on an empty site ,
the excitation energy is @xmath156 if the surrounding three hole - rich sites are singly occupied and it is @xmath157 if the surrounding three hole - rich sites are doubly occupied .
this is the reason why there are continuum weights from @xmath158 to @xmath159 in the spectra for all the momenta .
in addition , we can find the dispersion with bandwidth @xmath160 around @xmath161 .
this dispersion is formed from a empty zigzag chain along the @xmath65-axis , which is nearly separated from each other by the hole - rich sites .
the optical conductivity spectra calculated by the dmrg method for several sets of @xmath51 and @xmath53 are shown in fig .
[ opt ] . for all cases
, the spectral intensity for the @xmath65-direction seems to be a few times larger than that for the @xmath25-direction , which just reflects the difference between the interatomic distances @xmath162 in the @xmath65 and @xmath25 directions .
this leads to the intensity ratio @xmath163 .
the most easily comprehensible result should be the ones in the vertical co phase , i.e. , @xmath164 [ see fig . [ opt](b ) ] . the shapes of spectra for the @xmath65- and @xmath25-directions are essentially the same , although the intensities are different as mentioned above .
for both of the spectra , the lowest excitation appears around @xmath165 , which corresponds to the optical ( insulating ) gap , and the other excitations lie in much higher frequencies , @xmath166 . also , the total weight of the spectrum is smaller than that in the other co phases .
these results are consistent with the fact that the vertical co state is rather solid and the charge fluctuations are quite small .
the optical gap in the vertical co phase may be estimated as @xmath167 with the bare bandwidth @xmath168 .
let us then turn to the diagonal co phase , i.e. , @xmath169 [ see fig .
[ opt](a ) ] .
we find that the result for the @xmath65-direction is apparently different from that for the @xmath25-direction .
the spectrum for the @xmath65-direction consists of the first - excitation peak at @xmath170 and high - energy broad features , which is similar to that in the vertical co phase .
this result reflects the solid co and weak charge fluctuations along the @xmath65-direction , as shown in the dynamical density - density correlation functions . on the other hand , most of the spectral features for the @xmath25-direction appear in the lower - energy range .
the optical gap is of the order of @xmath171 , which is much smaller than that of the vertical co phase .
this is so because the diagonal co state can be rearranged to the randomly aligned ( or horizontal ) co state with small excitation energy .
thus , the charge fluctuations along the @xmath25-axis are much stronger than those along the @xmath65-axis . since the energy difference between the diagonal co phase and random alignment co state
is hardly changed even if @xmath53 increases , the low - energy structure is expected to be seen for any @xmath53 value .
lastly , we discuss the spectral aspects in the metallic regime around the three - fold co instability [ see fig .
[ opt](c)-(e ) ] . with the obc ,
the optical conductivity at @xmath136 is ruled out and the lowest - energy scale is limited to an order of @xmath172 .
@xcite therefore , the drude spectral weight appears around the minimum excitation energy . at @xmath173 ,
the spectra for the @xmath65- and @xmath25-directions are quite similar . for each spectrum ,
small peak structure around @xmath174 should correspond to the drude peaks .
the spectral weight exists continuously from the ` drude frequency ' to higher excitations and reaches the maximum around @xmath175 , and very little spectral weight exists at higher - energy range , @xmath176 . at @xmath131 , we can find some tendencies to the three - fold co state ; in the spectrum along the @xmath65-axis the intensity around @xmath175 is about half of that at @xmath173 and the high - energy broad features are relatively large .
a large peak around @xmath175 at @xmath173 splits into two peaks that indicate the appearance of individual excitations associated with the cos .
the spectrum along the @xmath65-axis remains almost unchanged , except the drude - like peak which becomes much smaller . at @xmath177
, we can clearly see the characteristic features of the three - fold co state : in the spectrum along the @xmath65-axis the weight is separated into the low - energy drude - like peak at @xmath174 and high - energy broad features .
we can understand that the former comes from the small fermi surface in the three - fold co phase and the latter comes from the excitations due to the three - fold charge fluctuations . on the other hand ,
the weight of the drude - like peak is nearly zero in the spectrum along the @xmath25-axis .
this means that the co is more solid along the @xmath25-axis than along the @xmath65-axis .
we can explain this as follows : 3/2-filled chains are alternated with zigzag empty chains along the @xmath25-axis [ see the inset of fig .
[ n_d ] ( c ) ] , whereas each chain along the @xmath25-axis is quarter filled , which means that the holes distribute uniformly along the @xmath65-axis . as mentioned above , the optical conductivity along the @xmath65- ( @xmath25- ) axis is concentrated in the lower - energy range when the three - fold ( diagonal ) charge fluctuations are enhanced .
we suggest that these features in the optical conductivity spectra should give a good criterion to examine the co patterns experimentally .
the extended hubbard model at quarter filling defined on the anisotropic triangular lattice have been studied by the lanczos exact - diagonalization and dmrg methods .
we determine the ground - state phase diagram based on the results of the hole density and double occupancy . in the phase diagram , there exist three substantial co phases ( diagonal , vertical , and three - fold ) and a metallic phase with uniform charge distribution .
we find that the charge fluctuations ( or instability to co ) with the three - fold periodicity appear for realistic parameters @xmath178 .
we suggest that the transition between the diagonal and three - fold co phases is derived by the flow of charges only via the nearest - neighbor hopping integrals , which may lead to the coexistence of the diagonal and three - fold charge fluctuations near the phase boundary .
the coexistence features are also found in the density - density correlation functions ; in the diagonal ( three - fold ) phase , not only the particular peak but also the low - energy peaks corresponding to the three - fold ( diagonal ) charge fluctuations are enhanced . in the vertical co phase ,
only the low - energy peak at the particular momentum is visible and there is no charge fluctuations associated with the other co states .
moreover , we find that there appear the low - energy collective - modes - like excitations in the entire brillouin zone when the three - fold charge fluctuations are very strong .
the wave numbers of the diagonal and three - fold charge fluctuations are equivalent to those of the two x - ray diffuse peaks @xmath179 and @xmath31 measured in @xmath10-(et)cs@xmath29(scn)@xmath3 with @xmath24co or zn .
if we could make the dihedral angle larger , the value of @xmath180 decreases and only the diagonal co fluctuation remains .
this is consistent with the fact that only a diffuse peak with wave number @xmath32 is observed in the compounds with @xmath24tl and rb which have a larger dihedral angle than those with @xmath24co or zn do .
we also study the single - particle excitation spectrum to see the evolution of the fundamental low - lying excitations in the presence of strongly frustrated correlations . when the onsite and intersite coulomb interactions compete ( @xmath181 ) , the interaction between holes is effectively diminished and the dispersion relation looks as if it is of the noninteracting case .
when the two intersite coulomb interactions compete ( @xmath182 ) , the carriers form very narrow conduction band with a small fermi surface . furthermore
, we examine the optical conductivity which reflects the characteristic features for each co phase . in the vertical co phase , the spectra for both the directions explicitly represent the insulating features and the optical gaps are large , estimated as @xmath183 . in the diagonal co phase , the spectrum for the @xmath65-direction is essentially the same as that in the vertical co phase but that for the @xmath25-direction is located in the lower - energy range .
this is so because the diagonal co state can be rearranged to the randomly aligned ( or horizontal ) co state with small excitation energy . in the three - fold co phase ,
the spectra indicate the presence of the separated low - energy drude - like peak and high - energy broad features , leading to the anomalous metallic states in the system .
the optical conductivity along the @xmath65- ( @xmath25- ) axis is concentrated in the lower - energy range when the three - fold ( diagonal ) charge fluctuations .
these spectral features may give a good criterion to examine the co patterns experimentally .
finally , we make a comment on the difference in the experimental spectra between @xmath39cszn and @xmath39rbzn . because we find that an increase in @xmath180 makes the diagonal co state unstable , it is the smallness of @xmath52 that is essential for the strong insulating features in the spectra of @xmath39cszn .
we hope that our studies will help us understand the mechanism of charge fluctuations in the extended hubbard model defined on the anisotropic triangular lattices and hence will offer useful suggestions to some aspects of the three - fold charge fluctuations observed in @xmath10-(et)@xmath11x .
this work was supported in part by grants - in - aid for scientific research ( nos .
18540338 , 18028008 , 18043006 , and 19014004 ) from the ministry of education , science , sports , and culture of japan .
a part of computations was carried out at the research center for computational science , okazaki research facilities , and the institute for solid state physics , university of tokyo .
99 present address : leibniz - institut fr festkrper- und werkstoffforschung dresden , p.o .
box 270116 , d-01171 dresden , germany y. tokura and n. nagaosa , science * 288 * , 462 ( 2000 ) .
e. j. w. verwey and p. w. haayman , and f. c. romeijn , j. chem . phys . *
15 * , 181 ( 1947 ) .
s. mori , c. h. chen , and s .- w .
cheong , nature * 392 * , 473 ( 1998 ) .
j. m. tranquada , b. j. sternlieb , j. d. axe , y. nakamura , and s. uchida , nature * 375 * , 561 ( 1995 ) . m. lang and j. mller , _ the physics of superconductors - vol.ii_ , edited by k .- h .
bennemann and j. b. ketterson ( springer , berlin , 2003 ) , p.453 .
h. mori , s. tanaka , and t. mori , , 12023 ( 1998 ) ; h. mori , t. okano , s. tanaka , m. tamura , y. nishio , k. kajita , and t. mori , j. phys .
. jpn . * 69 * , 1751 ( 2000 ) .
h. kobayashi , r. kato , a. kobayashi , y. nishio , k. kajita , and w. sasaki , chem .
1986 * , 789 ( 1986 ) ; * 1986 * 833 ( 1986 ) .
k. miyagawa , a. kawamoto , and k. kanoda , , r7679 ( 2000 ) .
r. chiba , h. yamamoto , k. hiraki , t. takahashi and t. nakamura , j. phys .
solids * 62 * , 389 ( 2001 ) .
m. watanabe , y. noda , y. nogami , and h. mori , synth . met . *
135 - 136 * , 665 ( 2003 ) .
m. watanabe , y. noda , y. nogami , and h. mori , j. phys .
. jpn . * 73 * , 116 ( 2004);*74 * , 2011 ( 2005 ) .
h. seo , j. phys .
* 69 * , 805 ( 2000 ) .
y. nogami , j .-
pouget , m. watanabe , k. oshima , h. mori , s. tanaka and t. mori , synth . met . *
103 * 1911 ( 1999 ) .
m. watanabe , y. nogami , k. oshima , h. mori , and s. tanaka , j. phys .
68 * , 2654 ( 1999 ) .
k. inagaki , i. terasaki , h. mori , and t. mori , j. phys .
* 73 * , 3364 ( 2004 ) .
t. nakamura , w. minagawa , r. kinami , and t. takahashi , j. phys .
. jpn . * 69 * , 504 ( 2000 ) .
r. chiba , k. hiraki , t. takahashi , h. m. yamamoto , and t. nakamura , synth . met . * 133 - 134 * , 305 ( 2003 ) .
f. sawano , i. terasaki , h. mori , t. mori , m. watanabe , n. ikeda , y. nogami , and y. noda , nature * 437 * , 522 ( 2005 ) .
y. takahide , t. konoike , k. enomoto , m. nishimura , t. terashima , s. uji , and h. m. yamamoto , , 136602 ( 2006 )
. n. l. wang , t. feng , z. j. chen , and h. mori , synth . met .
* 135 - 136 * , 701 ( 2003 ) .
k. suzuki , k. yamamoto , k. yakushi , and a. kawamoto , j. phys .
. jpn . * 74 * , 2631 ( 2005 ) .
m. kaneko and m. ogata , j. phys .
* 75 * , 014710 ( 2006 ) .
h. watanabe and m. ogata , j. phys .
. jpn . * 75 * , 063702 ( 2006 ) . c. hotta and n. furukawa , 193107 ( 2006 ) ; c. hotta , n. furukawa , a. nakagawa , and k. kubo , j. phys .
. jpn . * 75 * , 123704 ( 2006 ) .
k. kuroki , j. phys .
. jpn . * 75 * , 114716 ( 2006 ) .
h. seo , k. tsutsui , m. ogata , and j. merino , j. phys .
* 75 * , 114707 ( 2006 ) .
m. udagawa and y. motome , , 206405 ( 2007 ) .
m. shingai , s. nishimoto , and y. ohta , unpublished .
y. ohta , k. tsutsui , w. koshibae , and s. maekawa , , 13594 ( 1994 ) .
r. t. scalettar , e. y. loh , j. e. gubernatis , a. moreo , s. r. white , d.j .
scalapino , r. l. sugar , and e. dagotto , , 1407 ( 1989 ) ; a. moreo and d. j. scalapino , , 946 ( 1991 ) .
a. s. moskvin , j. mlek , m. knupfer , r. neudert , j. fink , r. hayn , s .- l .
drechsler , n. motoyama , h. eisaki , and s. uchida , , 037001 ( 2003 ) .
r. m. fye , m. j. martins , d. j. scalapino , j. wagner , and w. hanke , , 6909 ( 1991 ) | using the lanczos exact - diagonalization and density - matrix renormalization group methods , we study the extended hubbard model at quarter filling defined on the anisotropic triangular lattice .
we focus on charge ordering ( co ) phenomena induced by onsite and intersite coulomb interactions .
we determine the ground - state phase diagram including three co phases , i.e. , diagonal , vertical , and three - fold co phases , based on the calculated results of the hole density and double occupancy .
we also calculate the dynamical density - density correlation functions and find possible coexistence of the diagonal and three - fold charge fluctuations in a certain parameter region where the onsite and intersite interactions compete .
furthermore , the characteristic features of the optical conductivity for each co phase are discussed . |
this paper studies learning problems of the following form . consider a finite , but potentially very large , collection of binary - valued functions @xmath2 defined on a domain @xmath3 . in this paper , @xmath2 will be called the _ hypothesis space _ and @xmath3 will be called the _ query space_. each @xmath4 is a mapping from @xmath3 to @xmath5 . throughout the paper
we will let @xmath0 denote the cardinality of @xmath2 .
assume that the functions in @xmath2 are unique and that one function , @xmath6 , produces the correct binary labeling .
it is assumed that @xmath7 is fixed but unknown , and the goal is to determine @xmath7 through as few queries from @xmath3 as possible . for each query @xmath8 , the value @xmath9 , possibly corrupted with independently distributed binary noise , is observed .
the goal is to strategically select queries in a sequential fashion in order to identify @xmath7 as quickly as possible .
conditions are established under which gbs ( and a noise - tolerant variant ) have a near - optimal query complexity .
the main contributions of this paper are two - fold .
first , incoherence and geometric relations between the pair @xmath10 are studied to bound the number of queries required by gbs .
this leads to an easily verifiable sufficient condition that guarantees that gbs terminates with the correct hypothesis after no more than a constant times @xmath11 queries .
second , noise - tolerant versions of gbs are proposed .
the following noise model is considered .
the binary response @xmath12 to a query @xmath8 is an independent realization of the random variable @xmath13 satisfying @xmath14 , where @xmath15 denotes the underlying probability measure . in other words ,
the response to @xmath16 is only probably correct .
if a query @xmath16 is repeated more than once , then each response is an independent realization of @xmath13 . a new algorithm based on a weighted ( soft - decision ) gbs procedure
is shown to confidently identify @xmath7 after a constant times @xmath1 queries even in the presence of noise ( under the sufficient condition mentioned above ) .
an agnostic algorithm that performs well even if @xmath7 is not in the hypothesis space @xmath2 is also proposed .
the following notation will be used throughout the paper .
the hypothesis space @xmath2 is a finite collection of binary - valued functions defined on a domain @xmath3 , which is called the query space .
each @xmath4 is a mapping from @xmath3 to @xmath5 . for any subset @xmath17
, @xmath18 denotes the number of hypotheses in @xmath19 . the number of hypotheses in @xmath2 is denoted by @xmath20 .
the efficiency of classic binary search is due to the fact at each step there exists a query that splits the pool of viable hypotheses in half .
the existence of such queries is a result of the special ordered structure of the problem . because of ordering
, optimal query locations are easily identified by bisection . in the general
setting in which the query and hypothesis space are arbitrary it is impossible to order the hypotheses in a similar fashion and `` bisecting '' queries may not exist .
for example , consider hypotheses associated with halfspaces of @xmath21 .
each hypothesis takes the value @xmath22 on its halfspace and @xmath23 on the complement .
a bisecting query may not exist in this case . to address such situations
we next introduce a more general framework that does not require an ordered structure .
while it may not be possible to naturally order the hypotheses within @xmath3 , there does exist a similar local geometry that can be exploited in the search process .
observe that the query space @xmath3 can be partitioned into equivalence subsets such that every @xmath4 is constant for all queries in each such subset .
let @xmath24 denote the smallest such partition splits @xmath3 into two disjoint sets .
let @xmath25 and let @xmath26 denote its complement .
@xmath27 is the collection of all non - empty intersections of the form @xmath28 , where @xmath29 , and it is the smallest partition that refines the sets @xmath30 .
@xmath27 is known as the _ join _ of the sets @xmath30 . ] .
note that @xmath31 .
for every @xmath32 and @xmath4 , the value of @xmath33 is constant ( either @xmath22 or @xmath23 ) for all @xmath34 ; denote this value by @xmath35 .
observe that the query selection step in gbs is equivalent to an optimization over the partition cells in @xmath27 .
that is , it suffices to select a partition cell for the query according to @xmath36 .
the main results of this paper concern the query complexity of gbs , but before moving on let us comment on the computational complexity of the algorithm .
the query selection step is the main computational burden in gbs .
however , given @xmath27 the computational complexity of gbs is @xmath37 , up to a constant factor , where @xmath38 denotes the number of partition cells in @xmath27 .
the size and construction of @xmath27 is manageable in many practical situations .
for example , if @xmath3 is finite , then @xmath39 , where @xmath40 is the cardinality of @xmath3 . later , in section [ linear ] , we show that if @xmath2 is defined by @xmath0 halfspaces of @xmath41 , then @xmath38 grows like @xmath42 .
the partition @xmath27 provides a geometrical link between @xmath3 and @xmath2 .
the hypotheses induce a distance function on @xmath27 , and hence @xmath3 . for every pair @xmath43 the hamming distance between the response vectors @xmath44 and
@xmath45 provides a natural distance metric in @xmath3 .
two sets @xmath46 are said to be _ @xmath47-neighbors _ if @xmath47 or fewer hypotheses ( along with their complements , if they belong to @xmath2 ) output different values on @xmath48 and @xmath49 .
[ neighbor ] for example , suppose that @xmath2 is symmetric , so that @xmath4 implies @xmath50 .
then two sets @xmath48 and @xmath49 are @xmath47-neighbors if the hamming distance between their respective response vectors is less than or equal to @xmath51 . if @xmath2 is non - symmetric ( @xmath52 implies that @xmath53 is not in @xmath2 ) , then @xmath48 and @xmath49 are @xmath47-neighbors if the hamming distance between their respective response vectors is less than or equal to @xmath47 .
the pair @xmath10 is said to be _
@xmath47-neighborly _ if the @xmath47-neighborhood graph of @xmath27 is connected ( i.e. , for every pair of sets in @xmath27 there exists a sequence of @xmath47-neighbor sets that begins at one of the pair and ends with the other ) .
[ neighborly ] if @xmath10 is @xmath47-neighborly , then the distance between @xmath48 and @xmath49 is bounded by @xmath47 times the minimum path length between @xmath48 and @xmath49 . moreover
, the neighborly condition implies that there is an incremental way to move from one query to the another , moving a distance of at most @xmath47 at each step .
this local geometry guarantees that near - bisecting queries almost always exist , as shown in the following lemma .
_ assume that @xmath10 is @xmath47-neighborly and define the coherence parameter @xmath54 where the minimization is over all probability mass functions on @xmath27 . for every @xmath55 and any constant @xmath56 satisfying @xmath57 there exists an @xmath32 that approximately bisects @xmath19 @xmath58 or
the set @xmath19 is a small @xmath59 where @xmath18 denotes the cardinality of @xmath19 .
_ [ lemma1 ] _ proof : _ according to the definition of @xmath60 it follows that there exists a probability distribution @xmath61 such that @xmath62 this implies that there exists an @xmath32 such that @xmath63 or there exists a pair @xmath48 and @xmath49 such that @xmath64 in the former case , it follows that a query from @xmath48 will reduce the size of @xmath19 by a factor of at least @xmath65 ( i.e. , every query @xmath66 approximately bisects the subset @xmath19 ) . in latter case , an approximately bisecting query does not exist , but the @xmath47-neighborly condition implies that @xmath18 must be small . to see this note that the @xmath47-neighborly condition guarantees that there exists a sequence of @xmath47-neighbor sets beginning at @xmath48 and ending at @xmath49 . by assumption in this case , @xmath67 on every set and
the sign of @xmath68 must change at some point in the sequence .
it follows that there exist @xmath47-neighbor sets @xmath48 and @xmath49 such that @xmath69 and @xmath70 .
two inequalities follow from this observation .
first , @xmath71 .
second , @xmath72 .
note that if @xmath73 and its complement @xmath74 belong to @xmath19 , then their contributions to the quantity @xmath75 cancel each other . combining these inequalities yields @xmath76 . @xmath77 and @xmath78 . without loss of generality
we may assume that all hypotheses agree with @xmath7 at these two points .
the dashed path between the points @xmath79 and @xmath80 reveals a bisecting query location .
as the path crosses a decision boundary the corresponding hypothesis changes its output from @xmath22 to @xmath23 ( or vice - versa , depending on the direction followed ) . at a certain point , indicated by the shaded cell , half of the hypotheses output @xmath22 and half output @xmath23 .
selecting a query from this cell will _ bisect _ the collection of hypotheses .
, title="fig:",width=377 ] the coherence parameter @xmath60 quantifies the informativeness of queries .
the coherence parameter is optimized over the choice of @xmath61 , rather than sampled at random according to a specific distribution on @xmath3 , because the queries may be selected as needed from @xmath3 .
the minimizer in ( [ cstar ] ) exists because the minimization can be computed over the space of finite - dimensional probability mass functions over the elements of @xmath27 . for @xmath60 to be close to @xmath81 , there must exist a distribution @xmath61 on @xmath27 so that the moment of every @xmath82 is close to zero ( i.e. , for each @xmath4 the probabilities of the responses @xmath22 and @xmath23 are both close to @xmath83 ) .
this implies that there is a way to randomly sample queries so that the expected response of every hypothesis is close to zero . in this sense ,
the queries are incoherent with the hypotheses . in lemma [ lemma1 ] , @xmath60 bounds the proportion of the split of _ any _ subset @xmath19 generated by the best query ( i.e. , the degree to which the best query bisects any subset @xmath19 ) .
the coherence parameter @xmath84 leads to a bound on the number of queries required by gbs . _
if @xmath10 is @xmath47-neighborly , then gbs terminates with the correct hypothesis after at most @xmath85 queries , where @xmath86 .
_ [ thm1 ] _ proof : _ consider the @xmath87th step of the gbs algorithm .
lemma [ lemma1 ] shows that for any @xmath88 either there exists an approximately bisecting query and @xmath89 or @xmath90 .
the uniqueness of the hypotheses with respect to @xmath3 implies that there exists a query that eliminates at least one hypothesis .
therefore , @xmath91 .
it follows that each gbs query reduces the number of viable hypotheses by a factor of at least @xmath92 therefore , @xmath93 and gbs is guaranteed to terminate when @xmath87 satisfies @xmath94 .
taking the logarithm of this inequality produces the query complexity bound .
theorem [ thm1 ] demonstrates that if @xmath10 is neighborly , then the query complexity of gbs is near - optimal ; i.e. , within a constant factor of @xmath95 . the constant depends on coherence parameter @xmath60 and @xmath47 , and clearly it is desirable that both are as small as possible .
note that gbs does not require knowledge of @xmath60 or @xmath47 .
we also remark that the constant in the bound is not necessarily the best that one can obtain .
the proof involves selecting @xmath56 to balance splitting factor @xmath96 and the `` tail '' behavior @xmath97 , and this may not give the best bound .
the coherence parameter @xmath60 can be computed or bounded for many pairs @xmath10 that are commonly encountered in applications , as covered later in section [ coherence ] .
in noisy problems , the search must cope with erroneous responses .
specifically , assume that for any query @xmath8 the binary response @xmath12 is an independent realization of the random variable @xmath13 satisfying @xmath14 ( i.e. , the response is only probably correct ) .
if a query @xmath16 is repeated more than once , then each response is an independent realization of @xmath13 . define the _ noise - level _ for the query @xmath16 as @xmath98 . throughout the paper we will let @xmath99 and assume that @xmath100 . before presenting the main approach to noisy gbs , we first consider a simple strategy based on repetitive querying that will serve as a benchmark for comparison .
we begin by describing a simple noise - tolerant version of gbs .
the noise - tolerant algorithm is based on the simple idea of repeating each query of the gbs several times , in order to overcome the uncertainty introduced by the noise .
similar approaches are proposed in the work k " a " ari " ainen @xcite . karp and kleinberg @xcite analyze of this strategy for noise - tolerant classic binary search .
this is essentially like using a simple repetition code to communicate over a noisy channel .
this procedure is termed noise - tolerant gbs ( ngbs ) and is summarized in fig . [
fig : ngbs ] .
[ thm : ngbs ] consider a specific query @xmath101 repeated @xmath102 times , let @xmath103 denote the frequency of @xmath22 in the @xmath102 trials , and let @xmath104 $ ] .
the majority vote decision is correct if @xmath105 . by chernoff s
bound we have @xmath106 .
the results follows by the union bound . based on the bound above
, @xmath102 must satisfy @xmath107 to guarantee that the labels determined for all @xmath108 queries are correct with probability @xmath109 .
the query complexity of ngbs can thus be bounded by @xmath110 recall that @xmath111 , the cardinality of @xmath2 . if @xmath112 , then bound on the query complexity of ngbs is proportional to @xmath113 , a logarithmic factor worse than the query complexity in the noiseless setting . moreover ,
if an upper bound on @xmath108 is not known in advance , then one must assume the worst - case value , @xmath114 , in order to set @xmath102 .
that is , in order to guarantee that the correct hypothesis is determined with probability at least @xmath109 , the required number of repetitions of each query is @xmath115 in this situation , the bound on the query complexity of ngbs is proportional to @xmath116 , compared to @xmath117 in the noiseless setting .
it is conjectured that the extra logarithmic factor can not be removed from the query complexity ( i.e. , it is unavoidable using repetitive queries ) . as we show next ,
these problems can be eliminated by a more sophisticated approach to noisy gbs .
a more effective approach to noisy gbs is based on the following soft - decision procedure .
a similar procedure has been shown to be near - optimal for the noisy ( classic ) binary search problem by burnashev and zigangirov @xcite and later independently by karp and kleinberg @xcite .
the crucial distinction here is that gbs calls for a more general approach to query selection and a fundamentally different convergence analysis .
let @xmath118 be a known probability measure over @xmath2 .
that is , @xmath119 $ ] and @xmath120 .
the measure @xmath118 can be viewed as an initial weighting over the hypothesis class .
for example , taking @xmath118 to be the uniform distribution over @xmath2 expresses the fact that all hypothesis are equally reasonable prior to making queries .
we will assume that @xmath118 is uniform for the remainder of the paper , but the extension to other initial distributions is trivial .
note , however , that we still assume that @xmath6 is fixed but unknown .
after each query and response @xmath121 @xmath122 the distribution is updated according to @xmath123 where @xmath124 , @xmath125 is any constant satisfying @xmath126 , and @xmath127 is normalized to satisfy @xmath128 .
the update can be viewed as an application of bayes rule and its effect is simple ; the probability masses of hypotheses that agree with the label @xmath129 are boosted relative to those that disagree .
the parameter @xmath125 controls the size of the boost .
the hypothesis with the largest weight is selected at each step : @xmath130 if the maximizer is not unique , one of the maximizers is selected at random . note that , unlike the hard - decisions made by the gbs algorithm in fig .
[ fig : gbs ] , this procedure does not eliminate hypotheses that disagree with the observed labels , rather the weight assigned to each hypothesis is an indication of how successful its predictions have been .
thus , the procedure is termed soft - decision gbs ( sgbs ) and is summarized in fig . [
fig : sgbs ] .
the goal of sgbs is to drive the error @xmath131 to zero as quickly as possible by strategically selecting the queries .
the query selection at each step of sgbs must be informative with respect to the distribution @xmath132 .
in particular , if the _ weighted prediction _
@xmath133 is close to zero for a certain @xmath16 ( or @xmath48 ) , then a label at that point is informative due to the large disagreement among the hypotheses .
if multiple @xmath32 minimize @xmath134 , then one of the minimizers is selected uniformly at random . to analyze sgbs ,
define @xmath135 , @xmath136 .
the variable @xmath137 was also used by burnashev and zigangirov @xcite to analyze classic binary search .
it reflects the amount of mass that @xmath138 places on incorrect hypotheses .
let @xmath15 denotes the underlying probability measure governing noises and possible randomization in query selection , and let @xmath139 denote expectation with respect to @xmath15 .
note that by markov s inequality @xmath140 \ .
\label{markov } \end{aligned}\ ] ] at this point , the method of analyzing sgbs departs from that of burnashev and zigangirov @xcite which focused only on the classic binary search problem .
the lack of an ordered structure calls for a different attack on the problem , which is summarized in the following results and detailed in the appendix . _
consider any sequence of queries @xmath141 and the corresponding responses @xmath142 .
if @xmath143 , then @xmath144 is a nonnegative supermartingale with respect to @xmath145 ; i.e. , @xmath146 \leq c_n$ ] for all @xmath136 .
_ [ martingale ] the lemma is proved in the appendix .
the condition @xmath147 ensures that the update ( [ update1 ] ) is not overly aggressive .
it follows that @xmath148 \leq c_0 $ ] and by the martingale convergence theorem we have that @xmath149 exists and is finite ( for more information on martingale theory one can refer to the textbook by brmaud @xcite ) .
furthermore , we have the following theorem .
_ _ [ thm2 ] first observe that for every positive integer @xmath87 @xmath150 & = & \e[(\mm_n/\mm_{n-1 } ) \ , \mm_{n-1 } ] \ = \
\e\left[\e[(\mm_n/\mm_{n-1 } ) \ , \mm_{n-1}|p_{n-1}]\right ] \\ & = & \e\left[\mm_{n-1 } \ , \e[(\mm_n/\mm_{n-1})|p_{n-1}]\right ] \ \leq \ \e[\mm_{n-1 } ] \ , \max_{p_{n-1}}\e[(\mm_n/\mm_{n-1})|p_{n-1 } ] \\ & \leq & \mm_0 \left(\max_{i = 0,\dots , n-1 } \max_{p_i}\ , \e[(\mm_{i+1}/\mm_{i})|p_i]\right)^n \ .\end{aligned}\ ] ] in the proof of lemma [ martingale ] , it is shown that if @xmath151 , then @xmath152<1 $ ] for every @xmath153 and therefore @xmath154~<~1 $ ] .
it follows that the sequence @xmath155\right)^n$ ] is monotonically decreasing .
the result follows from ( [ markov ] ) .
note that if we can determine an upper bound for the sequence @xmath154\leq
1-\lambda < 1 $ ] , @xmath156 , then it follows that that @xmath157 .
the modified sgbs algorithm is outlined in fig .
[ fig : msgbs ] .
it is easily verified that lemma [ martingale ] and theorem [ thm2 ] also hold for the modified sgbs algorithm .
this follows since the modified query selection step is identical to that of the original sgbs algorithm , unless there exist two neighboring sets with strongly bipolar weighted responses . in the latter case ,
a query is randomly selected from one of these two sets with equal probability . for every @xmath32 and any probability measure @xmath158 on @xmath2 the _ weighted prediction _ on @xmath48
is defined to be @xmath159 , where @xmath35 is the constant value of @xmath73 for every @xmath34 .
the following lemma , which is the soft - decision analog of lemma [ lemma1 ] , plays a crucial role in the analysis of the modified sgbs algorithm . _
if @xmath10 is @xmath47-neighborly , then for every probability measure @xmath158 on @xmath2 there either exists a set @xmath32 such that @xmath160 or a pair of @xmath47-neighbor sets @xmath46 such that @xmath161 and @xmath162 . _ [ lemma2 ] suppose that @xmath163 . then there must exist @xmath46 such that @xmath164 and @xmath165 , otherwise @xmath60 can not be the incoherence parameter of @xmath2 , defined in ( [ cstar ] ) .
to see this suppose , for instance , that @xmath164 for all @xmath166 .
then for every distribution @xmath61 on @xmath3 we have @xmath167 .
this contradicts the definition of @xmath60 since @xmath168 .
the neighborly condition guarantees that there exists a sequence of @xmath47-neighbor sets beginning at @xmath48 and ending at @xmath49 . since @xmath169 on every set and
the sign of @xmath170 must change at some point in the sequence , it follows that there exist @xmath47-neighbor sets satisfying the claim .
the lemma guarantees that there exists either a set in @xmath27 on which the weighted hypotheses significantly disagree ( provided @xmath60 is significantly below @xmath171 ) or two neighboring sets in @xmath27 on which the weighted predictions are strongly bipolar . in either case , if a query is drawn randomly from these sets , then the weighted predictions are highly variable or uncertain , with respect to @xmath158 .
this makes the resulting label informative in either case .
if @xmath10 is @xmath171-neighborly , then the modified sgbs algorithm guarantees that @xmath172 exponentially fast .
the @xmath171-neighborly condition is required so that the expected boost to @xmath173 is significant at each step .
if this condition does not hold , then the boost could be arbitrarily small due to the effects of other hypotheses .
fortunately , as shown in section [ coherence ] , the @xmath171-neighborly condition holds in a wide range of common situations .
_ let @xmath15 denote the underlying probability measure ( governing noises and algorithm randomization ) . if @xmath151 and @xmath10 is @xmath171-neighborly , then the modified sgbs algorithm in fig .
[ fig : msgbs ] generates a sequence of hypotheses satisfying @xmath174 with exponential constant @xmath175 , where @xmath60 is defined in ( [ cstar ] ) .
_ [ thm3 ] the theorem is proved in the appendix .
the exponential convergence rate in the exponential rate parameter @xmath176 is a positive constant strictly less than @xmath171 . for a noise level @xmath177
this factor is maximized by a value @xmath178 which tends to @xmath179 as @xmath177 tends to @xmath83 . ]
is governed by the coherence parameter @xmath180 . as shown in section [ coherence ] , the value of @xmath60 is typically a small constant much less than @xmath171 that is independent of the size of @xmath2 . in such situations ,
the query complexity of modified sgbs is near - optimal .
the query complexity of the modified sgbs algorithm can be derived as follows .
let @xmath181 be a pre - specified confidence parameter .
the number of queries required to ensure that @xmath182 is @xmath183 , which is near - optimal . intuitively , about @xmath1 bits are required to encode each hypothesis .
more formally , the noisy classic binary search problem satisfies the assumptions of theorem [ thm3 ] ( as shown in section [ cbs ] ) , and hence it is a special case of the general problem . using information - theoretic methods ,
it has been shown by burnashev and zigangirov @xcite ( also see the work of karp and kleinberg @xcite ) that the query complexity for noisy classic binary search is also within a constant factor of @xmath184 .
in contrast , the query complexity bound for ngbs , based on repeating queries , is at least logarithmic factor worse .
we conclude this section with an example applying theorem [ thm3 ] to the halfspace learning problem .
_ consider learning multidimensional halfspaces .
let @xmath185 and consider hypotheses of the form @xmath186 where @xmath187 and @xmath188 parameterize the hypothesis @xmath189 and @xmath190 is the inner product in @xmath191 .
the following corollary characterizes the query complexity for this problem .
_ let @xmath2 be a finite collection of hypotheses of form ( [ halfspace ] ) and assume that the responses to each query are noisy , with noise bound @xmath100 .
then the hypotheses selected by modified sgbs with @xmath192 satisfy @xmath193 with @xmath194 .
moreover , @xmath195 can be computed in time polynomial in @xmath196 .
[ thms ] the error bound follows immediately from theorem [ thm3 ] since @xmath197 and @xmath198 is @xmath171-neighborly , as shown in section [ linear ] .
the polynomial - time computational complexity follows from the work of buck @xcite , as discussed in section [ linear ] .
suppose that @xmath2 is an @xmath199-dense set with respect to a uniform probability measure on a ball in @xmath191 ( i.e. , for _ any _ hyperplane of the form ( [ halfspace ] ) @xmath2 contains a hypothesis whose probability of error is within @xmath199 of it ) .
the size of such an @xmath2 satisfies @xmath200 , for a constant @xmath201 , which is the proportional to the minimum query complexity possible in this setting , as shown by balcan et al @xcite .
those authors also present an algorithm with roughly the same query complexity for this problem .
however , their algorithm is specifically designed for the linear threshold problem .
remarkably , near - optimal query complexity is achieved in polynomial - time by the general - purpose modified sgbs algorithm .
so far we have assumed that the correct hypothesis @xmath7 is in @xmath2 . in this section
we drop this assumption and consider _ agnostic _ algorithms guaranteed to find the best hypothesis in @xmath2 even if the correct hypothesis @xmath7 is not in @xmath2 and/or the assumptions of theorem [ thm1 ] or [ thm3 ] do not hold .
the best hypothesis in @xmath2 can be defined as the one that minimizes the error with respect to a probability measure on @xmath3 , denoted by @xmath202 , which can be arbitrary .
this notion of `` best '' commonly arises in machine problems where it is customary to measure the error or _
risk _ with respect to a distribution on @xmath3
. a common approach to hypothesis selection is _ empirical risk minimization _ ( erm ) , which uses queries randomly drawn according to @xmath202 and then selects the hypothesis in @xmath2 that minimizes the number of errors made on these queries . given a budget of @xmath87 queries , consider the following agnostic procedure . divide the query budget into three equal portions .
use gbs ( or ngbs or modified sgbs ) with one portion , erm ( queries randomly distributed according @xmath202 ) with another , and then allocate the third portion to queries from the subset of @xmath3 where the hypothesis selected by gbs ( or ngbs or modified sgbs ) and the hypothesis selected by erm disagree , with these queries randomly distributed according to the restriction of @xmath202 to this subset . finally , select the hypothesis that makes the fewest mistakes on the third portion as the final choice .
the sample complexity of this agnostic procedure is within a constant factor of that of the better of the two competing algorithms .
for example , if the conditions of theorems [ thm1 ] or [ thm3 ] hold , then the sample complexity of the agnostic algorithm is proportional to @xmath117 . in general , the sample complexity of the agnostic procedure is within a constant factor of that of erm alone . we formalize this as follows .
[ runoff]_let @xmath202 denote a probability measure on @xmath3 and for every @xmath82 let @xmath203 denote its probability of error with respect to @xmath202 .
consider two hypotheses @xmath204 and let @xmath205 denote the subset of queries for which @xmath206 and @xmath207 disagree ; i.e. , @xmath208 for all @xmath209 .
suppose that @xmath210 queries are drawn independently from @xmath211 , the restriction of @xmath202 to the set @xmath212 , let @xmath213 and @xmath214 denote average number of errors made by @xmath206 and @xmath207 on these queries , let @xmath215 $ ] and @xmath216 $ ] , and select @xmath217 .
then @xmath218 with probability less than @xmath219 . _ define @xmath220 and let @xmath221 . by hoeffding s inequality
we have @xmath222 $ ] with probability at least @xmath223 .
it follows that @xmath224 .
for example , if @xmath225 then since @xmath226 we may take @xmath227 to obtain @xmath228 .
the result follows since @xmath206 and @xmath207 agree on the complement of @xmath212 .
note that there is a distinct advantage to drawing queries from @xmath229 rather than @xmath202 , since the error exponent is proportional to @xmath230 which is greater than or equal to @xmath231 .
now to illustrate the idea , consider an agnostic procedure based on modified sgbs and erm .
the following theorem is proved in the appendix .
[ thm5 ] _ let @xmath202 denote a measure on @xmath3 and suppose we have a query budget of @xmath87 .
let @xmath206 denote the hypothesis selected by modified sgbs using @xmath232 of the queries and let @xmath207 denote the hypothesis selected by erm from @xmath232 queries drawn independently from @xmath202 . draw the remaining @xmath232 queries independently from @xmath229 , the restriction of @xmath202 to the set @xmath212 on which @xmath206 and @xmath207 disagree , and let @xmath213 and @xmath214 denote the average number of errors made by @xmath206 and @xmath207 on these queries .
select @xmath217 . then , in general , @xmath233 & \leq & \min\{\e[r(h_1)],\e[r(h_2)]\ } \
, + \ , \sqrt{3/n } \ , \end{aligned}\ ] ] where @xmath203 , @xmath82 , denotes the probability of error of @xmath73 with respect to @xmath202 . furthermore ,
if the assumptions of theorem [ thm3 ] hold and @xmath6 , then @xmath234 where @xmath177 is the noise bound . _ note that if the assumptions of theorem [ thm3 ] hold , then the agnostic procedure performs almost as well as modified sgbs alone .
in particular , the number of queries required to ensure that @xmath235 is proportional to @xmath184 ; optimal up to constant factors . also observe that the probability bound implies the following bound in expectation : @xmath233 & \leq & r(h^ * ) + \
ne^{-\lambda n/3 } \
+ \ 2 e^{-n|1 - 2\alpha|^2/6 } \\
& \leq & r(h^ * ) \ + \
\n e^{-c n } \ , \end{aligned}\ ] ] where @xmath201 is a constant depending on @xmath176 and @xmath177 .
the exponential convergence of the expected risk is much faster than the usual parametric rate for erm .
if the conditions of theorem [ thm3 ] are not met , then modified sgbs ( alone ) may perform poorly since it might select inappropriate queries and could even terminate with an incorrect hypothesis . however , the expected error of the agnostic selection @xmath236 is within @xmath237 of the expected error of erm , with no assumptions on the underlying distributions .
note that the expected error of erm is proportional to @xmath238 in the worst - case situation .
therefore , the agnostic procedure is near - optimal in general .
the agnostic procedure offers this safeguard on performance .
the same approach could be used to derive agnostic procedures from any _ active learning _
scheme ( i.e. , learning from adaptively selected queries ) , including gbs or ngbs
. we also note that the important element in the agnostic procedure is the selection of queries ; the proposed selection of @xmath236 based on those queries is convenient for proving the bounds , but not necessarily optimal .
in this section we examine a variety of common situations in which the neighborliness condition can be verified .
we will confine the discussion to gbs in the noise - free situation , and analogous results hold in the presence of noise . for a given pair @xmath10 ,
the effectiveness of gbs hinges on determining ( or bounding ) @xmath60 and establishing that @xmath10 are neighborly .
recall the definition of the bound @xmath60 from ( [ cstar ] ) and that @xmath111 , the cardinality of @xmath2 .
a _ trivial _ bound for @xmath60
is @xmath239 which is achieved by allocating @xmath240 mass to each hypothesis and evenly distributing @xmath241 mass to set(s ) @xmath48 where @xmath242 and @xmath241 on set(s ) @xmath48 where @xmath243 .
non - trivial coherence bounds are those for which there exists a @xmath61 and @xmath244 that does not depend on @xmath0 such that @xmath245 the coherence parameter @xmath60 is analytically determined or bounded in several illustrative applications below .
we also note that it may be known a priori that @xmath60 is bounded far way from @xmath171 .
suppose that for a certain @xmath61 on @xmath3 ( or @xmath27 ) the absolute value of the first - moment of the correct hypothesis ( w.r.t .
@xmath61 ) is known to be upper bounded by a constant @xmath246 .
then all hypotheses that violate the bound can be eliminated from consideration .
thus the constant @xmath56 is an upper bound on @xmath60 .
situations like this can arise , for example , in binary classification problems with side / prior knowledge that the marginal probabilities of the two classes are somewhat balanced .
then the moment of the correct hypothesis , with respect to the marginal probability distribution on @xmath3 , is bounded far away from @xmath171 and @xmath23 .
the neighborly condition can be numerically verified in a straightforward fashion .
enumerate the equivalence sets in @xmath27 as @xmath247 . to check whether the @xmath47-neighborhood graph is connected , form an @xmath248 matrix @xmath249
whose @xmath250 entry is @xmath171 if @xmath251 and @xmath252 are k - neighbors and @xmath81 otherwise .
normalize the rows of @xmath249 so that each sums to @xmath171 and denote the resulting _
matrix by @xmath253 .
the @xmath47-neighborhood graph is connected if and only if there exists an integer @xmath254 , @xmath255 , such that @xmath256 contains no zero entries .
this follows from the standard condition for state accessibility in markov chains ( for background on markov chains one can refer to the textbook by brmaud @xcite ) .
the smallest @xmath47 for which the @xmath47-neighborhood graph is connected can be determined using a binary search over the set @xmath257 , checking the condition above for each value of @xmath47 in the search .
this idea was suggested to me by clayton scott .
thus , the neighborly condition can be verified in polynomial - time in @xmath258 .
alternatively , in many cases the neighborly condition can be verified analytically , as demonstrated in following applications .
first we show that gbs reduces to classic binary search .
let @xmath259 be the collection of binary - valued functions on @xmath260 $ ] of the following form , @xmath261 for @xmath262 ( and @xmath263 ) .
assume that @xmath6 .
first consider the neighborly condition . recall that @xmath27 is the smallest partition of @xmath3 into equivalence sets induced by @xmath2 . in this case , each @xmath48 is an interval of the form @xmath264 , @xmath262 .
observe that only a single hypothesis , @xmath189 , has different responses to queries from @xmath251 and @xmath265 and so they are @xmath171-neighbors , for @xmath266 .
moreover , the @xmath171-neighborhood graph is connected in this case , and so @xmath10 is @xmath171-neighborly .
next consider coherence parameter @xmath60 .
take @xmath61 to be two point masses at @xmath267 and @xmath268 of probability @xmath83 each .
then @xmath269 for every @xmath4 , since @xmath270 and @xmath271 .
thus , @xmath272 . since @xmath273 and @xmath274 , we have @xmath275 and the query complexity of gbs is proportional to @xmath1 according to theorem [ thm1 ]
. the reduction factor of @xmath276 , instead of @xmath83 , arises because we allow the situation in which the number of hypotheses may be odd ( e.g. , given three hypotheses ) , the best query may eliminate just one ) .
if @xmath0 is even , then the query complexity is @xmath277 , which is information - theoretically optimal .
now let @xmath260 $ ] and consider a finite collection of hypotheses @xmath278 , where @xmath189 takes the value @xmath22 when @xmath279 , for a pair @xmath280 , and @xmath23 otherwise .
assume that @xmath6 .
the partition @xmath27 again consists of intervals , and the neighborly condition is satisfied with @xmath274 . to bound @xmath60 , note that the minimizing @xmath61 must place some mass within and outside each interval @xmath281 .
if the intervals all have length at least @xmath282 , then taking @xmath61 to be the uniform measure on @xmath283 $ ] yields that @xmath284 , regardless of the number of interval hypotheses under consideration .
therefore , in this setting theorem [ thm1 ] guarantees that gbs determines the correct hypothesis using at most a constant times @xmath1 steps .
however , consider the special case in which the intervals are disjoint .
then it is not hard to see that the best allocation of mass is to place @xmath285 mass in each subinterval , resulting in @xmath286 .
and so , theorem [ thm1 ] only guarantees that gbs is will terminate in at most @xmath0 steps ( the number of steps required by exhaustive linear search ) .
in fact , it is easy to see that no procedure can do better than linear search in this case and the query complexity of any method is proportional to @xmath287 .
however , note that if queries of a different form were allowed , then much better performance is possible . for example ,
if queries in the form of dyadic subinterval tests were allowed ( e.g. , tests that indicate whether or not the correct hypothesis is @xmath22-valued anywhere within a dyadic subinterval of choice ) , then the correct hypothesis can be identified through @xmath288 queries ( essentially a binary encoding of the correct hypothesis ) .
this underscores the importance of the geometrical relationship between @xmath3 and @xmath2 embodied in the neighborly condition and the incoherence parameter @xmath60 . optimizing the query space to the structure of @xmath2
is related to the notion of arbitrary queries examined in the work of kulkarni et al @xcite , and somewhat to the theory of compressed sensing developed by candes et al @xcite and donoho @xcite .
let @xmath289 be a collection of multidimensional threshold functions of the following form .
the threshold of each @xmath189 determined by ( possibly nonlinear ) decision surface in @xmath290-dimensional euclidean space and the queries are points in @xmath41 .
it suffices to consider linear decision surfaces of the form @xmath291 where @xmath187 , @xmath292 , the offset @xmath293 satisfies @xmath294 for some constant @xmath295 , and @xmath190 denotes the inner product in @xmath191 . each hypothesis is associated with a halfspace of @xmath191 . note that hypotheses of this form can be used to represent nonlinear decision surfaces , by first applying a mapping to an input space and then forming linear decision surfaces in the induced query space .
the problem of learning multidimensional threshold functions arises commonly in computer vision ( see the review of swain and stricker@xcite and applications by geman and jedynak @xcite and arkin et al @xcite ) , image processing studied by korostelev and kim @xcite , and active learning research ; for example the investigations by freund et al @xcite , dasgupta @xcite , balcan et al @xcite , and castro and nowak @xcite .
first we show that the pair @xmath198 is @xmath171-neighborly .
each @xmath166 is a polytope in @xmath191 .
these polytopes are generated by intersections of the halfspaces corresponding to the hypotheses .
any two polytopes that share a common face are @xmath171-neighbors ( the hypothesis whose decision boundary defines the face , and its complement if it exists , are the only ones that predict different values on these two sets ) . since the polytopes tessellate @xmath191 , the @xmath171-neighborhood graph of @xmath27 is connected .
we next show that the the coherence parameter @xmath197 .
since the offsets of the hypotheses are all less than @xmath296 in magnitude , it follows that the distance from the origin to the nearest point of the decision surface of every hypothesis is at most @xmath296 .
let @xmath297 denote the uniform probability distribution on a ball of radius @xmath298 centered at the origin in @xmath191 .
then for every @xmath73 of the form ( [ surface ] ) there exists a constant @xmath201 ( depending on @xmath296 ) such that @xmath299 and @xmath300 .
therefore @xmath272 , and it follows from theorem [ thm1 ] guarantees that gbs determines the correct multidimensional threshold in at most @xmath301 steps . to the best of our knowledge this is a new result in the theory of learning multidimensional threshold functions , although similar query complexity bounds have been established for the subclass of linear threshold functions with @xmath302 ( threshold boundaries passing through the origin ) ; see for example the work of balcan et al @xcite . these results are based on somewhat different learning algorithms , assumptions and analysis techniques . observe that if @xmath2 is an @xmath199-dense ( with respect lesbegue measure over a compact set in @xmath191 ) subset of the continuous class of threshold functions of the form ( [ surface ] ) , then the size of the @xmath2 satisfies @xmath303 .
therefore the query complexity of gbs is proportional to the metric entropy of the continuous class , and it follows from the results of kulkarni et al @xcite that no learning algorithm exists with a lower query complexity ( up to constant factors ) .
furthermore , note that the computational complexity of gbs for hypotheses of the form ( [ surface ] ) is proportional to the cardinality of @xmath27 , which is equal to the number of polytopes generated by intersections of half - spaces .
it is a well known fact ( see buck @xcite ) that @xmath304 .
therefore , gbs is a polynomial - time algorithm for this problem . in general
, the cardinality of @xmath27 could be as large as @xmath305 .
next let @xmath2 again be the hypotheses of the form ( [ surface ] ) , but let @xmath306^d$ ] , instead of all of @xmath191 .
this constraint on the query space affects the bound on the coherence @xmath60 . to bound @xmath60 ,
let @xmath61 be point masses of probability @xmath307 at each of the @xmath308 vertices of the cube @xmath309^d$ ] ( the natural generalization of the @xmath61 chosen in the case of classic binary search in section [ cbs ] above ) .
then @xmath310 for every @xmath4 , since for each @xmath73 there is at least one vertex on where it predicts @xmath22 and one where it predicts @xmath23 .
thus , @xmath311 . we conclude that the gbs determines the correct hypothesis in proportional to @xmath312 steps .
the dependence on @xmath308 is unavoidable , since it may be that each threshold function takes that value @xmath22 only at one of the @xmath308 vertices and so each vertex must be queried .
a noteworthy special case is arises when @xmath302 ( i.e. , the threshold boundaries pass through the origin ) . in this case , with @xmath61 as specified above , @xmath272 , since each hypothesis responds with @xmath22 at half of the vertices and @xmath23 on the other half .
therefore , the query complexity of gbs is at most @xmath301 , independent of the dimension . as discussed above , similar results for this special case
have been previously reported based on different algorithms and analyses ; see the results in the work of balcan et al @xcite and the references therein .
note that even if the threshold boundaries do not pass through the origin , and therefore the number of queries needed is proportional to @xmath1 so long as @xmath313 . the dependence on dimension @xmath290
can also be eliminated if it is known that for a certain distribution @xmath61 on @xmath3 the absolute value of the moment of the correct hypothesis w.r.t .
@xmath61 is known to be upper bounded by a constant @xmath246 , as discussed at the beginning of this section .
finally , we also mention hypotheses associated with axis - aligned rectangles in @xmath283^d$ ] , the multidimensional version of the interval hypotheses considered above .
an axis - aligned rectangle is defined by its boundary coordinates in each dimension , @xmath314 , @xmath315 .
the hypothesis associated with such a rectangle takes the value @xmath316 on the set @xmath317^d : \
a_j \leq x_j \leq b_j , \ j=1,\dots , d\}$ ] and @xmath23 otherwise . the complementary hypothesis may also be included .
consider a finite collection @xmath2 of hypotheses of this form .
if the rectangles associated with each @xmath82 have volume at least @xmath318 , then by taking @xmath61 to be the uniform measure on @xmath283^d$ ] it follows that the coherence parameter @xmath319 for this problem .
the cells of partition @xmath27 of @xmath283 $ ] associated with a collection of such hypotheses are rectangles themselves .
if the boundaries of the rectangles associated with the hypotheses are distinct , then the @xmath171-neighborhood graph of @xmath27 is connected .
theorem [ thm1 ] implies that the number of queries needed by gbs to determine the correct rectangle is proportional to @xmath320 . in many situations both the hypothesis and query spaces may be discrete .
a machine learning application , for example , may have access to a large ( but finite ) pool of unlabeled examples , any of which may be queried for a label . because obtaining labels can be costly , `` active '' learning algorithms select only those examples that are predicted to be highly informative for labeling .
theorem [ thm1 ] applies equally well to continuous or discrete query spaces .
for example , consider the linear separator case , but instead of the query space @xmath191 suppose that @xmath3 is a finite subset of points in @xmath191 .
the hypotheses again induce a partition of @xmath3 into subsets @xmath24 , but the number of subsets in the partition may be less than the number in @xmath321 . consequently , the neighborhood graph of @xmath24 depends on the specific points that are included in @xmath3 and may or may not be connected .
as discussed at the beginning of this section , the neighborly condition can be verified in polynomial - time ( polynomial in @xmath39 ) .
consider two illustrative examples .
let @xmath2 be a collection of linear separators as in ( [ surface ] ) above and first reconsider the partition @xmath321 .
recall that each set in @xmath321 is a polytope .
suppose that a discrete set @xmath3 contains at least one point inside each of the polytopes in @xmath321 .
then it follows from the results above that @xmath10 is @xmath171-neighborly .
second , consider a simple case in @xmath322 dimensions .
suppose @xmath3 consists of just three non - colinear points @xmath323 and suppose that @xmath2 is comprised of six classifiers , @xmath324 , satisfying @xmath325 , @xmath326 , @xmath327 , and @xmath328 , @xmath327 . in this case ,
@xmath329 and the responses to any pair of queries differ for four of the six hypotheses .
thus , the @xmath330-neighborhood graph of @xmath24 is connected , but the @xmath171-neighborhood is not . also note that a finite query space naturally limits the number of hypotheses that need be considered .
consider an uncountable collection of hypotheses .
the number of unique labeling assignments generated by these hypotheses can be bounded in terms of the vc dimension of the class ; see the book by vapnik for more information on vc theory @xcite . as a result
, it suffices to consider a finite subset of the hypotheses consisting of just one representative of each unique labeling assignment .
furthermore , the computational complexity of gbs is proportional to @xmath331 in this case .
generalized binary search can be viewed as a generalization of classic binary search , shannon - fano coding as noted by goodman and smyth @xcite , and channel coding with noiseless feedback as studied by horstein @xcite .
problems of this nature arise in many applications , including channel coding ( e.g. , the work of horstein @xcite and zigangirov @xcite ) , experimental design ( e.g. , as studied by rnyi @xcite ) , disease diagnosis ( e.g. , see the work of loveland @xcite ) , fault - tolerant computing ( e.g. , the work of feige et al @xcite ) , the scheduling problem considered by kosaraju et al @xcite , computer vision problems investigated by geman and jedynak @xcite and arkin et al @xcite ) , image processing problems studied by korostelev and kim @xcite , and active learning research ; for example the investigations by freund et al @xcite , dasgupta @xcite , balcan et al @xcite , and castro and nowak @xcite .
past work has provided a partial characterization of this problem .
if the responses to queries are noiseless , then selecting the sequence of queries from @xmath3 is equivalent to determining a binary decision tree , where a sequence of queries defines a path from the root of the tree ( corresponding to @xmath2 ) to a leaf ( corresponding to a single element of @xmath2 ) . in general the determination of the optimal ( worst- or average - case )
tree is np - complete as shown by hyafil and rivest @xcite .
however , there exists a greedy procedure that yields query sequences that are within a factor of @xmath1 of the optimal search tree depth ; this result has been discovered independently by several researchers including loveland @xcite , garey and graham @xcite , arkin et al @xcite , and dasgupta @xcite .
the greedy procedure is referred to here as _ generalized binary search _
( gbs ) or the _ splitting algorithm _ , and it reduces to classic binary search , as discussed in section [ cbs ] .
the number of queries an algorithm requires to determine @xmath7 is called the _ query complexity _ of the algorithm . since the hypotheses are assumed to be distinct , it is clear that the query complexity of gbs is at most @xmath0 ( because it is always possible to find query that eliminates at least one hypothesis at each step ) .
in fact , there are simple examples ( see section [ cbs ] ) demonstrating that this is the best one can hope to do in general .
however , it is also true that in many cases the performance of gbs can be much better , requiring as few as @xmath277 queries . in classic binary search , for example ,
half of the hypotheses are eliminated at each step ( e.g. , refer to the textbook by cormen et al @xcite ) . rnyi first considered a form of binary search with noise @xcite and explored its connections with information theory @xcite . in particular , the problem of sequential transmission over a binary symmetric channel with noiseless feedback , as formulated by horstein @xcite and studied by burnashev and zigangirov @xcite and more recently by pelc et al @xcite , is equivalent to a noisy binary search problem . there is a large literature on learning from queries ; see the review articles by angluin @xcite .
this paper focuses exclusively on membership queries ( i.e. , an @xmath332 is the query and the response is @xmath9 ) , although other types of queries ( equivalence , subset , superset , disjointness , and exhaustiveness ) are possible as discussed by angluin @xcite . _ arbitrary queries _
have also been investigated , in which the query is a subset of @xmath2 and the output is @xmath22 if @xmath7 belongs to the subset and @xmath23 otherwise .
a finite collection of hypotheses @xmath2 can be successively halved using arbitrary queries , and so it is possible to determine @xmath7 with @xmath95 arbitrary queries , the information - theoretically optimal query complexity discussed by kulkarni et al @xcite .
membership queries are the most natural in function learning problems , and because this paper deals only with this type we will simply refer to them as queries throughout the rest of the paper .
the number of queries required to determine a binary - valued function in a finite collection of hypotheses can be bounded ( above and below ) in terms of a combinatorial parameter of @xmath10 due to hegeds@xcite ( see the work of hellerstein et al @xcite for related work ) . due to its combinatorial nature , computing such bounds are generally np - hard . in contrast , the geometric relationship between @xmath3 and @xmath2 developed in this paper leads to an upper bound on the query complexity that can be determined analytically or computed in polynomial time in many cases of interest .
the term gbs is used in this paper to emphasize connections and similarities with classic binary search , which is a special case the general problem considered here .
classic binary search is equivalent to learning a one - dimensional binary - valued threshold function by selecting point evaluations of the function according to a bisection procedure .
consider the threshold function @xmath333 on the interval @xmath334 $ ] for some threshold value @xmath335 . throughout the paper
we adopt the convention that @xmath336 .
suppose that @xmath337 belongs to the discrete set @xmath338 and let @xmath2 denote the collection of threshold functions @xmath339 .
the value of @xmath337 can then determined from a constant times @xmath117 queries using a bisection procedure analogous to the game of twenty questions .
in fact , this is precisely what gbs performs in this case ( i.e. , gbs reduces to classic binary search in this setting ) .
if @xmath340 for some integer @xmath210 , then each point evaluation provides one bit in the @xmath210-bit binary expansion of @xmath337 .
thus , classic binary search is information - theoretically optimal ; see the book by traub , wasilkowski and wozniakowski @xcite for a nice treatment of classic bisection and binary search .
the main results of this paper generalize the salient aspects of classic binary search to a much broader class of problems .
in many ( if not most ) applications it is unrealistic to assume that the responses to queries are without error . a form of binary search with noise appears to have been first posed by rnyi @xcite .
the noisy binary search problem arises in sequential transmission over a binary symmetric channel with noiseless feedback studied by horstein @xcite and zigangirov @xcite .
the survey paper by pelc et al @xcite discusses the connections between search and coding problems . in channel coding with feedback
, each threshold corresponds to a unique binary codeword ( the binary expansion of @xmath337 ) .
thus , channel coding with noiseless feedback is equivalent to the problem of learning a one - dimensional threshold function in binary noise , as noted by burnashev and zigangirov @xcite .
the near - optimal solutions to the _ noisy binary search _ problem first appear in these two contexts .
discrete versions of horstein s probabilistic bisection procedure @xcite were shown to be information - theoretically optimal ( optimal decay of the error probability ) in the works of zigangirov and burnashev @xcite .
more recently , the same procedure was independently proposed and analyzed in the context of noise - tolerant versions of the classic binary search problem by karp and kleinberg @xcite , which was motivated by applications ranging from investment planning to admission control in queueing networks .
closely related approaches are considered in the work of feige et al @xcite .
the noisy binary search problem has found important applications in the minimax theory of sequential , adaptive sampling procedures proposed by korostelov and kim @xcite for image recovery and binary classification problems studied by castro and nowak @xcite .
we also mention the works of rivest et al @xcite , spencer @xcite and aslam and dhagat @xcite , and dhagat et al @xcite , which consider adversarial situations in which the total number of erroneous oracle responses is fixed in advance .
one straightforward approach to noisy gbs is to follow the gbs algorithm , but to repeat the query at each step multiple times in order to decide whether the response is more probably @xmath22 or @xmath23 .
this simple approach has been studied in the context of noisy versions of classic binary search and shown to perform significantly worse than other approaches in the work of karp and kleinberg @xcite ; perhaps not surprising since this is essentially a simple repetition code approach to communicating over a noisy channel .
a near - optimal noise - tolerant version of gbs was developed in this paper .
the algorithm can be viewed as a non - trivial generalization of horstein s probabilistic bisection procedure .
horstein s method relies on the special structure of classic binary search , namely that the hypotheses and queries can be naturally ordered together in the unit interval .
horstein s method is a sequential bayesian procedure .
it begins with uniform distribution over the set of hypotheses . at each step
, it queries at the point that bisects the probability mass of the current distribution over hypotheses , and then updates the distribution according to bayes rule .
horstein s procedure is nt directly applicable to situations in which the hypotheses and queries can not be ordered togetherl , but the geometric condition developed in this paper provides similar structure that is exploited here to devise a generalized probabilistic bisection procedure .
the key elements of the procedure and the analysis of its convergence are fundamentally different from those in the classic binary search work of burnashev and zigangirov @xcite and karp and kleinberg @xcite .
this paper investigated a generalization of classic binary search , called gbs , that extends it to arbitrary query and hypothesis spaces . while the gbs algorithm is well - known , past work
has only partially characterized its capabilities .
this paper developed new conditions under which gbs ( and a noise - tolerant variant ) achieve the information - theoretically optimal query complexity .
the new conditions are based on a novel geometric relation between the query and hypothesis spaces , which is verifiable analytically and/or computationally in many cases of practical interest .
the main results are applied to learning multidimensional threshold functions , a problem arising routinely in image processing and machine learning .
let us briefly consider some possible extensions and open problems .
first recall that in noisy situations it is assumed that the binary noise probability has a known upper bound @xmath100 . it is possible to accommodate situations in which the bound is unknown a priori .
this can be accomplished using an ngbs algorithm in which the number of repetitions of each query , @xmath102 , is determined adaptively to adjust to the unknown noise level .
this procedure was developed by the author in @xcite , and is based on a straightforward , iterated application of chernoff s bound .
similar strategies have been suggested as a general approach for devising noise - tolerant learning algorithms @xcite . using an adaptive procedure for adjusting the number of repetitions of each query yields an ngbs algorithm with query complexity bound proportional to @xmath341 , the same order as that of the ngbs algorithm discussed above which assumed a known bound @xmath177 . whether or not the additional logarithmic factor can be removed
if the noise bound @xmath177 is unknown is an open question .
adversarial noise models in which total number of errors is fixed in advance , like those considered by rivest et al @xcite and spencer @xcite , are also of interest in classic binary search problems .
repeating each query multiple times and taking the majority vote of the responses , as in the ngbs algorithm , is a standard approach to adversarial noise .
thus , ngbs provides an algorithm for generalized binary search with adversarial noise .
finally , we suggest that the salient features of the gbs algorithms could be extended to handle continuous , uncountable classes of hypotheses . for example , consider the continuous class of halfspace threshold functions on @xmath191 . this class is indexed parametrically and it is possible to associate a volume measure with the class ( and subsets of it ) by introducing a measure over the parameter space . at each step of a gbs - style algorithm ,
all inconsistent hypotheses are eliminated and the next query is selected to split the volume of the parameter space corresponding to the remaining hypotheses , mimicking the splitting criterion of the gbs algorithm presented here .
+ + acknowledgements .
the author thanks r. castro , a. gupta , c. scott , a. singh and the anonymous reviewers for helpful feedback and suggestions .
first we derive the precise form of @xmath342 is derived as follows .
let @xmath343 , the weighted proportion of hypotheses that agree with @xmath344 . the factor that normalizes the updated distribution in ( [ update1 ] )
is related to @xmath345 as follows .
note that @xmath346 .
thus , @xmath347 denote the reciprocal of the update factor for @xmath348 by @xmath349 where @xmath350 , and observe that @xmath351 .
thus , @xmath352 we prove that @xmath353\leq\mm_i$ ] by showing that @xmath354 \leq 1 $ ] . to accomplish this
, we will let @xmath153 be arbitrary .
for every @xmath355 and every @xmath4 let @xmath35 denote the value of @xmath73 on the set @xmath48 .
define @xmath356 , the proportion of hypotheses that take the value @xmath22 on @xmath48 .
let @xmath251 denote that set that @xmath357 is selected from , and consider the four possible situations : @xmath358 to bound @xmath359 $ ] it is helpful to condition on @xmath251 .
define @xmath360 .
if @xmath361 , then @xmath362 & = & \frac{(1-\delta_{a_i}^+)\beta+\delta_{a_i}^+ ( 1-\beta)}{1-\beta}(1-q_i)\ + \
\frac{\delta_{a_i}^+\beta+(1-\delta_{a_i}^+ ) ( 1-\beta)}{\beta}q_i \\ & = & \delta_{a_i}^+ + ( 1-\delta_{a_i}^+)\left[\frac{\beta(1-q_i)}{1-\beta}+\frac{q_i(1-\beta)}{\beta } \right ] \
.\end{aligned}\ ] ] define @xmath363 $ ] .
similarly , if @xmath364 , then @xmath365 & = & ( 1- \delta_{a_i}^+ ) + \delta_{a_i}^+ \left[\frac{\beta(1-q_i)}{1-\beta}+\frac{q_i(1-\beta)}{\beta } \right ] \
\gamma_i^{-}(a_i ) \end{aligned}\ ] ] by assumption @xmath366 , and since @xmath367 the factor @xmath368 ( strictly less than @xmath171 if @xmath369 ) .
define @xmath370 to obtain the bounds @xmath371 observe that for every @xmath48 we have @xmath372 , since at least one hypothesis takes the value @xmath23 on @xmath48 and @xmath373 for all @xmath82 .
therefore both @xmath374 and @xmath375 are less or equal to @xmath171 , and it follows that @xmath354\leq1 $ ] ( and strictly less than @xmath171 if @xmath151 )
. @xmath77 the proof amounts to obtaining upper bounds for @xmath374 and @xmath375 , defined above in ( [ pb ] ) and ( [ nb ] ) .
consider two distinct situations .
define @xmath376 .
first suppose that there do not exist neighboring sets @xmath48 and @xmath49 with @xmath377 and @xmath378 .
then by lemma [ lemma1 ] , this implies that @xmath379 , and according the query selection step of the modified sgbs algorithm , @xmath380 .
note that because @xmath381 , @xmath382 .
hence , both @xmath374 and @xmath375 are bounded above by @xmath383 .
now suppose that there exist neighboring sets @xmath48 and @xmath49 with @xmath377 and @xmath384 .
recall that in this case @xmath251 is randomly chosen to be @xmath48 or @xmath49 with equal probability .
note that @xmath385 and @xmath386 .
if @xmath387 , then applying ( [ pb ] ) results in @xmath388 & < & \frac{1}{2}(1 + \frac{1-b_i}{2}+\frac{1+b_i}{2}(1-\varepsilon_0 ) ) \ = \ \frac{1}{2}(2-\varepsilon_0\frac{1+b_i}{2 } ) \ \leq \ 1-\varepsilon_0/4 \ , \end{aligned}\ ] ] since @xmath389 .
similarly , if @xmath390 , then ( [ nb ] ) yields @xmath391 < 1-\varepsilon_0/4 $ ] . if @xmath392 on @xmath48 and @xmath393 , then applying ( [ nb ] ) on @xmath48 and ( [ pb ] ) on @xmath49 yields @xmath388 & \leq & \frac{1}{2}\left(\delta_{a}^+ ( 1-\varepsilon_0 ) + ( 1- \delta_{a}^+ ) + \delta_{a'}^+ + ( 1-\delta_{a'}^+)(1-\varepsilon_0)\right ) \\ & = & \frac{1}{2}(1-\delta_a^+ + \delta_{a'}^+ + ( 1-\varepsilon_0)(1+\delta_a^+ -\delta_{a'}^+ ) ) \\ & = & \frac{1}{2}(2-\varepsilon_0(1+\delta_a^+-\delta_{a'}^+ ) ) \\ & = & 1-\frac{\varepsilon_0}{2 } ( 1+\delta_a^+-\delta_{a'}^+ ) \ \leq
\ 1-\varepsilon_0/2 \ , \end{aligned}\ ] ] since @xmath394 .
the final possibility is that @xmath395 and @xmath396
. apply ( [ pb ] ) on @xmath48 and ( [ nb ] ) on @xmath49 to obtain @xmath388 & \leq & \hspace{-.1 in } \frac{1}{2}\left(\delta_a^+ + ( 1-\delta_a^+)(1-\varepsilon_0)+\delta_{a'}^+ ( 1-\varepsilon_0 ) + ( 1- \delta_{a'}^+)\right ) \\ & = & \frac{1}{2}(1+\delta_a^+-\delta_{a'}^+ + ( 1-\varepsilon_0)(1-\delta_a^++\delta_{a'}^+))\end{aligned}\ ] ] next , use the fact that because @xmath48 and @xmath49 are neighbors , @xmath397 ; if @xmath398 does not belong to @xmath2 , then @xmath399 .
hence , @xmath388 & \leq & \frac{1}{2}(1+\delta_a^+-\delta_{a'}^+ + ( 1-\epsilon_0)(1-\delta_a^++\delta_{a'}^+ ) ) \\ & = & \frac{1}{2}(1+p_i(h^*)-p_i(-h^ * ) + ( 1-\epsilon_0)(1-p_i(h^*)+p_i(-h^ * ) ) ) \\ & \leq & \frac{1}{2}(1+p_i(h^ * ) + ( 1-\epsilon_0)(1-p_i(h^ * ) ) ) \ = \ 1-\frac{\varepsilon_0}{2}(1-p_i(h^*))\ , \end{aligned}\ ] ] since the bound is maximized when @xmath400 .
now bound @xmath359 $ ] by the maximum of the conditional bounds above to obtain @xmath401 & \leq & \max \left\{1-\frac{\varepsilon_0}{2}(1-p_i(h^ * ) ) \ , , \ , 1-\frac{\varepsilon_0}{4 } \ , , \ , 1-(1-\ch)\frac{\varepsilon_0}{2 } \right\ } \ , \end{aligned}\ ] ] and thus it is easy to see that @xmath402 & = & \frac{\e\left [ \gamma_i|p_i\right]- p_i(h^ * ) } { 1-p_i(h^ * ) } \ \leq \ 1-\min\left\{\frac{\varepsilon_0}{2}(1-\ch),\frac{\varepsilon_0}{4}\right\ } \ .
\hspace{.5 in } \blacksquare \end{aligned}\ ] ] first consider the bound on @xmath403 $ ] .
let @xmath404 and consider the conditional expectation @xmath405 $ ] ; i.e. , expectation with respect to the @xmath232 queries drawn from the region @xmath212 , conditioned on the @xmath406 queries used to select @xmath206 and @xmath207 . by lemma [ runoff ] @xmath407 & \leq & ( 1-\delta_n ) \ , \min\{r(h_1),r(h_2)\ } \ + \ \delta_n \ , \max\{r(h_1),r(h_2)\ } \ , \\ & = & \min\{r(h_1),r(h_2)\ }
\ + \ \delta_n \
, \left[\max\{r(h_1),r(h_2)\}-\min\{r(h_1),r(h_2)\}\right ] \ , \\ & = & \min\{r(h_1),r(h_2)\ } \ + \ \delta_n \ , |r(h_1)-r(h_2)| \ , \\ & = & \min\{r(h_1),r(h_2)\ } \ + \ 2 \ , |r(h_1)-r(h_2)| \ , e^{-n|r_\delta(h_1)-r_\delta(h_2)|^2/6 } \ , \\ & \leq & \min\{r(h_1),r(h_2)\}+ 2 |r(h_1)-r(h_2)| \ , e^{-n|r(h_1)-r(h_2)|^2/6 } \ , \end{aligned}\ ] ] where the last inequality follows from the fact that @xmath408 .
the function @xmath409 attains its maximum at @xmath410 , and therefore @xmath407 & \leq & \min\{r(h_1),r(h_2)\ } \ , + \ ,
\sqrt{3/n } \ .\end{aligned}\ ] ] now taking the expectation with respect to @xmath206 and @xmath207 ( i.e. , with respect to the queries used for the selection of @xmath206 and @xmath207 ) @xmath233 & \leq & \e[\min\{r(h_1),r(h_2)\ } ] \ , + \
, \sqrt{3/n } \ , \\ & \leq & \min\{\e[r(h_1)],\e[r(h_2)]\ } \ , + \ ,
\sqrt{3/n } \ , \end{aligned}\ ] ] by jensen s inequality .
next consider the bound on @xmath411 .
this also follows from an application of lemma [ runoff ] .
note that if the conditions of theorem [ thm3 ] hold , then @xmath412 .
furthermore , if @xmath413 and @xmath414 , then @xmath415 .
the bound on @xmath416 follows by applying the union bound to the events @xmath413 and @xmath417 .
@xmath77 a. rnyi , `` on a problem in information theory , '' _ mta mat .
_ , p. 505516
, 1961 , reprinted in _ selected papers of alfred rnyi _ , vol . 2 , p. turan , ed . , pp .
631 - 638 .
akademiai kiado , budapest , 1976 .
e. j. cands , j. romberg , and t. tao , `` robust uncertainty principles : exact signal reconstruction from highly incomplete frequency information , '' _ ieee trans .
inform . theory _
52 , no . 2 ,
489509 , feb . 2006 . | this paper investigates the problem of determining a binary - valued function through a sequence of strategically selected queries .
the focus is an algorithm called generalized binary search ( gbs ) .
gbs is a well - known greedy algorithm for determining a binary - valued function through a sequence of strategically selected queries . at each step
, a query is selected that most evenly splits the hypotheses under consideration into two disjoint subsets , a natural generalization of the idea underlying classic binary search .
this paper develops novel incoherence and geometric conditions under which gbs achieves the information - theoretically optimal query complexity ; i.e. , given a collection of @xmath0 hypotheses , gbs terminates with the correct function after no more than a constant times @xmath1 queries .
furthermore , a noise - tolerant version of gbs is developed that also achieves the optimal query complexity .
these results are applied to learning halfspaces , a problem arising routinely in image processing and machine learning . |
it is well known that spherically - symmetric sources such as stars can be modelled in the simplest way by the interior and exterior schwarzschild solutions . but to include the radiation outside a star , more complicated solutions are required.@xcite these include the vaidya metric which uses a retarded time coordinate to describe a radiating atmosphere,@xcite the metrics of herrera and coworkers wherein spheres of matter are matched to exterior spacetimes,@xcite@xmath0@xcite and the metrics of glass and krisch which extend the vaidya solution to include both a radiation field and a string fluid.@xcite recently , liu and wesson presented a new kind of solutions in which the metric is time - dependent ( but not of the vaidya form ) and the energy - momentum tensor is of the form of a perfect fluid for radiation plus a radial heat flow.@xcite clearly , this solution describes sources which are time - dependent and spherically symmetric .
however , we wish to know specifically what kind of sources the solution represents . in this paper
we will show that it can describe * exterior * fields of a contracting ( or expanding ) star immersed in a homogenous and isotropic background radiation .
campbell s theorem says that any 4d einstein solution with a source can be locally embedded in a 5d manifold _ without _ sources whose field equations in terms of the ricci tensor are @xmath1.@xcite ( here and elsewhere lower case greek letter run 0,123 and uppercase latin letters run 0 , 123 , 4 , and we use units @xmath2 . ) a major application of campbell s theorem is to study embeddings of known 4d einstein solutions in 5d ricci flat manifolds.@xcite another application of the theorem is to generate new 4d solutions from known 5d ricci flat solutions.@xcite@xmath3@xcite in the following we will show briefly how a new 4d solution is generated in ref.[7 ] .
for the purpose of convenience , some of the equations and notations in ref .
[ 7 ] will be re - expressed . in reference
[ 7 ] , liu and wesson presented a class of 5d solutions .
this class of solutions is time - dependent in 5d and spherically symmetric in 3d with the 5d metric being @xmath4 -\left ( 1-\lambda t\right ) ^{-4}dy^2\ ; , \label{5metr}\ ] ] where @xmath5 is a constant and the two functions @xmath6 and @xmath7 are determined by @xmath8 where a prime denotes derivative with respect to @xmath9 .
clearly , ( [ bprime ] ) and ( [ aprime ] ) can be solved , at least numerically , by imposing boundary conditions on @xmath10 and @xmath11 .
this 5d solution satisfies the 5d equations @xmath1 and therefore is 5d empty .
however , the 4d part of the 5d metric ( [ 5metr ] ) , together with the two equations ( [ bprime ] ) and ( [ aprime ] ) , defines a 4d solution as shown in the following:@xcite @xmath12 \quad , \label{4metr}\ ] ] @xmath13 @xmath14 where @xmath15 is an effective or induced energy - momentum tensor with @xmath16 it was also shown in ref.[7 ] that this @xmath15 can be modeled as a perfect fluid plus a radial heat flow , @xmath17 were @xmath18 is the mass density , @xmath19 is the pressure , @xmath20 is the 4-velocity , @xmath21 is the heat - flux vector , and @xmath22 and @xmath23 obey the orthogonality condition @xmath24
. then equations ( [ t00])-([t01 ] ) yield @xmath25 equations ( [ 4metr])-([heat ] ) constitute a complete set of the 4d solution , from which we see that the equation of state of the 4d fluid is @xmath26 , so it represents a radiation or extra - relativistic particles accompanied by a radial heat flow .
the induced 4d energy - momentum tensor ( [ t - heat ] ) describes a thermodynamical system in which the radial heat current @xmath23 implies a radial temperature gradient .
now we wish to calculate this temperature distribution over the fields . the generalized relativistic relation between heat current and temperature gradient
can be found in refs.[11,12 ] with @xmath27 where @xmath28 is the coefficient of the thermal conductivity and @xmath29 is the projection tensor , @xmath30 to calculate equations ( [ q - t1 ] ) , we calculate @xmath31 firstly . with use of ( [ 4metr ] )
, we find the only non - vanishing @xmath31 being @xmath32 .
furthermore , we assume @xmath33
. then equations ( [ q - t1 ] ) reduce to @xmath34 and @xmath35 combining ( [ heat ] ) and ( [ q - t2 ] ) , we get @xmath36 we find that this equation can be integrated , giving an exact solution being @xmath37 where @xmath38 is a constant radius , and @xmath39 and @xmath40 are values of @xmath10 and @xmath41 at @xmath42 , respectively . from ( [ heat ] ) we see that @xmath43 corresponds to thermal equilibrium @xmath44
. then , from ( [ t - solu ] ) , if @xmath45 we get @xmath46constant .
thus we recover the conclusion that thermal equilibrium corresponds not to constant temperature , but to the redshifted temperature distribution @xmath47constant.@xcite generally we have @xmath48 and the exact solution ( [ t - solu ] ) determines both the space distribution and the time variation of the temperature @xmath49 over the fields .
the close - to - schwarzschild approximation of the solution ( [ 4metr ] ) was given in ref.[7 ] .
we find that it can be re - expressed in the following form : @xmath50 \quad , \label{clos - to1 } \\ b & = & 1-\frac{2gm}r+2\lambda ^2r^2+o\left ( \varepsilon ^3\right ) \quad , \label{clos - to2 } \\ a^{-1 } & = & 1-\frac{2gm}r-\lambda ^2r^2+o\left ( \varepsilon ^3\right ) \quad \;. \label{clos - to3}\end{aligned}\ ] ] here @xmath51 is a small quantity of the order of the newtonian potential @xmath52 , @xmath53 are terms of the order of @xmath54 or higher , and we have assumed , for practical usage , that @xmath55 is also a small quantity of the order @xmath51 , i.e. , @xmath56 to verify the accuracy of this solution , one can calculate @xmath57 and @xmath58 firstly , and then substitute them into the two equations ( [ bprime ] ) and ( [ aprime ] ) .
thus the approximate solutions ( [ clos - to1])-([clos - to3 ] ) are correct in the range ( [ small ] ) , or , equivalently , in @xmath59 we also conclude that ( [ clos - to1])-([clos - to3 ] ) are accurate up to the second order of @xmath51 and give back to the schwarzschild solution if @xmath45 .
so generally we can interpret the solution as an exterior solution of a star - like source .
we will show , in the next section , that in a wide range in our solar system the condition ( [ small ] ) , or , equivalently , ( [ r - range ] ) , is satisfied .
consider now the equation ( [ rho ] ) , which , by ( [ clos - to2 ] ) , reduces to @xmath60 \quad .
\label{rho - appr}\ ] ] here @xmath18 and @xmath19 constitute a perfect fluid with the equation of state @xmath26 , implying a property for radiation or ultra - relativistic particles . from ( [ rho - appr ] )
we see that neglecting higher - order terms in the square bracket in the rhs of ( [ rho - appr ] ) , the densities @xmath18 and @xmath19 are homogenous and isotropic .
thus we find that one can not interpret the fluid as a radiating atmosphere of a star such as in the vaidya metric.@xcite apparently , we can interpret it as to describe a star immersed in a homogenous and isotropic background radiation .
the existence of the heat flow term in the energy - momentum tensor ( [ t - heat ] ) implies that there must be a temperature gradient in the field and a heat interchanges between the star and the background radiation .
using ( [ clos - to2 ] ) and ( [ clos - to3 ] ) in ( [ t - solu ] ) gives @xmath61 \quad .
\label{t - appr}\ ] ] if we choose @xmath38 as the radius of the star , then in the exterior @xmath62 , equation ( [ t - appr ] ) implies that if @xmath63 then @xmath64 .
be aware that the thermal equilibrium corresponds to @xmath65 for which there is no heat flow.@xcite so we conclude that the heat flows outwards if @xmath63 and inwards if @xmath66 .
this agree with the original relation ( [ heat ] ) in which @xmath67 for @xmath63 and @xmath68 for @xmath66 since @xmath69 according to ( [ clos - to2 ] ) .
meanwhile , from the metric ( [ clos - to1 ] ) and the result ( [ rho - appr ] ) we also see that if @xmath70 , then as the time @xmath71 increases , the 3d space contracts and the energy density of the outside fluid increases .
all these properties are physically reasonable .
as we have concluded in section 4 that the 4d solution discussed in this paper can be interpreted as an exterior solution of a spherical source such as a star which has a non - zero temperature and is immersed in a homogenous and isotropic background radiation . a natural candidate for this kind of radiation
is the cosmic background radiation , for which the temperature is about @xmath72 at present days with an energy density around @xmath73 @xmath74 .
thus , by using ( [ rho ] ) , we determine the constant @xmath5 as @xmath75 so @xmath5 is of the order of the hubble constant .
now we wish to know if the close - to - schwarzschild solution given in ( [ clos - to1])-([clos - to3 ] ) can describe our solar system .
that is , we need to calculate and compare orders of the two terms @xmath52 and @xmath76 appeared in ( [ clos - to2 ] ) and ( [ clos - to3 ] ) .
the average distance between the sun and the nearest planet mercury is about @xmath77 au .
so we have @xmath78 the average distance between the sun and the pluto is about @xmath79 au .
so @xmath80 therefore we conclude that in a wide range in our solar system , we have @xmath81 .
so @xmath76 is of the order higher than the post - newtonian order .
this implies that the contributions of the cosmic background radiation to all of the known solar system experiments@xcite are negligible .
however , there may have other ways to detect possible new effects of the solution .
for example , according to the time - dependent metric ( [ clos - to1 ] ) and the value of @xmath5 in ( [ lambda ] ) , the radius of the central star should be contracting or expanding , depending on whether @xmath5 is positive or negative , with a relative rate at present time being @xmath82 it is worth to study whether this kind of effect is observable .
we thank paul wesson , mark roberts and guowen peng for comments . this work was supported by nsf of p. r. china under grant 19975007 . p.s .
wesson , _ space , time , matter _ , world scientific , singapore ( 1999 ) ; j.m .
overduin and p.s .
wesson , phys .
rep.*283 * , 303 ( 1997 ) , gr - qc/9805018 ; j. ponce de leon , gen .
. grav . * 20 * , 539 ( 1988 ) ; h. liu , b. mashhoon , ann . phys .
( leipzig ) * 4 * , 565 ( 1995 ) ; j.e .
lidsey , c. romero , r. tavakol , and s. rippl , class .
* 14 * , 865 ( 1997 ) , gr - qc/9907040 ; h. liu and p.s .
wesson , astrophys .
j. * 562 * , 1 ( 2001 ) , gr - qc/0107093 ; s.s .
seahra and p.s .
wesson , class .
* 19 * , 1139 ( 2002 ) , gr - qc/0202010 . | we study a time - dependent and spherically - symmetric solution with a star - like source .
we show that this solution can be interpreted as an exterior solution of a contracting star which has a decreasing temperature and is immersed in a homogenous and isotropic background radiation .
distribution of the temperature in the fields and close - to schwarzschild approximation of the solution are studied . by identifying the radiation with the cosmic background one , we find that the close - to - schwarzschild approximate solution is valid in a wide range in our solar system .
possible experimental tests of the solution are discussed briefly .
pacs number(s ) : 04.20.-q , 04.50.+h |
time - delay systems have attracted a lot of attention in recent years , in part due to the fact that multistability , i.e. the coexistence of multiple attractors , is a common occurrence when the delays are large typically , much larger than the response time of the system @xcite .
interest in multistability arises because multistable systems play a key role in pattern recognition processes @xcite and memory storage devices . by choosing appropriate initial conditions , prescribed periodic solutions
can be stored as oscillatory patterns of a time - delay system @xcite .
synchronization of chaotic time - delay systems has also received attention , since it has potential applications to secure communications @xcite .
perez and cerdeira @xcite have shown that , in low - dimensional chaotic systems , a hidden message can be unmasked by the dynamical reconstruction of the chaotic signal using nonlinear dynamical methods .
encrypting a message in the chaotic output of a time - delay system has the advantage that the dynamics is in this case high - dimensional ( the dimension increases linearly with the delay @xcite ) but , in spite of this fact , synchronization can be achieved by transmitting a single scalar signal .
unfortunately , this method is not as secure as initially expected , since it has been shown that by using a special embedding space , the delay time can be identified , and the message can be successfully unmasked @xcite .
coupled oscillators with time delays in the coupling , which represent interactions being transmitted at finite speed , have been extensively studied as well @xcite .
recently , a new effect of delayed coupling was reported by voss @xcite , who showed the existence of an anticipating synchronization regime . in this regime ,
the slave system becomes synchronized to the chaotic future state of the master system .
anticipation occurs when the coupling is delayed , and results from the interplay of memory effects and relaxation mechanisms .
numerically , this regime was found in coupled semiconductor lasers with optical feedback @xcite . in this paper
we investigate the existence and stability of anticipated synchronization in coupled time - delay maps , which have the advantage of allowing for analytical calculations . in the next section ,
we introduce a system of two time - delay coupled maps in a master - slave configuration , and present analytical results on the stability of anticipated and retarded synchronization for generic maps . in section 3 , we apply the results to a delay map that arises from the discretization of the ikeda delay - differential equation .
section 4 presents a summary and the conclusions .
we consider a one - dimensional map of the form @xmath3 for @xmath4 , the first term in the right - hand side represents a relaxation mechanism . under its sole action
@xmath5 would asymptotically vanish .
this relaxation , however , competes with the effect of the nonlinear function @xmath6 , which has the form of a time - delayed feedback with delay @xmath0 . the ( master ) map ( [ x ] ) is used to partially drive the evolution of a new ( slave ) system @xmath7 .
the dynamics of this system is , in principle , the same as for @xmath5 , except that a part of the nonlinear component is replaced by the evolution of @xmath5 , namely @xmath8 the parameter @xmath9 $ ] measures the strength of the driving action .
note that @xmath5 enters the dynamics of @xmath7 with a delay @xmath1 .
this form of unidirectional time - delay coupling is not only an extension to maps of the configuration proposed by voss for continuous - time systems @xcite , but also generalizes the master - slave interaction by introducing a parameter ( @xmath10 ) that controls its strength .
voss s configuration corresponds to the extreme value @xmath11 . the structure of eqs .
( [ x ] ) and ( [ y ] ) suggests that , if driving is effective enough , @xmath7 may become synchronized to the driving signal , in the form @xmath12 if this synchronization is realized , the dynamics of @xmath7 coincides with that of @xmath5 up to a time shift @xmath2 . if @xmath13 the slave system anticipates the evolution of the master system , and we have anticipated synchronization . on the other hand , if @xmath14 the synchronization is retarded .
linear stability of the synchronized state ( [ sync ] ) can be analyzed in the usual way .
taking @xmath15 , the evolution of the deviation @xmath16 in the limit of infinitesimal deviations is readily derived from ( [ y ] ) : @xmath17 with @xmath18 .
note that , if @xmath4 and for @xmath11 , the deviation decreases exponentially with time and synchronization is therefore stable . in view of well - known properties of synchronization in coupled maps without delays , it is possible to advance that , if the dynamics of @xmath5 is chaotic , the synchronized state ( [ sync ] ) will be stable above a certain threshold @xmath19 . on the other hand ,
if @xmath5 is nonchaotic , synchronization should be stable for any @xmath20 . for @xmath21 , eq .
( [ delta ] ) can be formally integrated by introducing a linear @xmath22-dimensional map for a variable @xmath23 , with @xmath24 .
this equivalent map is given by @xmath25 where the elements of the matrix @xmath26 read @xmath27 the deviations given by eq .
( [ delta ] ) can then be obtained from the solution of the equivalent map , @xmath28 thus , synchronization is linearly stable if all the eigenvalues of the evolution matrix @xmath29 vanish for @xmath30 . for @xmath11
the eigenvalues are @xmath31 . for @xmath32 ,
on the other hand , @xmath29 and its eigenvalues must be calculated numerically , since they involve the successive states of the variable @xmath5 , cf .
( [ mat ] ) .
it is interesting to point out that stability is determined by the asymptotic properties of @xmath29 in the limit @xmath33 , where the driving system @xmath5 has exhaustively explored the accessible region of phase space . in this limit , thus , the delay @xmath1 of eq .
( [ y ] ) plays no role in the determination the eigenvalues of @xmath29 . in consequence
, @xmath1 is irrelevant to the stability of the synchronized state .
in this section we present numerical results corresponding to maps ( [ x ] ) and ( [ y ] ) , with @xmath34 . with this choice of @xmath6 , we refer to eq .
( [ x ] ) as the ikeda delay map , since it arises from the discretization of the ikeda delay - differential equation @xcite , @xmath35 this equation was introduced to describe the dynamics of an optical bistable resonator , where the finiteness of the speed of light is relevant , and therefore require a description that incorporates explicitly the round - trip time of light in an optical cavity .
physically , @xmath36 is the phase lag of the electric field across the resonator , @xmath37 is the laser intensity injected into the system , and @xmath38 is the round trip time of the light in the resonator . writing
@xmath39/ \delta t$ ] , @xmath40 , and @xmath41 , the ikeda delay map is obtained from eq .
( [ ik ] ) with @xmath42 , @xmath43 , and @xmath44 .
the ikeda delay map is not to be confused with the discrete ikeda map @xcite , which is obtained from ( [ ik ] ) by discretizing time in units of @xmath38 in the singular limit where the delay - to - response time ratio diverges . in the following ,
we focus the attention on the case @xmath45 .
our numerical calculations are restricted to random initial condition in @xmath46 both for the master and the slave system . in fig .
[ f1 ] we illustrate anticipated synchronization , for @xmath47 and @xmath48 .
the slave system ( dashed line ) anticipates in @xmath49 steps the state of the master system ( solid line ) . after a transient
, the difference @xmath50 ( dotted line ) decays to zero .
we have verified that this behavior is independent of the value of @xmath1 .
the critical value @xmath51 of the coupling strength , above which the synchronized state is stable , has been calculated from the analytical results of the previous section by means of a numerical evaluation of the matrix @xmath29 in eq .
( [ u ] ) .
since the delay @xmath1 is irrelevant to this calculation , we can take @xmath48 in eq .
( [ mat ] ) . from a given initial condition for the master system ,
the matrix @xmath29 is evaluated for successive values of @xmath52 . in order to avoid the calculation of its eigenvalues , which is specially troublesome for large @xmath0 ,
@xmath29 is multiplied at each step by a randomly generated vector , @xmath53 , of unitary modulus . if , for a given value of @xmath10 , the modulus of the product remains above a certain upper threshold @xmath54 for sufficiently large @xmath52 , @xmath55 for @xmath56 , synchronization is considered unstable and @xmath57 . if , on the other hand , @xmath58 for @xmath56 , where @xmath59 is a suitable lower threshold , synchronization is considered stable and @xmath60 . with this criterion ,
@xmath51 can be found by decimation within the interval @xmath61 up to a certain previously fixed precision . in our calculations ,
we have taken @xmath62 , @xmath63 , and @xmath64 .
figure [ f2 ] shows @xmath51 as a function of the delay @xmath0 and for several values of the parameter @xmath65 .
note that for small @xmath0 specifically , for @xmath66 the critical value @xmath51 can vanish , which indicates that the master evolution is nonchaotic .
notice , furthermore , that for some values of @xmath65 and @xmath66 we have plotted more than one value of @xmath51 .
this is due to the fact that , in such cases , the master system is multistable with , typically , two chaotic attractors .
the value of @xmath51 depends thus on the initial conditions for both @xmath5 and @xmath7 .
whereas the behavior is quite irregular for small @xmath0 , for larger delays the critical coupling becomes practically independent of @xmath0 .
though it has not been possible to prove this analytically , we conjecture that @xmath51 approaches a finite limit below unity as @xmath67 .
the dependence on @xmath65 in the investigated interval is also moderately weak . as expected , since the dynamics of the ikeda delay map becomes more irregular as @xmath65 grows i.e .
the lyapunov exponent is higher @xmath51 increases accordingly .
the degree of anticipated of retarded synchronization can be quantified by calculating the similarity function @xmath68 , defined as @xcite : @xmath69 ^ 2\rangle } \over { [ \langle x_n^2\rangle \langle y_n^2\rangle ] ^{1/2 } } } .\ ] ] if @xmath5 and @xmath7 are independent time series with similar mean value and dispersion , the average of their square difference is of order @xmath70 , and thus @xmath71 . if , on the other hand , there is perfect anticipated or retarded synchronization , the difference @xmath72 vanishes , and @xmath73 . in fig .
[ f3 ] , the similarity function is shown for the same parameters as in fig . [ f1 ] .
figure [ f4 ] shows @xmath74 i.e .
the minimum value of @xmath75 recorded during sufficiently long realizations of the evolution for a given set of parameters in the @xmath76-plane , with all the other parameters as in fig .
the region where anticipated synchronization occurs , @xmath77 , is clearly visible for large values of @xmath10 .
the position of its boundary is in good qualitative agreement with the values of @xmath51 shown in fig .
[ f2 ] for @xmath78 .
we have extended the study of anticipated synchronization , advanced by voss for unidirectionally coupled differential equations with time delays @xcite , to delayed coupled chaotic maps . while the nature of anticipated synchronization of maps and differential equations is the same , delay discrete - time dynamics admits an analytical treatment which can not be carried out for continuous - time systems .
in fact , ordinary differential equations with finite time delays constitute an infinite - dimensional problem . on the other hand ,
since times delays in maps must be discrete , the dimensionality of the problem remains finite .
taking advantage of this situation , we have analytically studied the stability of anticipated and retarded synchronization in a generic master - slave configuration . in the absence of coupling , master and slave dynamics
are identical and involve an intrinsic delay @xmath0 . coupling consists in the replacement of a part of the slave dynamics by that of the master system , with a delay @xmath1 .
we have shown that the stability of synchronization is independent of @xmath1 .
the structure of the linearized problem , eqs .
( [ delta]-[u ] ) , suggest meanwhile a strong though not transparent dependence on the lyapunov exponent of the master system , as expected . in practice
, the linearized problem has to be treated numerically , but it only involves the realization of the master system and the successive application of the linear - evolution operator , being thus a purely algebraic process .
these results have been applied to the ikeda delay map , which derives from the application of the euler integration scheme to the ikeda delay - differential equation .
we have calculated the critical coupling intensity above which synchronization is stable , as a function of the delay @xmath0 and of a parameter @xmath65 that controls the chaotic dynamics of the map .
it is found that , whereas for small values of @xmath0 the critical coupling can vary considerably due to the irregular appearance and disappearance of chaotic and nonchaotic attractors the dependence for large @xmath0 is much smoother . in fact , as @xmath0 grows , the critical coupling seems to approach a constant value .
the dependence with the dynamical parameter @xmath65 is also moderate in the considered range . in this work
we have focused the attention on exact anticipated synchronization .
however , it has been previously shown that approximate anticipated synchronization is possible in coupled differential equations , even in the absence of intrinsic delays @xcite .
the study of approximate anticipated synchronization in coupled maps constitutes therefore a line for future work . in particular , it would be interesting to investigate in detail the connection between the degree of synchronization , and the irregularity of the dynamics as measured by the lyapunov exponents . for delay differential equations , the number of positive lyapunov exponents and the fractal dimension increase linearly with the delay , while the metric entropy remains roughly constant @xcite .
we therefore conjecture that the metric entropy might be a good indicator of the possibility of synchronizing with anticipation , and thus predicting , chaotic dynamics .
this work was supported by proyecto de desarrollo de ciencias bsicas ( pedeciba ) and by comisin sectorial de investigacin cientfica ( csic ) , uruguay . | we study the synchronization of two chaotic maps with unidirectional ( master - slave ) coupling .
both maps have an intrinsic delay @xmath0 , and coupling acts with a delay @xmath1 .
depending on the sign of the difference @xmath2 , the slave map can synchronize to a future or a past state of the master system .
the stability properties of the synchronized state are studied analytically , and we find that they are independent of the coupling delay @xmath1 . these results are compared with numerical simulations of a delayed map that arises from discretization of the ikeda delay - differential equation .
we show that the critical value of the coupling strength above which synchronization is stable becomes independent of the delay @xmath0 for large delays . , chaos synchronization , time - delayed systems 05.45.xt , 05.65.+b |
the successful deployment of the _ hubble space telescope ( hst ) _ has opened many windows for the study of the high redshift universe . in the spectroscopic arena , the _ hst _ faint object spectrograph ( fos ) observation of the luminous radio - quiet qso hs 1700 + 6416 ( @xmath1 = 2.72 ) by reimers et al .
( 1992 ) has provided an extraordinary spectrum which demonstrates the potential of the _ hst _ for the study of high @xmath14 qso absorption lines .
this fos spectrum has yielded results on topics about which little or nothing was known before the launch of the _ hst _ , including ( 1 ) the helium content of qso absorbers at @xmath15 1.84 , and ( 2 ) an ultra - highly ionized phase of qso absorbers traced by species such as ( see also hamann et al .
1995 ) , which may reveal the presence of gas with @xmath16 k ( verner , tytler , & barthel 1994 ) .
these new topics are observationally challenging because the resonance lines of , , and the ultra - highly ionized species ( e.g. , , , and ) all occur at rest wavelengths @xmath17 912 , and only a handful of qsos are currently known with continuum flux detected at @xmath18 912 ( c.f .
beaver et al .
1991 ; jakobsen et al .
1994 ; 6.1 in tripp , bechtold , & green 1994 ; reimers et al .
1995b ; lyons et al .
1995 ; tytler et al .
the continua of background qsos with sufficient redshift to show these absorption lines are usually undetectable due to attenuation by intervening absorption , either optically thick lyman limit ( ll ) absorption or the cumulative effect of many absorbers along the line of sight ( c.f . ,
mller & jakobsen 1990 ; picard & jakobsen 1993 ) .
the hs 1700 + 6416 sight line has its share of ll absorbers seven ll systems are well - detected in the reimers et al
. fos spectrum but remarkably all of the lls are optically thin ( @xmath19 1 ) and the quasar continuum is detected from @xmath51200 to 3250 , which corresponds to @xmath5320 to 870 in the qso restframe .
in addition to the ground - breaking _ hst _ spectrum , this sight line is intriguing because a deep rosat pspc image obtained by reimers et al .
( 1995a ) shows two extended x - ray sources within a few arcminutes of the qso .
both of these x - ray objects are flagged as greater than 3@xmath20 extended sources by the sharc survey ( see romer et al .
1996a , 1996b ) .
one of these x - ray sources is abell 2246 , a galaxy cluster at @xmath14 = 0.225 .
the other extended source is less easily identified and may be a higher redshift galaxy cluster .
the close proximity of these x - ray objects to the sight line offers the tantalizing possibility of studying the detailed properties of galaxy intracluster media detected in absorption in the spectrum of hs 1700 + 6416 .
hs 1700 + 6416 is also an excellent target for ground - based studies of ll absorbers .
since the ll systems in the _ hst _ spectrum are all optically thin , the lyman continuum optical depth can be measured , and this provides an accurate estimate of the column density .
previous studies of ll absorbers ( e.g. , steidel 1990 ) have mostly used samples of optically thick ll systems and have been forced to rely on elaborate methods to estimate @xmath21 ( ) . to study the abundances and physical conditions in ll systems , and to complement space - based programs , we have obtained a high resolution ( fwhm = 20 ) high signal - to - noise optical spectrum of hs 1700 + 6416 , and in this paper we present the data and analyses of the absorption systems . the paper is organized as follows .
we briefly review the observations and data reduction in 2 and the line selection and column density measurements in 3 . in
4 we provide some comments on the individual absorption systems including assessments of absorption line saturation . in this section
we also comment on a dense cluster of absorbers at 2.432 @xmath3 @xmath3 2.441 in which the weak lines of some absorbers are well - aligned with the strong lines of different doublets . in 5 we discuss the physical conditions and abundances in the absorption systems . in
6 we examine the number of absorbers detected per unit redshift including weak absorbers .
after some comments on the abundances derived by vogel & reimers ( 1995 ) from the low resolution fos spectrum in 7 , we summarize the paper in 8 .
the observations and data reduction techniques are fully described in tripp et al .
( 1996 ) ; we summarize a few of the pertinent details here .
hs 1700 + 6416 was observed during darktime ( 1820 may 1993 ) with the echelle spectrograph and the t2 kb 2048@xmath222048 ccd on the 4 m telescope at kitt peak national observatory ( kpno ) . with a 1@xmath23 entrance aperture ,
the instrument setup provided a resolution of 20 with @xmath52 pixels / fwhm and spectral coverage from 4300 to 8350 with no gaps between echelle orders .
five one - hour observations were obtained over the course of three nights .
the preliminary data reduction ( overscan and bias subtraction , flatfielding , etc . ) was done with iraf following the standard procedure .
the `` optimal '' spectrum extraction routine described by horne ( 1986 ; see also robertson 1986 ) was used to extract the final spectra from the individual observations , and then the individual spectra were rebinned to the same wavelength scale and coadded .
in addition to the flux spectrum , the optimal extraction program yields a 1@xmath20 error spectrum which includes the contribution from read noise as well as statistical noise , and this error spectrum was propagated throughtout the data reduction and analysis procedures .
the coadded spectrum was normalized by fitting cubic splines to regions free of absorption lines .
overlapping regions of adjacent orders were weighted by their signal - to - noise ratios and coadded .
the final normalized spectrum of hs 1700 + 6416 is shown in figure 1 .
the signal - to - noise ratio of the final spectrum ranges from @xmath530 to 70 per resolution element at @xmath24 4550 ( the s / n is somewhat lower within the forest ) .
absorption lines in the hs 1700 + 6416 spectrum were selected for analysis based on the statistical significance of the observed equivalent width @xmath25 .
absorption line equivalent widths were measured using the automated method described in tripp et al .
telluric absorption lines were removed from the line list by comparing the normalized spectrum of hs 1700 + 6416 to the normalized spectrum of hs 1946 + 7658 obtained during the same observing run . since the hs 1700 + 6416 and hs 1946 + 7658 spectra have similar s / n levels and the same resolution , atmospheric absorption lines are effectively removed in a ratio of the two spectra ( see figure 3 in tripp et al .
we have used this ratio of the hs 1700 + 6416 spectrum to the hs 1946 + 7658 spectrum to identify telluric lines _ only _ ; for all other measurements reported in this paper we use the coadded spectrum shown in figure 1 ( which has not been divided by the hs 1946 + 7658 spectrum ) .
table 1 lists all statistically significant ( @xmath26 ) extragalactic absorption lines detected in the kpno hs 1700 + 6416 spectrum . table 1 also lists the heliocentric vacuum wavelength of the centroid of each absorption line and a line number which identifies the absorption line in figure 1 .
identifications of the line species derived from the morton ( 1991 ) finding list are tabulated in column 4 of table 1 , and table 2 summarizes by redshift the heavy element systems we detect in the spectrum of this quasar . we have detected 13 heavy element absorbers in the direction of this quasar , 6 of which also produce detectable lyman limits in the fos spectrum of hs 1700 + 6416 obtained by reimers et al .
there are several absorption features in table 1 for which we could not find positive identifications ( labeled `` uid '' in column 4 of table 1 ) .
most of these unidentified lines are weak ( @xmath27 ) , so they could be the 1548.2 line with the corresponding 1550.8 line falling below our detection threshold . similarly , some of these uids could be the stronger line in the doublet with the weaker line undetected .
cccllclc 1 & 4404.79 & [email protected] & 1548.2 & 1.84515 & [email protected] & [email protected] & @xmath1514.32 2 & 4412.26 & [email protected] & 1550.8 & 1.84515 & [email protected] & [email protected] & [email protected] 3 & 4599.17 & [email protected] & 1238.8 & 2.71248 & [email protected] & [email protected] & [email protected] 4 & 4613.85 & [email protected] & 1242.8 & 2.71248 & [email protected] & [email protected] & [email protected] 5 & 4620.93 & [email protected] & 1393.8 & 2.31457 & [email protected] & [email protected] & [email protected] & & & & 2.31499 & [email protected] & [email protected] & [email protected] & & & & 2.31550 & [email protected] & [email protected] & [email protected] & & & & 2.31642 & [email protected] & [email protected] & [email protected] 6 & 4631.32 & [email protected] & uid & & & & 7 & 4650.72 & [email protected] & 1402.8 & 2.31457 & [email protected] & [email protected] & & & & & 2.31499 & [email protected] & [email protected] & [email protected] & & & & 2.31550 & [email protected] & [email protected] & [email protected] & & & & 2.31642 & [email protected] & [email protected] & [email protected] 8 & 4679.89 & [email protected] & uid & & & & 9 & 4752.37 & [email protected] & uid & & & & 10 & 4785.28 & [email protected] & 1393.8 & 2.43339 & [email protected] & [email protected] & [email protected] 11 & 4815.90 & [email protected] & 2796.4 & 0.72219 & [email protected] & [email protected] & [email protected] 12 & 4828.16 & [email protected] & 2803.5 & 0.72219 & [email protected] & [email protected] & [email protected] 13 & 4842.45 & [email protected] & uid & & & & 14 & 4867.35 & [email protected] & uid & & & & 15 & 4904.41 & [email protected] & 1548.2 & 2.16797 & [email protected] & [email protected] & [email protected] 16 & 4912.92 & [email protected] & 1550.8 & 2.16797 & [email protected] & [email protected] & [email protected] 17 & 4952.66 & [email protected] & uid & & & & 18 & 5061.03 & [email protected] & 1526.7 & 2.31499 & [email protected] * & [email protected] & [email protected] 19 & 5061.87 & [email protected] & 1526.7 & 2.31553 & & & [email protected] 20 & 5105.72 & [email protected] & uid & & & & 21 & 5121.89 & [email protected] & 1548.2 & 2.30812 & [email protected] & [email protected] & & & & uid & & & & 22 & 5130.23 & [email protected] & 1550.8 & 2.30812 & [email protected] & [email protected] & 23 & 5133.14 & [email protected] & 1548.2 & 2.31458 & [email protected] & [email protected] & & & & 2.31505 & [email protected] & [email protected] & & & & 2.31568 & [email protected] & [email protected] & & & & 2.31638 & [email protected] & [email protected] & [ 0pt ] } @xmath1514.74 24 & 5141.71 & [email protected] & 1550.2 & 2.31458 & [email protected] & [email protected] & & & & 2.31505 & [email protected] & [email protected] & & & & 2.31568 & [email protected] & [email protected] & & & & 2.31638 & [email protected] & [email protected] & [ 0pt ] } @xmath1514.93 25 & 5213.84 & [email protected] & 2796.4 & 0.86431 & [email protected] & [email protected] & & & & uid & & & & 26 & 5232.72 & [email protected] & 1548.2 & 2.37995 & [email protected] & [email protected] & [email protected] 27 & 5241.48 & [email protected] & 1550.8 & 2.37995 & [email protected] & [email protected] & [email protected] 28 & 5313.49 & [email protected] & 1548.2 & 2.43204 & [email protected] & [email protected] & [email protected] 29 & 5315.29 & [email protected] & 1548.2 & 2.43306 & [email protected] & [email protected] & & & & 2.43329 & [email protected] & [email protected] & [ 0pt ] } [email protected] 30 & 5323.70 & [email protected] & 1548.2 & 2.43785 & [email protected] & [email protected] & & & & & 2.43877 & [email protected] & [email protected] & & & & 1550.8 & 2.43204 & [email protected] & [email protected] & & & & & 2.43306 & [email protected] & [email protected] & & & & & 2.43329 & [email protected] & [email protected] & 31 & 5326.70 & [email protected] & 1548.2 & 2.44059 & [email protected] & [email protected] & [email protected] 32 & 5332.45 & [email protected] & 1550.8 & 2.43785 & [email protected] & [email protected] & 12.70@xmath28 & & & & 2.43877 & [email protected] & [email protected] & [email protected] 33 & 5335.48 & [email protected] & 1550.8 & 2.44059 & [email protected] & [email protected] & [email protected] 34 & 5524.54 & [email protected] & uid & & & & 35 & 5538.74 & [email protected] & 1670.8 & 2.31498 & [email protected] * & [email protected] & [email protected] 36 & 5540.05 & [email protected] & 1548.2 & 2.57813 & [email protected] & [email protected] & & & & 2.57884 & [email protected] & [email protected] & [ 0pt ] } [email protected] 37 & 5549.41 & [email protected] & 1550.8 & 2.57813 & [email protected] & [email protected] & & & & 2.57884 & [email protected] & [email protected] & [ 0pt ] } [email protected] 38 & 5747.49 & [email protected] & 1548.2 & 2.71245 & [email protected] & [email protected] & [email protected] 39 & 5757.38 & [email protected] & 1550.8 & 2.71245 & [email protected] & [email protected] & [email protected] 40 & 6032.45 & [email protected] & 2796.4 & 1.15729 & [email protected] & [email protected] & [email protected] 41 & 6047.98 & [email protected] & 2803.5 & 1.15729 & [email protected] & [email protected] & [email protected] 42 & 6148.36 & [email protected] & 1854.7 & 2.31499 & [email protected] * & [email protected] & [email protected] 43 & 6149.34 & [email protected] & 1854.7 & 2.31551 & [email protected] * & [email protected] & [email protected] 44 & 7955.79 & [email protected] & 2796.4 & 1.84506 & [email protected] & [email protected] & [email protected] 45 & 7976.21 & [email protected] & 2803.5 & 1.84506 & [email protected] & [email protected] & [email protected] column densities of identified absorption lines were measured using two techniques , the apparent column density method and profile fitting . once again , these measurements are discussed in some detail in tripp et al .
( 1996 ) . in the apparent column density method ,
the `` apparent '' ( i.e. , observed ) optical depth per unit velocity , @xmath29 , is used to calculate the apparent column density as a function of velocity according to the following relation : @xmath30 where @xmath31 is the oscillator strength and @xmath32 is the wavelength of the transition ( in ) , and the other symbols have their usual meanings .
the apparent optical depth is determined from the observed normalized absorption line profile , @xmath33 , according to the usual relation , @xmath29 = ln[1/@xmath33 ] . if the line profile is fully resolved
, then provides a good estimate of the true column density per unit velocity .
even if the profile is not well - resolved , still provides a good representation of the true column density , broadened by the spectral spread function , _ if _ the profile is not affected by unresolved saturation . however
, if the profile contains unresolved saturated absorption , then will be an underestimate of the true column density .
if two or more resonance lines of a species differing in the quantity @xmath35 are available , then comparison of the @xmath36 profiles of the different lines can be used to search for unresolved saturation .
if the absorption profile contains unresolved saturated components , then the of stronger lines will be less than the of weaker lines at velocities where the saturation occurs .
if there is no unresolved saturation , the resonance lines will yield the same . finally , if the absorption profile is not affected by unresolved saturated components , then @xmath36 can be integrated to obtain a good measurement of the total column density , @xmath21(total ) = @xmath37 = @xmath38 .
the error in due to uncertainty of the continuum placement is estimated as described in tripp et al .
( 1996 ) , and this error is added in quadrature to the error from photon counting and read noise to get the total uncertainty in .
integrated apparent column densities of all metals detected in the hs 1700 + 6416 spectrum are given in column 8 of table 1 .
for further information on the apparent column density method , see savage & sembach ( 1991 ) , jenkins ( 1996 ) , and references therein . for most of the lines in table 1 we use oscillator strengths and wavelengths from morton ( 1991 ) . however , for some transitions we use the oscillator strength revisions summarized in tripp et al .
( 1996 ) .
we have also used the profile fitting routine described by lanzetta & bowen ( 1992 ) to estimate column densities by fitting voigt profiles to the hs 1700 + 6416 absorption lines , and we refer the reader to their paper for a description of the minimized @xmath39 profile fitting procedure . for the purposes of this paper ,
we have fitted different lines from the same ion ( e.g. , 1548.2 and 1550.8 ) simultaneously , but we have fitted different elements and different ionization stages independently .
the column densities , doppler parameters , and redshifts obtained by fitting voigt profiles in this fashion are listed in columns 5 , 6 , and 7 of table 1 .
voigt profile fitting corrects for saturated absorption to some degree , but direct integration of does not correct for saturation effects unless special efforts are made ( savage & sembach 1991 ; jenkins 1996 ) .
therefore for a specific line , comparison of the column density derived from profile fitting to the column density given by direct integration of can also be used to assess the impact of unresolved saturated absorption .
if the column density from profile fitting is significantly larger than the integrated apparent column density , then the absorption line may be affected by unresolved saturation .
in the hs 1700 + 6416 discovery paper , reimers et al .
( 1989 ) identified four metal absorption systems at @xmath40 = 2.308 , 2.315 , 2.433 , and 2.440 on the basis of lines detected in a high resolution but modest s / n optical spectrum .
subsequently , reimers et al .
( 1992 ) reobserved the quasar with the _ hst _ fos and detected a wide range of ionization stages of many heavy elements at these redshifts .
they also reported detections of metals at @xmath40 = 0.8642 , 1.1572 , 1.725 , 1.8465 , 2.1678 , and 2.579 .
reimers et al .
( 1992 ) detected lyman continuum absorption in seven of these systems , and all of the lyman limits have moderate optical depth ( @xmath41 1 ) .
this is important because with detection of flux shortward of the lyman limit edge , the lyman continuum optical depth @xmath42 at wavelength @xmath43 912 can be reliably measured .
since the cross section for continuum absorption is well known , it is then straightforward to estimate the column density , @xmath44 ^{3}.\ ] ] because of the low resolution of the fos spectrum , this provides a measurement of the _ total _ column density in the lyman limit system and no information about the kinematics of the absorber .
more recently , hs 1700 + 6416 has been observed from the ground by sanz et al .
( 1993 ) , rodrguez - pascual et al .
( 1995 ) , and petitjean , riediger , & rauch ( 1996 ) .
sanz et al .
( 1993 ) and rodrguez - pascual et al .
( 1995 ) have concentrated their efforts on the forest ; they obtained high resolution spectra of the forest and low resolution spectra longward of the emission line .
petitjean et al . (
1995 ) have observed the quasar longward of emission but with moderate resolution ( fwhm @xmath5 95 ) .
clcc 0.72219 & 2796.4,2803.5 & & 0.86431 & 2796.4,2803.5 & @xmath45 & 16.35 1.15729 & 2796.4,2803.5 & @xmath45 & 16.85 1.725 & & @xmath45 & 17.05 1.84506 & 2796.4,2803.5 & @xmath45 & 16.75 & 1548.2,1550.8 & & 2.16797 & 1548.2,1550.8 & @xmath45 & 16.85 2.30812 & 1548.2,1550.8 & & 2.315,2.316 & 1526.7 & @xmath45 & 16.85 & 1670.8 & & & 1854.7,1862.8 & & & 1393.8,1402.8 & & & 1548.2,1550.8 & & 2.37995 & 1548.2,1550.8 & & 2.433 & 1393.8 & @xmath45 & 16.7 & 1548.2,1550.8 & & 2.439 & 1548.2,1550.8 & & 2.44059 & 1548.2,1550.8 & & 2.57813,2.57884 & 1548.2,1550.8 & & 2.7125 & 1548.2,1550.8 & & & 1238.8,1242.8 & & table 2 summarizes the heavy element systems we have identified in the kpno optical spectrum shown in figure 1 .
table 2 also lists the column densities determined by reimers et al .
( 1992 ) from @xmath46 as discussed above .
we detect 13 metal systems including new systems at @xmath40 = 2.37995 and 2.44059 . in this section
we comment on these metal systems including assessments of absorption saturation effects on the column density measurements .
our kpno spectrum covers the portion of the forest between @xmath47 4300 and @xmath32 = 4520 , so in principle we can measure the column densities of the metal systems at @xmath40 = 2.57813 , 2.57884 and = 2.7125 from the absorption profile .
however , the lines at 2.57813 and 2.57884 are badly blended and saturated , so the values of @xmath21 ( ) are not usefully constrained by our spectrum .
the highest redshift metal absorber is found at = 2.7125 , but there are at least five additional lines at 2.7125 @xmath3 @xmath0 2.74427 which are not detected in any metal lines despite high s / n in spectral regions where these metal lines would be observed .
reimers et al .
( 1992 ) did not recognize this metal system in their initial analysis of the _ hst _ fos spectrum , but it was identified in the reanalysis by vogel & reimers ( 1995 ) .
confirmation of this system was provided by petitjean et al .
( 1996 ) who detected the lines at 2796.4 and 2803.5 .
we also detect this doublet ( lines 11 and 12 in figure 1 ) , and comparison of the apparent column densities of the lines indicates that there is some unresolved saturation in the core of the 2796.4 profile . the 2796.4 line ( the stronger member of the doublet ) is mildly blended with 1402.8 absorption from the = 2.433 system ; this blending could mask some saturation in the profile because the excess optical depth from the line artificially boosts the 2796.4 .
however , the weaker 2803.5 line yields an integrated column density in good agreement with the column density from profile fitting ( see table 1 ) , so this magnesium doublet is only weakly saturated .
reimers et al .
( 1992 ) do not report detection of lyman continuum absorption at this redshift , but the lyman limit occurs in a region of the fos spectrum which has very low s / n .
this system produces partial lyman continuum absorption in the _ hst _ fos spectrum , and from the lyman limit optical depth reimers et al .
( 1992 ) estimate that log @xmath21 ( ) = 16.35 [ unfortunately , reimers et al .
( 1992 ) do not provide the uncertainties in their measurements of @xmath21 ( ) ] . we detect a 6.6@xmath20 absorption feature ( line 25 in fig .
1 ) at the expected wavelength of 2796.4 at this redshift .
we consider this to be a probable , but not definite , identification .
we detect a 2.3@xmath20 absorption line at the expected wavelength ( 5226.65 ) of 2803.5 , but this weak line does not line up well with the stronger line : the stronger line is best - fit with = 0.86452 while the weaker line indicates that = 0.86431 .
however , the stronger line shows evidence of two components , and one of the components is reasonably fit with = 0.86431 .
on this basis , we tentatively conclude that the stronger line is blended with an unidentified absorption feature .
the profiles of these lines are rather noisy so it is difficult to assess the impact of unresolved saturation , but these lines are weak and unlikely to be saturated .
this metal system produces obvious lyman continuum absorption in the _ hst _ spectrum , and reimers et al .
( 1992 ) find that log @xmath21 ( ) = 16.85 .
the 2796.4 and 2803.5 lines at this redshift are recorded at high significance levels in the kpno spectrum ( see lines 40 and 41 in figure 1 and table 1 ) .
the integrated apparent column densities of these lines are in agreement , but inspection of the profiles suggests that they are undersampled , and the higher column density from profile fitting indicates that the lines may be moderately affected by unresolved saturation .
no metals are unambiguously detected in this absorption system , which is identified on the basis of obvious lyman limit absorption in the _ hst _ spectrum with log @xmath21 ( ) = 17.05 ( reimers et al .
the redshift is not favorable for clear detection of many metal lines from the ground ; only the , , , and resonance lines are sufficiently redshifted to escape from the forest .
unfortunately , the 2796.4 , 2803.5 doublet at this redshift falls within the o@xmath49 atmospheric a band ( see figure 1 ) , but after removal of the a band using the hs 1946 + 7658 spectrum as a template ( see 3.1 ) , we still do not see any convincing absorption which can be attributed to . the strongest line at 2382.8 is redshifted into a cleaner region of the optical spectrum ;
the weak telluric lines near the expected wavelength of this fe line are effectively removed after division by the template , and the strong line is not evident after removal of the atmospheric lines .
petitjean et al .
( 1996 ) point out that the strong and broad line at 4551.8 could be 1670.8 at = 1.7243 , but this line occurs at the edge of the forest and petitjean et al .
acknowledge that it could be at = 2.744 instead .
we strongly prefer the identification for this line because such a strong and broad line(al ii ) = 13.1 with @xmath10 = 41.9 . ]
should be accompanied by absorption which is easily detected ( even within the telluric a band ) , but strong absorption is not seen .
we can not confidently place upper limits on the column densities of metals we do not detect in this system because the redshift is not adequately constrained by the _ hst _ spectrum ( see 7 ) .
the 2796.4 and 2803.5 lines are among the strongest resonance lines typically detected in neutral interstellar gas in the milky way ( see , for example , 3.4 in lockman & savage 1995 ) , so the absence of these lines in absorption suggests that this lyman limit system is highly ionized .
reimers et al .
( 1992 ) and vogel & reimers ( 1995 ) identify a lyman limit system at = 1.8465 in the _ hst _ fos spectrum of hs 1700 + 6416 , and from the measured @xmath50 they estimate that log @xmath21 ( ) = 16.75 .
we do not detect any metals at = 1.8465 , but we do detect the 2796.4 , 2803.5 doublet at = 1.84506 ( lines 44 and 45 in figure 1 ) at high significance levels . our kpno spectrum covers the doublet at this redshift , and two strong absorption features are present at the expected wavelengths of the 1548.2 and 1550.8 transitions ( see lines 1 and 2 in figure 1 ) .
these lines occur within the forest and thus are prone to confusion due to blending with absorption from different redshifts ; however , the velocity extents , centroids , and even the profiles derived from lines 1 and 2 are in reasonable agreement , and we believe these absorption features are primarily due to at this redshift .
comparison of the profiles indicates that these magnesium lines are significantly affected by unresolved saturated absorption .
similarly , comparison of the profiles suggests that the column density may be slightly underestimated due to unresolved saturation , but it is important to bear in mind that the column might also be overestimated due to contamination by forest lines .
the 1548.2 and 1550.8 lines are clearly detected at this redshift ( lines 15 and 16 in figure 1 ) , and reimers et al .
( 1992 ) report lyman limit absorption as well with log @xmath21 ( ) = 16.85 .
inspection of the profiles indicates that these lines are not significantly affected by saturation , and this conclusion is supported by the good agreement of the column densities from profile fitting and integration ( see table 1 ) .
therefore @xmath21()/@xmath21 ( ) = 1.23 @xmath51 .
this column density ratio is similar to the ratio measured in lower @xmath21 ( ) clouds by cowie et al .
however , we report this ratio with some trepidation because the _ hst _ lyman limit provides an estimation of the _ total _ column density , and while some of this absorption is likely to be associated with the observed lines , there may be additional gas which does not produce absorption .
reimers et al .
( 1992 ) identified this system based on detection of lyman series lines as well as various metal lines . at first glance
, lines 21 and 22 in figure 1 appear to be at the expected wavelengths of the 1548.2 , 1550.8 doublet at this redshift .
however , closer inspection of these lines casts some doubt on this identification .
first , line 21 shows two components while line 22 shows only one .
line 22 is close to the strong and complex 1548.2 absorption at = 2.315 , but it is separated enough to at least partially show two components and the second component is not apparent .
the second component of line 21 is not telluric , but it could be an unidentified line from a different redshift . however , line 22 is significantly broader than the stronger component of line 21 , and most importantly , lines 21 and 22 do not line up as well : profile fitting gives = 2.30805 for the stronger component of line 21 and = 2.30817 for line 22 .
we were unable to find a better identification for these lines , so we tentatively suggest that these are the lines , substantially confused by blending , at the redshift of the _ hst _ system .
line 22 could be an additional component of the multicomponent 1548.2 profile at = 2.315 , but in this case a corresponding line should be detected in the 1550.8 profile at = 2.315 , and nothing is apparent at the expected wavelength ( see figure 2 ) .
this absorption system , which is detected in , , , , and , is the most prominent heavy element system in our optical spectrum of hs 1700 + 6416 .
this system also shows lyman limit absorption in the fos spectrum , and from this reimers et al .
( 1992 ) measure log @xmath21 ( ) = 16.85 .
figure 2 shows the normalized absorption profiles of all metals we detect in this system ( excluding profiles which occur in the forest ) .
two components are apparent in the and absorption profiles and 4.5@xmath20 levels , respectively , in the al iii 1854.7 transition . given the weakness of these lines , it is not surprising that they are not significantly detected in the weaker al iii 1862.8 line of this doublet .
] , and at least three components are clearly present in the profiles of and
. we shall refer to these main components at = 2.3150 , 2.3155 , and 2.3164 as components 1 , 2 , and 3 respectively ( as labeled in figure 2 ) .
in addition , the 1548.2 and 1550.8 profiles both show absorption at @xmath4 2.3146 ( i.e. , at @xmath52 40 in figure 2 ) , and very weak absorption is present at this redshift in the 1393.8 profile
. there are some indications of further component structure in the profiles ( see below ) , but the combined s / n and resolution of these data do not warrant addition of more components to the profile fits . unfortunately , the 1670.8 absorption occurs close to the 1548.2 absorption line at = 2.578 . furthermore , the = 2.578 absorber shows multiple components ( = 2.57813 and = 2.57884 ) , so the identification of the absorption line at 5538.7 is ambiguous ; this absorption could be due to in component 1 ( see figure 2 ) , or it could be a third component of the absorber at = 2.57746 . the 1550.8 profile at = 2.578 does not show a third component at = 2.57746 , so the line at 5538.7 is probably due to .
however , we can not rule out the identification because the line is weak , and it may fall below our detection threshold in the weaker 1550.8 profile . in figure 3
we compare the apparent column density profiles of the 1393.8 and 1402.8 lines ( panel a ) as well as the 1548.2 and 1550.8 lines ( panel b ) and the 1854.7 line .
not surprisingly , this shows that the profile is badly saturated in components 1 and 2 .
the profile is also significantly saturated in component 3 ; the weaker line in the doublet ( filled circles in figure 3 ) yields a significantly higher than the strong line in this component .
this high degree of saturation indicates that component 3 is intrinsically rather narrow , and this in turn indicates that the gas is relatively cool .
the profile is less affected by saturation . from figure 3
we see that the weak and strong line profiles are in good agreement in components 2 and 3 , which indicates that the column density is reliably measured in these components . in component 1 , on the other hand , the weak line has a higher apparent column density than the strong line , so this component is affected by unresolved saturation .
however , figure 3 indicates that is only mildly saturated in component 1 , so a correction can be applied to the weak line apparent column density to obtain a reliable estimate of @xmath21 ( ) [ savage & sembach 1991 ; jenkins 1996 ] . using the method of savage & sembach ( 1991 )
, we estimate that the correction required for the integrated of the 1402.8 line is 0.08 dex
. therefore log @xmath53 ( ) = 13.68 + 0.08 = 13.76 in component 1 .
this corrected column density is in excellent agreement with the column density from profile fitting , log @xmath21 ( ) = [email protected] , which gives us confidence in the measurement .
for the , , and profiles we do not have the benefit of additional transitions for comparison , but these lines are all rather weak so saturation is not likely to be a problem . from figure 3 we see that a very weak absorption feature is present at @xmath52 + 118 in the profile .
this is close to component 3 but is shifted slightly blueward compared to and .
similarly , from figure 2 we see that there may be weak absorption near but slightly blueward of component 3 .
in 5 we discuss the physical conditions in this lyman limit absorber , and we use standard photoionization models to examine the heavy element abundances . this new metal system is detected only in the 1548.2 , 1550.8 doublet ( lines 26 and 27 in figure 1 and table 1 ) .
comparison of the profiles indicates that these lines are not substantially affected by unresolved saturated component structure .
similarly , the column densities from profile fitting and integration agree within the noise , so saturation is not a problem . a prominent complex of doublets is detected at 5313 @xmath3 @xmath3 5336 ( lines 2833 in figure 1 ) , which corresponds to 2.432 @xmath3 @xmath3 2.441 .
figure 4 shows an expanded plot of this region of the kpno spectrum along with the component fit which gives the profile parameters listed in columns 5 - 7 in table 1 . to show the clustering of lines , we also plot at the top of figure 4 the redshifts at which we detect absorbers in the optical spectrum of hs 1700 + 6416 .
this absorption complex has two interesting features : \(1 ) there is a very high density of absorption lines in this small redshift interval .
inspection of the spectrum in figure 4 reveals that there are at least 4 doublets in this complex within the redshift range @xmath55 = 0.00855 , and this implies that the number of absorbers per unit redshift is @xmath56 450 . in addition , profile fitting requires a broad component with a substantial column density at = 2.43329 ; this broad component is not well constrained by the present data and may split up into subcomponents when observed at higher resolution .
also , the strength of the line at 5322.4 indicates that another narrow component is present ( see below ) .
in contrast , petitjean et al . (
1994 ) find @xmath57 7.7 and tripp et al .
( 1996 ) derive @xmath12 = [email protected] using samples of intervening systems with much larger redshift intervals of @xmath55 = 1.3 and @xmath55 = 2.4 , respectively .
the petitjean et al . and tripp et al .
samples include _ all _ lines with rest equivalent width @xmath58 0.03 along multiple sight lines , and therefore these samples provide a measure of the _ average _ @xmath12 of systems at moderately high redshifts
. however , petitjean et al . and tripp et al .
treat any doublets separated by less than 200 as a single absorber . applying this treatment to the data in figure 4
, we still find that there are 2 or 3 systems in this small redshift range implying that @xmath59 200 .
the high @xmath12 measured in this cluster only indicates that the _ local _ density is very high and is not in conflict with previous estimates of @xmath12 over broad redshift intervals .
\(2 ) the weak line ( 1550.8 ) of the doublet at = 2.43306 is remarkably well - aligned with the strong line of the doublet at = 2.43877 ( see figure 4 ) . to within what velocity tolerance
are these lines aligned ? traditionally the quantity @xmath60 has been used to search for alignments of absorbers at redshifts @xmath61 and @xmath62 ; if the absorbers are aligned in some doublet with wavelengths @xmath63 and @xmath64 , then @xmath65 .
thus for the doublet we expect @xmath66 = 1550.770/1548.195 = 1.001663 , and with @xmath61 = [email protected] and @xmath62 = [email protected] we obtain @xmath66 = [email protected] .
these redshifts and uncertainties were obtained by fitting the profiles of all lines shown in figure 4 simultaneously , and the uncertainties in the profile parameters may be greater than the uncertainties formally estimated by the fitting program due to the complexity of the profiles .
however , the 1548.2 line of the @xmath61 system and the 1550.8 line of the @xmath62 system are free from confusion from this alignment , so independently fitting these lines provides an estimate of the redshifts which is not affected by the severe blending due to the redshift overlap . the @xmath61 1548.2 line is blended with a broad component at @xmath67 2.4333 ( see figure 4 ) , and it is necessary to fit the broad component and @xmath61 simultaneously to obtain an acceptable fit , but the core of the @xmath61 line is narrow and deep enough that @xmath61 is well - constrained despite the blend . the @xmath62 1550.8 line is not blended . from this independent fitting exercise ,
we obtain @xmath61 = [email protected] and @xmath62 = [email protected] . from these redshifts
we estimate that @xmath61 and @xmath62 are aligned to within 10 at the 3@xmath20 level .
there is another pair of weak doublets in figure 4 which show the weak line of one absorber aligned with the strong line of another , although this case is not as strong .
the line at 5313.49 is attributed to 1548.2 at = 2.43204 .
the corresponding weak ( 1550.8 ) line is detected at the expected wavelength ( 5322.3 ) , but it is considerably stronger than the 1548.2 line , and fitting the profiles of the 5313.5 and 5322.3 lines with only one component yields a very poor fit . since the line at 5313.5 only allows one component , we attribute the line at 5322.3 to the 1550.8 line at = 2.43204 blended with a 1548.2 line at = 2.43785 , and we find that there is a weak absorption line at the expected wavelength of 1550.8 at = 2.43785 .
since these doublets are weaker than the aligned doublets discussed in the previous paragraph , we can not constrain the velocity alignment as tightly .
we estimate that the weak line of the = 2.43204 absorber is aligned to within @xmath5 25 with the strong line of the = 2.43785 absorber .
this alignment of the weak line from one redshift with the strong line from a different redshift is reminiscent of the absorption `` line locking '' phenomenon which was once a topic of considerable interest and controversy ( perry , burbidge , & burbidge 1978 ; weymann , carswell , & smith 1981 and references therein ) .
some interesting examples of aligned weak and strong doublet lines from different redshift systems have appeared in more recent literature ( e.g. , foltz et al .
1987 ; wampler 1991 ; wampler , bergeron , & petitjean 1993 ) . in some cases
these occur in broad absorption line ( bal ) and `` associated '' absorption systems with @xmath4 @xmath1 , which is perhaps not surprising since radiative acceleration provides a natural explanation of these alignments , and the presumably close proximity of the @xmath4 @xmath1 absorbers to the qso would subject them to an extreme non - isotropic radiation field .
several models of bal outflows have recently been proposed in which radiation pressure plays a critical role ( e.g. , arav , li , & begelman 1994 ; murray et al .
1995 ; scoville & norman 1995 ) , and arav et al .
( 1995 ) have shown that a double - trough spectral signature predicted by their model is convincingly detected in the types of qsos where it is expected to be present .
however , the complex shown in figure 4 is definitely not a bal system , and it is displaced from @xmath1 by @xmath524000 , so it usually would not be considered an associated system either .
bal outflows do attain velocities this large ( turnshek 1988 ) , so perhaps this complex is close to the qso and was accelerated in an earlier bal outflow that has since mostly dissipated ( a scenario proposed by turnshek 1988 ) .
alternatively , this may be a bal outflow which is just beginning to form with the high velocity absorption showing up first to be followed later by lower velocity absorption with closer to @xmath1 . before pursuing these speculations , it is important to ascertain whether or not the doublet alignments in figure 4 could simply be coincidental . if we consider randomly placing some number of absorbers in a specified velocity interval , then we can use the binomial distribution to estimate the probability that the weak line of one absorber could by chance be aligned with the strong line of a different absorber to within some velocity tolerance .
it is not immediately obvious what velocity interval or number of absorbers should be assumed for this calculation , however , and these assumptions significantly affect the probability .
the _ minimum _ velocity interval ( which translates to an _ upper limit _ on the probability ) is the difference between the velocity of the strong line of the highest redshift doublet in the cluster and the velocity of the strong line of the lowest redshift doublet ( if the velocity interval were less than this , then the complex shown in figure 4 would never occur because some of the doublets would be outside of the velocity interval ) .
this corresponds to a wavelength interval of @xmath68 = 13.2 or a velocity interval of @xmath69 = 746 .
there are four well - detected absorbers ( = 2.43204 , 2.43306 2.43877 , and 2.44059 ) in this @xmath32 range , and as noted above , it is likely that there is another at = 2.43785 , and there is a broad component at = 2.43329 which could be due to multiple blended lines . with four absorbers and @xmath69 = 746 , the probability of one weak / strong line alignment to within 10 is 15% .
if we include the second apparent weak / strong line alignment , then we must relax the velocity overlap tolerance to 25 , and we must randomly place at least five absorbers in the velocity interval , so in this case the probability of two chance alignments is 12% .
if we increase the number of absorbers to , say , 7 , then the probability of two alignments to within 25 increases to 28% .
evidently it is possible that these apparently line locked systems are just chance coincidences in an overdense complex .
nevertheless , since similar highly displaced line alignments have been reported elsewhere ( wampler et al . 1993 ) , it is worthwhile to make a note of this .
if highly displaced line alignments show up in future high resolution qso absorber studies , then this may reveal an interesting new qso phenomenon or evolutionary stage .
it is also interesting to note that sargent , boksenberg , & steidel ( 1988 ) , hamann , barlow , & junkkarinen ( 1997a ) , and jannuzi et al .
( 1996 ) have reported evidence of a new class of qso absorption systems which are similar to bals but highly displaced from the qso systemic redshift and lacking the @xmath70 absorption usually detected in bals .
these displaced bals may be related to the complex discussed in this section .
it may be possible to establish that this complex is close to the qso by monitoring the absorption profile ; if the profile shows temporal variability , then the absorbing gas must be close to the qso .
reimers et al .
( 1992 ) report detection of this system in several metals , but they do not detect lyman limit absorption .
however , the spectral coverage of the fos observation barely reaches the expected wavelength of this lyman limit , and their s / n drops off precipitously in this region ( see figure 1 in vogel & reimers 1995 ) , so the fos spectrum is not sensitive to lyman continuum absorption at this redshift . we detect the 1548.2 , 1550.8 doublet in this system , and two components are obvious in the profiles of both lines ( see lines 36 and 37 in figure 1 ) .
comparison of the profiles of the doublet does not show unresolved saturation in either component .
the 1548.2 profile shows a third component , but there is no evidence of this component in the 1550.8 profile and , as discussed above , this third component = 7.3 and log n(c iv ) = 12.66 . ]
may be 1670.8 at = 2.3150 .
the = 2.3150 system shows complex component structure ( see figure 2 ) , so the 1548.2 profile could be contaminated with absorption from = 2.3155 .
reimers et al .
( 1992 ) and vogel & reimers ( 1995 ) do not report detections of any metals at this redshift , but as discussed by petitjean et al .
( 1996 ) , some of the heavy element lines at this redshift in the fos spectrum may have been misidentified by vogel & reimers .
for example , petitjean et al . point out that the line vogel & reimers attribute to 833.3 at = 2.433 could instead be 770.4 at = 2.712 . in the kpno optical spectrum
we detect the 1238.8 , 1242.8 and 1548.2 , 1550.8 doublets at this redshift ( see lines 3 - 4 and 38 - 39 , respectively , in table 1 and figure 1 ) .
a 4.7@xmath20 absorption line is present at 4842.45 , the expected wavelength of 1304.4 at = 2.71249 .
however , this identification probably is not correct because other lines should be detected as well , and they are not .
for example , the 1260.4 line is considerably stronger than the 1304.4 line , but the 1260.4 line is not detected .
there is a weak unidentified feature recorded at 4679.89 ( line 8 in figure 1 ) , but this is shifted redward from the expected wavelength of the 1260.4 line by @xmath55 = + 0.00047 , and furthermore this unidentified line is much weaker than the expected 1260.4 line ( based on the strength of the line at 4842.45 , assuming it is due to 1304.4 ) .
similarly , the 1526.7 line should be easily detected , but no line is apparent at the expected wavelength ( see figure 1 ) .
the apparent column density profiles of the and doublets at = 2.7125 are shown in figure 5 .
this figure suggests that the absorption profile is affected by unresolved saturation ; the weak line ( 1242.8 ) has a significantly higher than the strong line in the core of the profile , and the integrated 1242.8 column density , log @xmath21 ( ) = 13.87 , is 0.09 dex greater than the integrated column of the 1238.8 line .
alternatively , figure 5 may indicate that the absorbing gas does not completely cover the flux source ( see below ) .
the column density from profile fitting agrees with the integrated 1242.8 column density , but inspection of the fitted profiles shows that the fit is not very good and the observed optical depth in the core of the 1242.8 line exceeds the profile fit , so the true column density could be somewhat higher . from figure 5
we see that , on the other hand , is not significantly affected by unresolved saturation ( or incomplete covering ) at this redshift ; the weak and strong line profiles are in good agreement as are the column densities from profile fitting and integration ( see table 1 ) .
this absorption system occurs within 5000 of the qso emission redshift and thus qualifies as an `` associated '' absorber .
there are some indications that these are a special class of qso absorbers which may be intrinsically related to the qso nucleus or host galaxy , but there is also evidence that some absorption systems within 5000 of @xmath1 are at least a few hundred kpc away from the qso ( turnshek , weymann , & williams 1979 ; morris et al .
1986 ; tripp et al .
probably this class of absorbers includes a mix of intrinsic and intervening absorption systems . in
5.2.3 we use photoionization models to study the abundances and ionization in this absorption system .
recently it has been suggested that the gas in some associated absorbers does not completely cover the qso flux source ( wampler et al .
1993 ; petitjean et al . 1994 ; hamann et al .
1997a , b ; barlow & sargent 1997 ) .
if this is true , then these associated absorbers must be close to the qso nucleus .
the case presented by barlow & sargent ( 1997 ) is particularly convincing because the profile has an obvious flat bottom indicative of strong saturation , but the flux in the flat bottom of the profile is well - detected and is not zero .
while quite interesting , this complicates the column density measurements because even if the lines are unsaturated , the incomplete covering will introduce errors in column densities from profile fitting and integration .
the discrepancy between the 1238.8 and 1242.8 profiles shown in figure 5 could indicate that the absorbing gas does not completely cover the flux source . in principle , one can attempt to correct the absorption profiles for partial covering ( c.f . , hamann et al .
1997b ; barlow & sargent 1997 ) , but in this case this is difficult since the absorption lines are strong enough that a mix of partial covering and saturation could affect the measurements .
for the column densities in the = 2.7125 absorber , we adopt the conservative approach of simply treating the integrated column density of the weak line as a lower limit on @xmath21 ( ) .
in this section we examine the physical conditions and abundances in the metal systems of hs 1700 + 6416 .
we begin with simple inferences drawn from the absorption line measurements , and then we apply standard photoionization models to estimate abundances .
however , we find some indications that the standard single - zone photoionization models are too simple to explain the data and multi - zone models may be required ( see below ) . the observed line widths provide one of the most straightforward probes of physical conditions .
if the absorption profile is dominated by doppler broadening ( i.e. , thermal motions ) , then the temperature @xmath71 of the gas can be calculated from the measured doppler parameter , @xmath10 = @xmath72 = 0.129@xmath73 , where @xmath74 is the atomic weight of the element and @xmath20 is the radial velocity dispersion of the gas ( spitzer 1978 ) . in qso absorbers , the gas is often cool enough that other factors such as bulk motions , turbulence , or multiple unresolved components can significantly broaden the line , so the temperature derived from this equation must be treated as an upper limit .
since @xmath10 goes as @xmath75 , lighter elements like carbon are more useful for constraining the temperature of the gas . in table 3
we list upper limits on the gas temperature derived from the doppler parameters substantially smaller than the 20 resolution .
] of all lines detected in the kpno optical spectrum .
ccc 2.16797 & [email protected] & @xmath32.0@xmath76 2.30812 & [email protected] & @xmath34.7@xmath76 2.31458 & [email protected] & @xmath31.5@xmath76 2.31505 & [email protected] & @xmath39.2@xmath77 2.31568 & [email protected] & @xmath35.9@xmath76 2.31638 & [email protected] & @xmath31.1@xmath77 2.37995 & [email protected] & @xmath32.9@xmath77 2.43204 & [email protected] & @xmath34.6@xmath77 2.43306 & [email protected] & @xmath31.2@xmath77 2.43329 & [email protected] & @xmath31.4@xmath78 2.43785 & [email protected] & @xmath31.1@xmath76 2.43877 & [email protected] & @xmath32.4@xmath76 2.44059 & [email protected] & @xmath33.0@xmath76 2.57813 & [email protected] & @xmath32.6@xmath76 2.57884 & [email protected] & @xmath38.8@xmath76 2.71245 & [email protected] & @xmath39.4@xmath77 an issue of considerable interest is the question of the ionization mechanism(s ) in the various types of qso absorbers . are intervening qso absorbers photoionized by the uv background radiation from qsos and agns , or do local photoionizing sources and/or collisional ionization play a role ? from table 3
we see that four of the absorption lines in the hs 1700 + 6416 optical spectrum have @xmath79 50000 k. at these temperatures very little is present in gas in collisional ionization equilibrium ( sutherland & dopita 1993 ) , so these systems are probably photoionized
. however , it is also possible that the narrow lines occur in nonequilibrium gas which has cooled more rapidly than it has recombined ; this is predicted in some renditions of galactic fountain theory ( edgar & chevalier 1986 ; benjamin & shapiro 1997 ) .
it is important to note that in some cases these lines are moderately weak and noise could cause the profiles to appear artificially narrow . on the other hand ,
some of the lines are so broad as to indicate that the line profiles are probably _ not _ dominated by thermal doppler broadening . for example , the absorber at = 2.57884 has @xmath80 , which implies that @xmath81k if the profile is dominated by thermal motions ; at this temperature carbon is predominantly in higher ionization stages than in collisional ionization equilibrium ( sutherland & dopita 1993 ) . the high ion column density ratios in the lyman limit absorber at = 2.3150 , 2.3155 indicate that the ionization conditions differ significantly from the conditions in the gaseous halo of the milky way . from figure 3
we see that @xmath21()/@xmath21 ( ) @xmath6 10 in components 2 and 3 of this absorber and could be significantly larger since the profile is affected by strong saturation . for comparison , sembach & savage ( 1992 )
measure @xmath21()/@xmath21 ( ) = [email protected] in the milky way gaseous halo .
more recently sembach , savage , & tripp ( 1997 ) have compiled the best measurements made with the _ hst _ and _ international ultraviolet explorer ( iue ) _ of high ion column densities in the galactic ism , and they find @xmath21()/@xmath21 ( ) = [email protected] with the vast majority of the ism / ionic ratios being well below 10 ( see their figure 8) .
this high ionic ratio in the ll absorber is not expected on the basis of cooling galactic fountain theory but could be produced in supernova remnants , conductive interfaces between warm and hot gas , or `` turbulent mixing layers '' ( see 8 in sembach et al .
1997 and references therein ) .
this higher ratio could also be due to the significantly increased intensity of the extragalactic uv background radiation at @xmath82 2 compared to @xmath82 0 ( c.f . , bechtold 1995 and references therein ) .
since the temperature of the gas producing the absorption line in this ll absorber at = 2.31638 ( i.e. , component 3 , see figure 2 ) is evidently less than roughly 11000 k ( table 3 ) , it seems likely that this system is photoionized .
components 1 and 2 may or may not be photoionized ( the line widths are large enough to allow collisional ionization ) , but given the small total column density for all components derived from @xmath83 by reimers et al . (
1992 ) [ log n ( ) = 16.85 ] , photoionization by the extragalactic uv background radiation is likely to be significant in all three components ( c.f .
, viegas 1995 ) , so modeling is required to estimate elemental abundances . following the seminal paper of bergeron & stasiska ( 1986 ) ,
it has become customary to use photoionization models to infer abundances in moderate column density qso absorbers , and the physical conditions discussed above indicate that this is appropriate . in this section
we use models constructed with the photoionization code cloudy ( version 84.12 ; ferland 1993 ) to estimate the abundances in the lyman limit and associated absorption systems .
we model the ll absorbers as a constant density plane parallel slab photoionized predominantly by the madau ( 1992 ) extragalactic background radiation from qsos / agns , which includes the effects of h and he opacity in intervening clouds and lyman limit systems .
the slab is illuminated on one side .
we assume the mean intensity at the hydrogen lyman limit is @xmath84(ll ) = 10@xmath85 ergs s@xmath86 @xmath87 hz@xmath86 sr@xmath86 . for the associated absorber ,
the model assumes the gas is predominantly photoionized by the flux from hs 1700 + 6416 itself ( see 5.2.3 ) . once the radiation field has been specified , other parameters which can be modified to fit the observed column densities are the overall metallicity [ m / h ] , the relative heavy element abundances , the ionization parameter @xmath88 = h ionizing photon density / total hydrogen number density ) , and the column density . we express the logarithmic abundance of element x relative to element y using the usual notation , [ x / y ] = log ( x / y ) - log ( x / y)@xmath89 , and we use [ m / h ] to indicate the overall abundance of the metals relative to solar . throughout this paper
we use the solar abundances from grevesse & anders ( 1989 ) .
we begin by placing lower limits on the heavy element abundances , and then for absorbers detected in several species and ionization stages , we use the models to more tightly constrain the metallicities . as discussed by hamann ( 1997 ) , even if we have measured column densities of _ only one _ ionization stage of metal x ( e.g. , ) and , we can still place a lower limit on [ x / h ] in the context of the photoionization model . the abundance
[ x / h ] can be expressed as @xmath90 = { \rm log}\left ( \frac{n({\rm x}_{i})}{n({\rm h \
i)}}\right ) + { \rm log}\left ( \frac{f({\rm h \
i})}{f({\rm x}_{i})}\right ) - { \rm log}\left ( \frac{\rm x}{\rm h}\right ) _ { \odot}\ ] ] where @xmath91 is element x in ionization stage @xmath92 , @xmath21 is the column density , @xmath31 is the ionization fraction , and ( x / h)@xmath89 is the abundance ratio in the sun . by choosing the model ionization parameter which minimizes @xmath93 , we can derive a lower limit on [ x / h ] ; any other value of @xmath94 will lead to higher [ x / h ] . in practice
our method is to calculate @xmath21(x@xmath95 ) vs. @xmath94 for a series of models with decreasing values of [ x / h ] and fixed @xmath21 ( ) ; eventually [ x / h ] drops low enough so that the peak of the @xmath21(x@xmath95 ) vs. @xmath94 curve is lower than the observed column density of x@xmath95 . for several of the hs 1700 + 6416 ll absorbers , table 4
summarizes the lower limits on [ x / h ] which agree with the measured column densities , to within the 2@xmath20 uncertainties , at the peak of the @xmath21(x@xmath95 ) vs. @xmath94 curve calculated by the model described above .
cccccc 1.15729 & @xmath6 0.22 & & & & 1.84506 & @xmath6 0.45 & @xmath6 2.26 & & & 2.16797 & & @xmath6 2.81 & & 2.31499 & & & & @xmath6 0.95 & @xmath6 0.96 2.31551 & & & & @xmath6 1.35 & @xmath6 1.40 & & & & & & 2.7125 & & @xmath6 0.82 & @xmath6 0.65 & & & & & & & & 2.7125 & & @xmath6 0.25 & @xmath6 + 0.19 & & the abundance lower limits derived using equation 3 as discussed above are [ si / h ] @xmath6 0.95 and [ al / h ] @xmath6 0.96 in component 1 ( = 2.3150 ) and [ si / h ] @xmath6 1.35 and [ al / h ] @xmath6 1.40 in component 2 ( = 2.3155 ) .
these lower limits were calculated assuming log @xmath21 ( ) = 16.85 _ for each component_. there are at least three absorption components in this ll system ( see figure 2 ) , and log @xmath21 ( ) = 16.85 is the _ total _ column ( i.e. , integrated across all components ) derived from the lyman limit optical depth .
therefore these are rather conservative lower limits ; @xmath21 ( ) in the _ individual _ components is almost certainly lower than the assumed value so the abundances are almost certainly higher than these lower limts . in this absorber
we detect absorption lines from several species including multiple ionization stages of si and al ( see table 2 ) , so we may be able to more accurately pin down the abundances using photoionization models which simultaneously match all of the measured heavy element column densities . however , with these models one must bear in mind that different ionization stages such as and may not occur in the same gas , i.e. , the absorber may be a multiphase medium
. we will start with the single zone model in which all of the absorption lines are assumed to arise in the same gas because it is the simplest model . with this caveat ,
consider component 1 ( = 2.3150 ) .
based on the heavy element column densities , we estimate that log @xmath21 ( ) @xmath4 16.6 for component 1 assuming components 1 and 2 have roughly the same metallicity and contain the bulk of the neutral hydrogen .
our best estimations of the metal column densities in this component are log @xmath21 ( ) = [email protected] , log @xmath21 ( ) = [email protected] , log @xmath21 ( ) = [email protected] , and log @xmath21 ( ) = [email protected] ( see 4.8 ) .
a very crude lower limit of log @xmath21 ( ) @xmath15 14.6 is obtained by integrating the apparent column density of 1550.8 over the velocity range of component 1 . to simultaneously produce enough and in component 1 with log @xmath21 ( ) = 16.6 ( ignoring for the moment the higher ionization stages ) ,
the model must have [ m / h ] @xmath6 0.73 .
since and are the dominant ionization stages of si and al in interstellar regions , this lower limit holds even if the higher ion absorption does not occur in the same gas . however to simultaneously match @xmath21 ( ) and @xmath21 ( ) , the model must have a higher metallicity . to search for a model which agrees with _ all _ of the observed column densities , we have carried out a series of cloudy calculations with metallicities ranging from [ m / h ] = 0.60 to [ m / h ] = 0.30 .
figure 6 shows a sample cloudy calculation , and figure 7 summarizes the results of these modeling efforts .
figure 6 shows the column densities of these species , as a function of @xmath94 , calculated by a photoionization model with [ m / h ] = 0.45 .
we present this particular cloudy calculation as a sample because this model provides the single best fit . from the figure
we see that the model is consistent with the observed column densities of , , and , to within the 2@xmath20 uncertainties , when 2.73 @xmath0 log @xmath97 2.46 , and it is consistent with the lower limit on @xmath21 ( ) when the ionization parameter exceeds 2.54 .
the dashed line in figure 6 shows the mean temperature of the model as a function of @xmath94 , and we see that in this ionization parameter range the model is also in agreement with the observational constraints on the gas temperature .
however , over this ionization parameter range the calculation predicts too much by 0.40.5 dex , which is greater than the 3@xmath20 uncertainty of @xmath21 ( ) . this problem is aggravated by the possible contamination of the absorption from 1548.2 at = 2.57746 ( see 4.8 and figure 2 ) ; if this absorption is partly or mostly due to at a different redshift , then the discrepancy between the model and the observed @xmath21 ( ) is even greater .
figure 7 summarizes the results from five photoionization models with 0.60 @xmath0 log @xmath97 0.40 . for each species ,
the labeled horizontal bar in figure 7 shows the ionization parameter range which yields model column densities in 2@xmath20 agreement with the measured values . from figure 7
we see that _ none _ of the models are able to fit all of the observed column densities for any value of @xmath94 .
models with 0.50 @xmath98 [ m / h ] @xmath98 0.40 can fit all of the species except . when [ m / h ] exceeds 0.40 , the model fails to fit several of the observed column densities at any given value of @xmath94 .
there are several possible explanations for the difficulty in fitting including the following .
( 1 ) aluminum could be underabundant in the gas - phase relative to si and c , due either to depletion onto dust grains or relative metal abundances which are intrinsically different from solar . in the ism of the milky way , aluminum is strongly depleted from the gas - phase by incorporation into dust grains ( jenkins 1987 ) while si and c are moderately or lightly depleted ( sofia , cardelli , & savage 1994 ) , and therefore depletion could cause al to be underabundant . alternatively , the intrinsic al / si and al / c abundances could be different from solar . in low metallicity stars in the milky way halo ,
si is overabundant relative to aluminum , and this is attributed to different nucleosynthesis time scales for si and al ( mcwilliam et al .
1995 ; ryan , norris , & beers 1996 ; wheeler , sneden , & truran 1989 and references therein ) .
either of these phenomena could alleviate the problem with in the model .
if al is underabundant by @xmath50.2 dex relative to si and c , for example , then the model shown in figure 6 could fit all of the metal column densities at log @xmath94 = 2.54 .
( 2 ) the absorption lines could arise in a multiphase medium in which the low ionization species are associated with the but some or all of the higher ion absorption occurs in a physically distinct region . in this case
the single - zone models of figures 6 and 7 are not expected to succeed ; more complex multi - zone models are required .
( 3 ) there may be some problems with the al atomic data in the model , especially the al dielectronic recombination rates ( see 4.1 in petitjean et al . 1994 ) .
( 4 ) we may be deceived by noise since the line is marginally detected .
the presence of multiple components in this absorber suggests that it may not be appropriate to model the total column densities as a single constant density slab ; the multiple components indicate that the density changes along the line of sight . however , some previous studies of abundances in lyman limit absorbers ( e.g. , vogel & reimers 1995 ) are based on total column densities due to the low resolution of the instruments used , so if we wish to compare our abundances to these studies ( see 7 ) , we should start with total column densities .
the column densities integrated across the velocity range of the main low ion absorption ( components 1 and 2 ) in this absorber are log @xmath21 ( ) = [email protected] , log @xmath21 ( ) = [email protected] , and log @xmath21 ( ) = [email protected] . in this case , the column density appropriate for the model is the total , log @xmath21 ( ) = 16.85 . with these column densities ,
we find that the photoionization model agrees with the measurements to within the 2@xmath20 uncertainties when 0.62 @xmath0 [ m / h ] @xmath0 0.43 . throughout the rest of this section we will use [ m / h ] @xmath4 0.45 as a rough estimate of the metallicity in this absorber since this value for [ m / h ] provides the best fit to the component 1 column densities and agrees with the constraints set in the previous paragraph from the integrated column densities .
how does the metallicity derived for this ll absorber compare to other intervening qso absorption systems at similar redshifts ? at first glance ,
[ m / h ] @xmath4 0.45 seems like a high metallicity for = 2.315 .
however , it may not be .
recently , several high quality abundance measurements for intervening damped qso absorbers have appeared in the literature ( lu et al .
1996 and references therein , see their table 16 ) . in figure 8
we show the si and al abundances measured in these damped absorbers with open circles .
the filled triangles in figure 8 indicate the _ conservative _ lower limits on [ si / h ] and [ al / h ] at = 2.3150 and 2.3155 from table 4 .
we also show in figure 8a the si abundances derived by steidel ( 1990 ) in a study of several high redshift ll absorbers ( open squares ) . from this figure
we see that the gas - phase si and al abundances in this ll absorber are comparable to the gas - phase abundances in a significant fraction of the damped absorbers at similar redshifts .
the open star at @xmath14 = 0 in figure 8a is the diffuse cloud gas - phase abundance of si observed in the galactic halo ( savage & sembach 1996 ) .
it is interesting to note that at @xmath4 2 , [ si / h ] in many intervening qso absorbers is already approaching the value observed in the gaseous halo of the milky way .
it is also interesting to note that [ si / h ] appears to increase significantly when @xmath98 3 .
( 1996 ) suggest that the sudden increase in metallicity at @xmath82 3 is due to chemical enrichment from the first wide - spread star formation in high redshift galaxies .
however the sample is small , and more abundance measurements are needed .
some additional properties of this ll absorber can be derived from the models .
since we have assumed a value for the mean intensity at the lyman limit @xmath84(ll ) , the hydrogen number density can be determined from the ionization parameter .
the top axis of figure 6 indicates how @xmath99 changes with @xmath94 .
the models also specify the thickness @xmath100 of the absorber , and this in turn can be used to estimate the cloud mass , assuming the absorber is spherical ( see vic in steidel 1990 ) .
the model with [ m / h ] = 0.45 and log @xmath94 = 2.6 has @xmath99 = 0.02 @xmath8 , @xmath100 = 240 pc , and cloud mass @xmath101 = 1.0 @xmath22 10@xmath102 .
these rough estimates of @xmath100 and @xmath101 are close to the values predicted by the recent two - phase gaseous halo model of mo & miralda - escud ( 1996 ) for a cold photoionized phase embedded in a hot halo .
we only detect or absorption lines associated with the rest of the ll absorbers ( see table 2 ) , so photoionization models can not tightly constrain the elemental abundances ( the level of ionization is unknown ) .
nevertheless , we can use cloudy models to place lower limits on the metallicities of these absorbers using equation 3 as discussed above . applying this method with the madau ( 1992 ) extragalactic radiation field , we derive [ c / h ] @xmath6 2.26 for the ll absorber at = 1.84506 and [ c / h ] @xmath6
2.81 for the ll absorber at = 2.16797 .
however , for the = 1.84506 system we also detect , and we derive a lower limit of [ mg / h ] @xmath6 0.45 . if we assume that the , , and absorption lines at = 1.84506 all originate in the same gas , then the photoionization model requires [ mg / h ] = [ c / h ] = 0.43 .
this must also be treated as a lower limit on [ m / h ] since both and are affected by absorption saturation at this redshift ( see 4.5 ) .
finally , applying this same method to the absorber at = 1.15729 , we derive [ mg / h ] @xmath6 0.22 .
once again we see indications that high metallicities are already present in moderate redshift intervening absorbers .
we detect and rather strong absorption at = 2.7125 ( lines 38 - 39 and 3 - 4 in figure 1 ) .
we can also place good upper limits on the column densities of , , , , , and at this redshift since these lines are not in the forest .
no absorption lines are apparent at the expected wavelengths ; upper limits on these species along with the and column densities are summarized in table 5 .
the and absorption lines are displaced by @xmath5600 from the emission redshift of the qso and thus belong in the `` associated '' absorber class .
recent high resolution studies have shown that associated absorbers are often characterized by strong high ion absorption ( especially ) and overall metallicities comparable to or greater than solar ( wampler 1991 ; petitjean et al .
1994 ; savaglio et al . 1994 ; hamann et al . 1995 ; tripp et al . 1996
; hamann 1997 ) .
the ionization level of the hs 1700 + 6416 associated system inferred from the high ion column density ratios is similar to other associated absorbers and rather different from the ism of the milky way .
the = 2.7125 ionic ratios are @xmath21()/@xmath21 ( ) @xmath3 0.2 and @xmath21()/@xmath21 ( ) @xmath15 5.2 , whereas sembach et al .
( 1997 ) find that the median ratios in the galactic ism are @xmath21()/@xmath21 ( ) = [email protected] and @xmath21()/@xmath21 ( ) = [email protected] based on 40 sight lines studied with _
hst _ and the _ iue_. for comparison , we note that petitjean et al .
( 1994 ) report @xmath21()/@xmath21 ( ) @xmath5 0.3 and 0.4 for associated absorbers of pks 0424 - 131 and q 0450 - 131 respectively .
this high degree of ionization is perhaps not surprising since associated absorbers are probably physically close to rather uv luminous qsos , either somehow associated with the qso nucleus / host galaxy or possibly occuring in the halos of galaxies in a cluster near the quasar , and the gas is predominantly photoionized by the qso .
given the observed column density ( see below ) , ionization corrections in the hs 1700 + 6416 associated system are certain to be large , so once again photoionization modeling is required to infer abundances .
lcc ...... & 1215.670 & [email protected] ..... & 1302.168 & @xmath313.74 ..... & 1334.532 & @xmath313.28 .... & 1670.787 & @xmath311.68 .... & 1526.707 & @xmath312.97 .... & 1608.451 & @xmath313.21 .... & 1393.755 & @xmath312.46 .... & 1548.195 , 1550.770 & [email protected] ..... & 1238.821 , 1242.804 & @xmath613.87 in this associated absorber , we do not have the benefit of optically thin lyman continuum absorption for the determination of @xmath21 ( ) , so we have used the absorption profile ( the line at 4513.2 in figure 1 ) to constrain the column density .
direct integration of the apparent column density profile at this redshift yields log @xmath21 ( ) = [email protected] . for the purpose of this photoionization modeling effort
, we assume that this column is not underestimated due to unresolved absorption saturation , but at any rate this provides a firm lower limit .
if the absorption occurs in the same gas that produces the and absorption ( which is assumed when abundances are derived from a constant density single slab photoionized model ) , then at the likely gas temperatures the profile is well enough resolved so that any saturation should be apparent . instead , the profile is broad and the apparent optical depth is only @xmath103 = 0.7 in the deepest part of the profile .
furthermore , if this profile contains one or more components which are narrow enough to cause unresolved saturation , then these cold components might be detectable in low ionization stages such as , and they are not apparent ( see figure 1 and table 5 ) .
therefore , we shall proceed with the photoionization modeling using log @xmath21 ( ) = 13.6 .
however , we noted in 4.12 that the profiles show evidence of unresolved saturation or incomplete covering of the qso flux source .
if the absorbing gas does nt completely cover the qso flux source , then the column density from direct integration of may require a correction .
this would increase the measured column density , and in the case of , this might decrease the metallicity required by the photoionization model to match the heavy element column densities .
however , the heavy element column densities may be underestimated as well , so the corrections could push the model to higher or lower metallicities .
since the qso itself is likely to be the dominant source of ionization , we have used the observed spectral energy distribution ( sed ) of hs 1700 + 6416 for the input radiation field for cloudy abundance estimations .
the qso continuum is _ detected _ from @xmath104 = 912 ( in the qso frame ) down to @xmath105 320 , which is the region of the spectrum which governs the ionization of most low and intermediate ionization stages and is close to the critical region of the spectrum where and are photoionized , i.e. , 125 @xmath106 300 .
however , since the observed qso flux is attenuated by at least seven optically thin lyman limit absorbers ( see figure 1 in reimers et al .
1992 ) , the true photoionizing flux impinging on the associated absorber gas is certainly greater than the observed flux .
this true flux is difficult to estimate because the observed flux must be corrected for ( 1 ) the seven optically thin ll absorbers , ( 2 ) the net reddening introduced by any dust in intervening absorbers ( which is smeared by the various redshifts of the intervening systems ) , and ( 3 ) attenuation by the forest and the ` valley ' , which includes the cumulative lyman continuum absorption due to all of the intervening absorbers ( mller & jakobsen 1990 ) . the correction for the seven ll systems is straightforward ( see equation 2 ) , but we have little information on the magnitude and wavelength dependence of the other corrections .
several recent high resolution studies have suggested that there may be very little dust in intervening damped absorbers ( lu et al .
1995 ; meyer , lanzetta , & wolfe 1995 ; prochaska & wolfe 1996 ; lu et al .
however , this conclusion has been contested by pettini et al .
( 1997 ) who argue that the observed abundances are consistent with a modest amount of dust in the qso absorbers . to derive a first - order estimation of the associated system metallicity
, we have used the observed sed of hs 1700 + 6416 from siemiginowska et al .
( 1996 , see their figure 2 ) for the cloudy radiation field .
this sed is based on the _ hst _ fos data from reimers et al .
( 1992 ) combined with ground - based optical spectrophotometry , the rosat pspc data from reimers et al .
( 1995a ) , and an iras detection of the qso at 60 @xmath107 m .
siemiginowska et al . have corrected the _ hst _ uv spectrum for the seven ll absorbers , but they have not applied any corrections for intervening dust or forest / valley attenuation .
therefore the siemiginowska et al .
sed provides a firm lower limit on the photoionizing flux ; various corrections could make the true flux greater , but the flux can not be less than the siemiginowska et al . sed .
we show in figure 9 the column densities calculated by this model as a function of the ionization parameter @xmath94 for [ m / h ] = 0.65 with solar relative abundances and log @xmath21 ( ) = 13.6 .
we show this particular model because it provides a lower limit on [ n / h ] ( see below ) .
we also show in this figure how the mean temperature ( dotted line ) of the model changes with @xmath94 , and we plot the observed column density as well as the lower limit on @xmath21 ( ) and the upper limits on @xmath21 ( ) and @xmath21 ( ) near ( arbitrarily ) log @xmath94 = 2.2
. the constraints on the low ions and are easily satisfied by the model , and the model temperature is compatible with the upper limit set by the line width ( see table 3 ) .
we see from figure 9 that the model agrees with the observed column density to within 2@xmath20 over a broad range in ionization parameter , 1.4 @xmath0 log @xmath108 + 0.4 .
however , over most of this ionization parameter range , the model falls far short of the lower limit on @xmath21 ( ) . to produce enough , the model requires log @xmath94 = + 0.4 , and even then the model column is just barely large enough , which sets a lower limit of [ n / h ] @xmath6 0.65 . for carbon
the model requires [ c / h ] @xmath6 0.82 .
however , the true column density may be significantly larger than the lower limit in table 5 , and to obtain a higher / ratio , @xmath94 must be increased ( see figure 9 ) . at higher values of @xmath94
the and column densities predicted by the model decrease , so the metallicity must in turn be increased as well to agree with the measured column densities .
these limits assume solar n / c relative abundances .
hamann & ferland ( 1993 ) have argued , on the basis of qso _ emission _ line intensities , that nitrogen is overabundant relative to carbon in the immediate vicinity of a qso .
this may provide another means to boost the / ratio in this associated absorber _ if _
the absorbing gas is close to the quasar , but in this case models with lower ionization parameters and lower metallicities can be brought into agreement with the measured column densities .
associated system abundance measurements based on and are sensitive to the shape of the ionizing continuum assumed for the model .
hamann ( 1997 ) has recently assembled a suite of photoionization models for the study of abundances in associated and broad absorption lines using a variety of continuum shapes , and we can use his work to illustrate this sensitivity to continuum shape .
his continuum shape ` a ' with a uv cutoff temperature @xmath109 k ( see appendix in hamann 1997 ) is consistent with the observed sed of hs 1700 + 6416 from siemiginowska et al .
( 1996 ) in the fuv but has significantly higher flux in the ionizing ultraviolet where siemiginowska et al . did not measure the qso flux . using this continuum shape ` a ' , we derive [ c / h ] @xmath6 0.25 and [ n / h ] @xmath6 + 0.19 for the associated system of hs 1700 + 6416 .
the continuum shape we used in the previous paragraph and in figure 9 ( based on siemiginowska et al . ) provides firm lower limits on [ c / h ] and [ n / h ] because this continuum is the lower limit on the ionizing flux .
there are various corrections which could make the sed harder and hence increase the lower limits on [ c / h ] and [ n / h ] . finally , we comment that these photoionization models are highly simplified , and there are some indications that the real absorbers are more complicated than the assumed models .
consider the doppler parameters derived from profile fitting in table 1 .
these @xmath10-values can be expressed as a sum with a term due to thermal motions ( @xmath110 ) and a term which includes all non - thermal motions ( @xmath111 ) such as bulk motions and turbulence , @xmath112 where @xmath113 is the mass and @xmath74 is the atomic weight of the element and @xmath10 is measured in . for the photoionized models presented above , we assumed that all of the absorption lines of various species in a given component arise in a constant density gas slab .
therefore one might expect different elements to have the same values for @xmath111 and @xmath71 , and higher mass elements should have lower @xmath10-values . in principle one can solve equation 4 for @xmath111 and @xmath71 given @xmath10-values for two or more elements .
however , in this associated absorber profile fitting yields @xmath10 = [email protected] for and @xmath10 = [email protected] for .
this is not consistent with the assumption that @xmath111 and @xmath71 are the same for the and absorption lines because this would require an unphysical negative temperature .
this suggests that ( 1 ) there is some problem with the doppler parameter measurements or uncertainties , or ( 2 ) the properties of the gas causing the absorption are different from the properties of the gas producing the absorption .
the disparity between @xmath10 ( ) and @xmath10 ( ) can not be blamed on the profile fitting procedure ; attempts to fit the and profiles with compatible @xmath10-values , assuming @xmath111 and @xmath71 are the same , yield unacceptably bad fits . the nitrogen line would be broader than the line if is strongly saturated , but profile fitting should account for such saturation and yield consistent @xmath10-values . therefore in this associated absorber ,
the gas and gas do not have the same non - thermal motions and temperature .
it is possible that the absorption and _ some _ of the absorption occurs in the same gas cloud with additional absorption arising in a separate cloud which does not produce absorption .
some support for this suggestion is provided by the fact that the profiles are affected by unresolved saturation and/or incomplete covering of the qso flux source , while the profiles show no evidence of these effects ( see figure 5 ) .
future studies of this type of qso absorber may require more intricate models .
based on a survey of absorption lines in the spectra of 103 qsos , steidel & sargent ( 1992 ) conclude that there is paucity of _ weak _ lines , and they assert that `` at least 80% of the absorbers are accounted for in a sample sensitive to @xmath114 = 0.3 '' ( @xmath115 is the minimum rest equivalent width in the sample ) . we draw a different ( albeit tentative ) conclusion from our high resolution high s / n spectra of hs 1700 + 6416 and hs 1946 + 7658 ( tripp et al .
the sensitivity of our spectra of these qsos is adequate for detection of lines with restframe equivalent width @xmath116 as small as 0.03 ( at the 4@xmath20 level ) over a total redshift path @xmath55 = 2.56 .
we detect six definite absorbers in the spectra of these qsos , and only one of these lines has @xmath117 0.3 . to expand the sample size
, we can combine our data with the similar quality spectra of pks 0424131 ( @xmath1 = 2.1657 ) and q 0450131 ( @xmath1 = 2.2535 ) obtained by petitjean et al .
( 1994)20 and sufficient s / n for detection of all lines with @xmath118 0.03 . ] . from this combined sample of four quasars , we derive a mean number of absorbers per unit redshift of @xmath12 = [email protected] . from their sample of absorbers with @xmath118 0.3 ,
steidel & sargent ( 1992 ) derive @xmath12 = [email protected] .
it is important to emphasize that @xmath12 derived from our sample of four quasars is highly uncertain due to the small sample .
however , womble ( 1995 ) has recently detected a similar excess of _ weak _ systems ( @xmath119 ) in a high s / n high resolution keck spectrum of pg 1634 + 706 ( @xmath1 = 1.334 ) .
furthermore , steidel & sargent find that @xmath12 increases smoothly as the minimum @xmath116 in the sample decreases from 1.0 to 0.3 ( see their figure 5 ) .
this also suggests that @xmath12 is dominated by weak systems .
this excess of weak lines may indicate that the cross section of the absorbers is large ; womble estimates that the required radius is @xmath590 @xmath120 kpc ( @xmath121 = h@xmath122/100 mpc@xmath86 ) for lines with @xmath123 .
an excess of weak absorbers may also have implications for the study of galaxies _ selected _ because they produce absorption in the spectrum of the background qso .
for example , steidel , dickinson , & persson ( 1994 ) find that their sample of absorbers are mostly associated with normal luminous galaxies , but their sample may be biased against underluminous galaxies because of the lower equivalent width cutoff ( they select systems with @xmath118 0.3 for follow - up imaging and redshift measurements ) . with the echelle spectrograph on the keck telescope
, it should be straightforward to confirm or refute this excess of weak absorbers with a larger sample . as discussed by petitjean & bergeron ( 1990 ) and steidel & sargent ( 1992 ) , there are some indications that the fractions of absorbers which are weak and strong may evolve with redshift in the sense that there are more strong lines at higher redshift .
this probably is not the reason that the @xmath12 obtained by steidel & sargent is lower than the @xmath12 we have derived because the redshift range probed by steidel & sargent , 0.2 @xmath124 2.2 , is similar to the redshift range probed by our sample , 0.4 @xmath124 2.1 .
the _ hst _ fos observation of hs 1700 + 6416 obtained by reimers et al .
( 1992 ) is ground - breaking work , and the spectrum has provided an excellent reconnaissance of this extraordinary sight line .
however , the data have some serious limitations due to the low resolution ( fwhm @xmath5300 ) of the fos , and we suspect that the abundances derived from the fos data by vogel & reimers ( 1995 ) are unreliable .
several problems are introduced by the low resolution of the data combined with the high redshift of the qso ( which leads to a very high line density in the fos spectral region including metal lines , he lines , and many lyman series lines ) : ( 1 ) in many cases , the absorption lines are strongly blended with several lines of different species at different redshifts ( see figure 4 in vogel & reimers 1995 ) , and extracting accurate column densities from the blends is rather difficult .
( 2 ) as lipman , pettini , & hunstead ( 1995 ) have noted ( see their figure 1 ) , narrow lines are unlikely to be detected in the fos spectrum , and the fos line list will be biased toward strongly saturated lines which do not provide reliable column densities .
( 3 ) quasar absorption line profiles frequently show multiple components which will be completely unresolved at the fos resolution , and consequently absorption by low and high ionization species which appear to be at the same redshift in the fos data may not actually be associated ( i.e. , the low and high ion absorption occur in different unresolved components ) .
confusion due to multiple unresolved components will be especially severe in the absorption line complex at 2.432 @xmath3 @xmath3 2.441 ( see 4.10 ) .
this paper confirms that the vogel & reimers ( 1995 ) abundances may be problematic .
for example , we estimate that the strongest component of the ll absorber at = 2.3150 has [ si / h ] @xmath6 0.95 and [ al / h ] @xmath6 0.96 and is best fit with [ m / h ] @xmath4 0.45 ( see 5.2.1 ) , while vogel & reimers report [ n / h ] @xmath3 1.04 and [ o / h ] = 1.52
. the low resolution of the fos also yields redshifts which may not be accurate enough to facilitate line identifications .
consider the ll system for which vogel & reimers give = 1.8465 .
we detect the doublet at high significance levels at = 1.84506 ( lines 44 and 45 in figure 1 ) , which differs from the vogel & reimers redshift by @xmath5150 .
this amount of uncertainty in @xmath14 can make line identification difficult given the high density of lines , and indeed petitjean et al .
( 1996 ) have pointed out several lines that vogel & reimers may have incorrectly identified because they were not aware of the associated absorption system of hs 1700 + 6416 .
similarly , the extended x - ray sources near the sight line ( see 1 ) may produce absorption lines , at = 0.225 in the case of abell 2246 , and these low redshift absorption lines could have been misidentified as well .
these problems are generic drawbacks of low resolution data which should be greatly alleviated when hs 1700 + 6416 is observed with the space telescope imaging spectrograph .
the paper is summarized as follows .
\(1 ) we have obtained a high s / n high resolution ( 20 fwhm ) spectrum of hs 1700 + 6416 ( @xmath1 = 2.72 ) with complete spectral coverage from 4300 to 8350 .
thirteen absorption systems with metals are detected in this optical spectrum , including six absorbers which cause optically thin lyman limit absorption in the _ hst _ spectrum obtained by reimers et al .
( 1992 ) .
\(3 ) a dense cluster of doublets is detected at 2.432 @xmath3 @xmath3 2.441 , which corresponds to a displacement of @xmath524000 from the qso emission redshift . despite the large displacement from @xmath1 , two pairs of doublets are apparently line locked , i.e.
, the strong line of one absorber is aligned with the weak line of a different doublet , to within 10 in one case and to within 25 in the other case
. this may be the remnant ( or precursor ) of a bal outflow . however , it is possible that these alignments are coincidental rather than true line locking .
\(4 ) the column density ratios of high ionization stages in the lyman limit absorber at = 2.3150 differ significantly from the ratios observed in the gaseous halo of the milky way . for example , @xmath21()/@xmath21 ( ) @xmath6 10 in two components of this absorber .
this high ionic ratio in the ll absorber is not expected in cooling galactic fountain gas but could be produced in supernova remnants , conductive interfaces between warm and hot gas , turbulent mixing layers , or from photoionization by the diffuse extragalactic background radiation .
\(5 ) photoionization modeling of the lyman limit absorber at = 2.3150 places conservative lower limits on [ si / h ] and [ al / h ] . the model
which provides the best fit has [ m / h ] @xmath4 0.45 .
this is substantially larger than the metallicity derived by vogel & reimers ( 1995 ) from the _ hst _ fos spectrum .
the discrepancy is probably due to the low resolution of the fos data which causes severe blending and saturation problems .
\(6 ) an associated absorber at = 2.7125 shows and strong lines .
the lines are unsaturated , but the apparent column density profiles indicate that is affected by unresolved saturation , or the absorbing gas does not completely cover the qso flux source .
\(7 ) despite coverage of several low ionization stages of abundant elements with high @xmath35 values , no low ion absorption is evident in the associated absorber . also , the high ion column density ratios differ significantly from the ratios observed in the milky way ism .
this indicates that the absorber is highly ionized , presumably by the qso .
photoionization modeling using the qso spectral energy distribution for the input radiation field requires [ n / h ] @xmath6 0.65 and [ c / h ] @xmath6 0.82 .
these are firm lower limits which could be increased by plausible corrections to the ionizing continuum shape .
however , profile fitting yields @xmath10 ( ) = [email protected] and @xmath10 ( ) = [email protected] , and therefore the gas which produces the absorption does not have the same temperature and/or non - thermal motions as the gas , and a more complex photoionization model may be required for this absorber .
\(8 ) we briefly examine the number of absorbers detected per unit redshift .
the line density appears to be dominated by weak lines with @xmath13 0.3 , but the sample suffers from small number statistics .
it is a pleasure to thank fred hamann for several excellent comments which significantly improved this paper .
we acknowledge gary ferland for use of his photoionization code cloudy , and we thank ken lanzetta for sharing software which was used for profile fitting .
kathy romer and david cohen provided helpful comments on the rosat data and the nature of the extended x - ray sources near hs 1700 + 6416 .
t.m.t . received support from nasa grant ngt51003 .
this research is also supported by nasa through grant hf1062.01 - 94a for l.l . from the space telescope science institute , which is operated by the association of universities for research in astronomy ,
, for nasa under contract nas5 - 26555 .
b.d.s . acknowledges support from nasa grant nag5 - 1852 .
arav , n. , korista , k. t. , barlow , t. a. , & begelman , m. c. 1995 , nature , 376 , 576 arav , n. , li , z .- y . , & begelman , m. c. 1994 , apj , 432 , 62 barlow , t. a. , & sargent , w. l. w. 1997 , aj , 113 , 136 beaver , e. a. , et al .
1991 , apj , 377 , l1 bechtold , j. 1995 in qso absorption lines , ed . g. meylan , ( berlin : springer ) , 299 benjamin , r. a. , & shapiro , p. r. 1997 , apjs , submitted bergeron , j. , & stasiska , g. 1986 , a&a , 169 , 1 cowie , l. l. , songaila , a. , kim , t .- s . , &
hu , e. m. 1995 , aj , 109 , 1522 edgar , r. j. , & chevalier , r. a. 1986 , apj , 310 , l27 ferland , g. j. 1993 , university of kentucky department of physics and astronomy internal report foltz , c. b. , chaffee , f. h. , weymann , r. j. , & anderson , s. f. 1988 , in quasar absorption lines : probing the universe , ed . j. c. blades , d. a. turnshek , & c. a. norman ( cambridge : cambridge university press ) , 53 foltz , c. b. , weymann , r. j. , morris , s. l. , & turnshek , d. a. 1987 , apj , 317 , 450 grevesse , n. , & anders , e. 1989 , in cosmic abundances of matter , ed . c. j. waddington , ( new york : american institute of physics ) , 1 hamann , f. 1997 , apjs , in press hamann , f. , barlow , t. a. , beaver , e. a. , burbidge , e. m. , cohen , r. d. , junkkarinen , v. , & lyons , r. 1995 , apj , 443 , 606 hamann , f. , barlow , t. a. , & junkkarinen , v. 1997a , apj , in press hamann , f. , barlow , t. a. , junkkarinen , v. , & burbidge , e. m. 1997b , apj , in press hamann , f. , & ferland , g. 1993 , apj , 418 , 11 horne , k. 1986 , pasp , 98 , 609 jakobsen , p. , boksenberg , a. , deharveng , j. m. , greenfield , p. , jedrzejewski , r. , & paresce , f. 1994 , nature , 370 , 35 jannuzi , b. t. , et al .
1996 , apj , 470 , l11 jenkins , e. b. 1987 , in interstellar processes , ed .
d. j. hollenbach & h. a. thronson ( dordrecht : reidel ) , 533 jenkins , e. b. 1996 , apj , 471 , 292 lanzetta , k. m. , & bowen , d. v. 1992 , apj , 391 , 48 lipman , k. , pettini , m. , & hunstead , r. w. 1995 , in qso absorption lines , ed . g. meylan , ( berlin : springer ) , 89 lockman , f. j. , & savage , b. d. 1995 , apjs , 97 , 1 lyons , r. w. , cohen , r. d. , junkkarinen , v. t. , burbidge , e. m. , & beaver , e. a. 1995 , aj , 110 , 1544 lu , l. , sargent , w. l. w. , & barlow , t. a. , churchill , c. w. , & vogt , s. 1996 , apjs , 107 , 475 lu , l , savage , b. d. , tripp , t. m. , & meyer , d. m. 1995 , apj , 447 , 597 madau , p. 1992 , apj , 389 , l1 mcwilliam , a. , preston , g. , sneden , c. , & searle , l. 1995 , aj , 109 , 2757 meyer , d. m. , lanzetta , k. m. , & wolfe , a. m. 1995 , apj , 451 , l13 mo , h. j. , & miralda - escud , j. 1996 , apj , 469 , 589 mller , p. , & jakobsen , p. 1990 , a&a , 228 , 299 morris , s. l. , weymann , r. j. , foltz , c. b. , turnshek , d. a. , shectman , s. , price , c. , & boroson , t. a. 1986 , apj , 310 , 40 morton , d. c. 1991 , apjs , 77 , 119 murray , n. , chaing , j. , grossman , s. a. , & voit , g. m. 1995 , apj , 451 , 498 perry , j. j. , burbidge , e. m. , & burbidge , g. r. 1978 , pasp , 90 , 337 petitjean , p. , & bergeron , j. 1990 , 231 , 309 petitjean , p. , rauch , m. , & carswell , r. f. , 1994 , a&a , 291 , 29 petitjean , p. , riediger , r. , & rauch , m. 1996 , a&a , 307 , 417 pettini , m. , king , d. l. , smith , l. j. , & hunstead , r. w. 1997 , apj , submitted pettini , m. , smith , l. j. , hunstead , r. w. , & king , d. l. 1994 , apj , 426 , 79 picard , a. , & jakobsen , p. 1993 , a&a , 276 , 331 prochaska , j. x. , & wolfe , a. m. 1996 , apj , 470 , 474 , 140 rauch , m. , sargent , w. l. w. , womble , d. s. , & barlow , t. a. 1996 , apj , 467 , l5 reimers , d. , bade , n. , schartel , n. , hagen , h .- j . , engles , d. , & toussaint , f. 1995a , a&a , 296 , l49 reimers , d. , clavel , j. , groote , d. , engels , h. j. , naylor , t. , wamsteker , w. , & hopp , u. 1989 , a&a , 218 , 71 reimers , d. , rodriguez - pascual , p. , hagen , h .- j . , & wisotzki , l. 1995b , a&a , 293 , l21 reimers , d. , vogel , s. , hagen , h .- j .
, engels , d. , groote , d. , wamsteker , w. , clavel , j. , & rosa , m. r. 1992 , nature , 360 , 561 robertson , j. g. 1986 , pasp , 98 , 1220 rodrguez - pascual , p. m. , de la fuente , a. , sanz , j. l. , recondo , m. c. , clavel , j. , santos - lle , m. , & wamsteker , w. 1995 , apj , 448 , 575 romer , a. k. 1996a , private communication romer , a. k. , ulmer , m. p. , nichol , b. c. , holden , b. p. , burke , d. , & collins , c. a. 1996b , baas , 28 , 853 ryan , s. g. , norris , j. e. , & beers , t. c. 1996 , apj , 471 , 254 sanz , j. l. , clavel , j. , naylor , t. , & wamsteker , w. 1993 , mnras , 260 , 468 sargent , w. l. w. , boksenberg , a. , & steidel , c. c. 1988 , apjs , 68 , 539 savage , b. d. , & sembach , k. r. 1991 , apj , 379 , 245 savage , b. d. , & sembach , k. r. 1996 , , 34 , 279 savaglio , s. , dodorico , s. , & mller , p. 1994 ,
a&a , 281 , 331 scoville , n. , & norman , c. 1995 , apj , 451 , 510 sembach , k. r. , & savage , b. d. 1992 , apjs , 83 , 147 sembach , k. r. , savage , b. d. , & tripp , t. m. 1997 , apj , in press siemiginowska , a. , bechtold , j. , tran , k .- v . , & dobrzycki , a. 1996 , in science with the hubble space telescope - ii , ed .
p. benvenuti , f. d. macchetto , & e. j. schreier , stsci / st - ecf workshop , 181 sofia , u. j. , cardelli , j. a. , & savage , b. d. 1994 , apj , 430 , 650 spitzer , l. 1978 , physical processes in the interstellar medium , ( new york : john wiley & sons ) steidel , c. c. , 1990 , apjs , 74 , 37 steidel , c. c. , dickinson , m. , & persson , s. e. 1994 , apj , 437 , l75 steidel , c. c. , & sargent , w. l. w. 1989 , apj , 343 , l33 steidel , c. c. , & sargent , w. l. w. 1992 , apjs , 80 , 1 sutherland , r. s. , & dopita , m. a. 1993 , apjs , 88 , 253 tripp , t. m. , bechtold , j. , & green , r. f. 1994 , apj , 433 , 533 tripp , t. m. , lu , l. , & savage , b. d. 1996 , apjs , 102 , 239 turnshek , d. a. 1988 in quasar absorption lines : probing the universe , ed . j. c. blades , d. a. turnshek , & c. a. norman ( cambridge : cambridge university press ) , 17 turnshek , d. a. , weymann , r. j. , & williams , r. e. 1979 , apj , 230 , 330 tytler , d. , fan , x .- m . , burles , s. , cottrell , l. , davis , c. , kirkman , d. , & zuo , l. 1995 , in qso absorption lines , ed . g. meylan , ( berlin : springer ) , 289 verner , d. a. , tytler , d. , & barthel , p. d. 1994 , apj , 430 , 186 viegas , s. m. 1995 , mnras , 276 , 268 vogel , s. , & reimers , d. 1995 , a&a , 294 , 377 wampler , e. j. 1991 , 368 , 40 wampler , e. j. , bergeron , j. , & petitjean , p. 1993 , a&a , 273 , 15 weymann , r. j. , carswell , r. f. , & smith , m. g. 1981 , ara&a , 19 , 41 wheeler , j. c. , sneden , c. , & truran , j. w. 1989 , ara&a , 27 , 279 womble , d. s. 1995 in eso workshop on qso absorption lines , ed . g. meylan , ( berlin : springer ) , 157 | we have obtained a high signal - to - noise ( 30 @xmath0 s / n @xmath0 70 ) high resolution ( fwhm = 20 ) spectrum of the radio - quiet qso hs 1700 + 6416 ( @xmath1 = 2.72 ) with the echelle spectrograph on the kpno 4 m telescope .
we detect 13 metal systems in the optical spectrum of this qso , including six systems with associated optically thin lyman limit absorption in the _ hst _ spectrum obtained by reimers et al .
we use the apparent column density technique and profile fitting to measure the heavy element column densities and to evaluate the impact of unresolved absorption saturation .
profile fitting indicates that four of the systems are narrow with @xmath2 8 , which implies that these absorbers are relatively cool and probably photoionized .
the dense cluster of doublets at 2.432 @xmath3 @xmath3 2.441 shows the weak line of one absorber apparently aligned with the strong line of a different doublet , i.e. , line locked , for two pairs of absorbers .
line locking has been detected previously in @xmath4 @xmath1 absorbers where radiation pressure is likely to play a role , but it is surprising in this case since this complex is displaced by @xmath524000 from the qso emission redshift .
this may be the remnant ( or precursor ) of a broad absorption line ( bal ) outflow .
however , it is possible that these alignments are chance alignments rather than true line locking .
the high ion column density ratios in the multicomponent lyman limit absorber at = 2.3150 suggest that the ionization conditions in this absorber differ significantly from the conditions in the gaseous halo of the milky way . from photoionization models we derive [ si / h ]
@xmath6 0.95 and [ al / h ] @xmath6 0.96 for the stongest component of this absorber . these are conservative lower limits derived from lower ionization stages only ; photoionization models in agreement with the observed low and high ionization stages require [ m / h ] @xmath4 0.45 .
in contrast , vogel & reimers derive [ n / h ] @xmath3 1.04 and [ o / h ] = 1.52 for this absorber .
we suggest that the discrepancy is due to the low resolution of the vogel & reimers data ( fwhm @xmath4 300 ) which introduces serious blending and saturation problems .
the photoionized model with [ m / h ] = 0.45 has a particle density @xmath7 0.02 @xmath8 , a size of a few hundred pc , and a mass of roughly 1 @xmath9 , assuming the absorber is spherical .
we detect unsaturated and rather strong `` associated '' absorption at = 2.7125 .
the apparent column density of the weak 1242.8 line is greater than the apparent column density of the stronger 1238.8 line in this absorber , which indicates that the profile is affected by unresolved saturation or that the absorbing gas does not completely cover the qso emission source .
if the latter interpretation is correct , then the associated absorbing gas must be close to the qso .
we have used the observed spectral energy distribution of the qso , corrected for intervening lyman limit absorption , for photoionization modeling of the associated absorber , and we derive [ n / h ] @xmath6 0.65 and [ c / h ] @xmath6 0.82 . other corrections ( e.g. , to account for dust in intervening absorbers or ` lyman valley ' attenuation ) will make the spectral energy distribution harder and increase the metallicity estimates .
however ,
the absorption profiles suggest that the constant density single slab model is too simplistic : we obtain @xmath10 ( ) = [email protected] and @xmath10 ( ) = [email protected] from profile fitting , and therefore the gas which produces the absorption does not have the same temperature and non - thermal motions as the gas .
finally , we briefly examine the number of systems detected per unit redshift , and we tentatively conclude that @xmath12 is dominated by weak lines with @xmath13 0.3 . |
understanding when and how the most massive galaxies in the universe formed is one of the most outstanding problems in cosmology and galaxy formation .
massive galaxies are predicted in cold dark matter based models of structure formation to form gradually with time through the merging of smaller systems ( e.g , .
white & rees 1978 ) . while there is some evidence for this process ( e.g. , le fevre et al .
2000 ; patton et al .
2002 ; conselice et al .
2003a , b ; bridge et al . 2007 ) , many details are still lacking . alternatively ,
massive galaxies , which are mostly ellipticals in today s universe ( e.g. , conselice 2006a ) , may have formed in a very rapid collapse of gas ( e.g. , larson 1974 ) .
observational evidence suggests that passively evolving massive galaxies exist at @xmath24 , as well as at even early times , @xmath25 ( e.g. , dunlop et al . 1996 ; spinrad et al . 1997 ; fontana et al . 2004 ; daddi et al . 2004
; glazebrook et al . 2004 ; treu et al .
2005 ; labbe et al .
2005 ; papovich et al . 2006 ; grazian et al . 2006 ; kriek et al . 2006 ; lane et al . 2007 ) .
these systems are a subset of the massive galaxy population at high redshift , yet their nature , and whether or not they can account for all of the most massive galaxies in today s universe is still unknown .
we also do not yet know if these massive galaxies are still forming when the universe was half its current age ( at @xmath24 ) .
massive galaxies are largely the test - bed for galaxy models , and therefore understanding their evolution observationally is an important test of the physics behind galaxy formation .
a major part of this problem is determining when massive galaxies form .
some studies claim that massive galaxies are all in place by @xmath24 ( e.g. , glazebrook et al .
however , since star formation and merging activity has been seen in ellipticals from @xmath26 to @xmath24 ( stanford et al . 2004 ; lin et al .
2004 ; teplitz et al .
2006 ) , a definite answer remains elusive . if it were possible to date every star in nearby massive galaxies , we could in principle determine the formation epoch and time - scales of these systems by examining their individual stars .
we can not however resolve all the stars in galaxies , and their integrated stellar properties , such as colours , become degenerate after about 5 gyrs ( e.g. , worthey 1994 ) .
stellar ages in massive galaxies also do not necessarily correlate with the assembly of mass through , for example , merging activity ( conselice 2006b ; de lucia et al .
2006 ; trujillo et al .
an alternative approach towards understanding massive galaxies and their evolution is empirically measuring the number densities , morphologies , star formation rates , and stellar masses of the most massive systems at some fiducial time and to compare these to similar quantities at different times ( redshifts ) , and with models .
this has however traditionally been a challenging problem since the most massive galaxies are typically very red , either due to evolved stellar populations , or by formation in dusty starbursts ( e.g. , graham & dey 1996 ) .
because of the shape of the spectral energy distributions of evolved galaxies , they are identifiable at @xmath24 due to their extremely red near - infrared to optical colours . these objects are sometimes known as extremely red objects ( elston et al . 1988 ) , which sample a fraction of @xmath27 massive galaxies ( e.g. , moustakas et al . 2004 ; conselice et al .
conversely if one can obtain complete redshift samples , and combine these with deep near - infrared imaging , it is possible to locate the most massive galaxies up to @xmath15 without relying on assumptions concerning their spectral energy distributions , morphologies , or star formation histories .
tracing the total evolution of stellar mass through cosmic time has been attempted in several fields , including the hubble deep fields ( dickinson et al .
2003 ) , the gems / combo-17 field ( borch et al .
2006 ) , and the goods fields ( bundy et al . 2005 ) .
recently there have also been claims for the determination of the stellar mass evolution of galaxies at @xmath28 ( dickinson et al .
2003 ; fontana et al .
2004 ; bundy et al . 2006
; rudnick et al . 2006 ) .
some of these studies claim that many massive galaxies are formed by @xmath29 , and that these systems prove to be a complication for cold dark matter models of structure formation ( e.g. , glazebrook et al .
there are however several limitations to these studies .
the two most important being the use of photometric redshifts , and the small field sizes used in these earlier studies .
what is now needed is a large spectroscopic survey over a large area , with deep near - infrared imaging to measure accurate stellar masses of galaxies found when the universe was younger than half its current age .
we provide such a study in this paper by utilising @xmath13 1.5 deg@xmath2 palomar nir imaging of the deep2 fields ( davis et al .
we also go beyond previous work by utilising ancillary data on a purely mass selected sample to determine the morphologies , colours , and star formation rates of massive galaxies , and how these properties evolved down to @xmath16 .
our galaxies are selected from the palomar observatory wide - field infrared survey , which is designed to determine robustly the stellar mass evolution of galaxies , and identify the most massive galaxies , at @xmath30 .
previously , we have utilised spectroscopic redshifts from the deep2 redshift survey ( davis et al . 2003 ; faber et al . 2005 ) to probe the mass function down to m@xmath31 + 3 ( bundy et al .
2005 , 2006 ) , and to study the properties of distant red galaxies ( conselice et al .
our goal in this paper is specifically to understand the evolutionary history and properties of high redshift massive galaxies , defined as systems with m@xmath0 at @xmath20 .
one of our primary objectives is to study systems at @xmath32 , where we have ancillary information from other telescopes .
this paper is organised as follows .
2 is a presentation of our data .
3 gives an overview of the methodology used to determine the quantities used in this paper , while 4 is a discussion of the properties and evolution of massive galaxies at @xmath19 .
5 is a discussion of our results in terms of galaxy evolution scenarios , and 6 is a summary . throughout this paper
we use a standard cosmology of h@xmath33 km s@xmath34 mpc@xmath34 , and @xmath35 .
all quoted magnitudes are in the vega systems , unless otherwise specified .
the galaxies we study in this paper consists of those in the fields covered by the palomar observatory wide - field infrared survey ( powir , table 1 ) , excluding the goods north field discussed in bundy et al .
the powir survey was designed to obtain deep k - band and j - band data over a significant ( @xmath131.5 deg@xmath36 ) area .
observations were carried out between september 2002 and october 2005 over a total of @xmath37 nights .
this survey covers the goods field north ( giavalisco et al .
2004 ; bundy et al .
2005 ) , the extended groth strip ( davis et al .
2006 ) , and three other fields that the deep2 team has observed with the deimos spectrograph ( davis et al .
the total area we cover in the k - band is 5524 arcmin@xmath2 = 1.53 deg@xmath2 , with half of this area imaged in the j - band .
our goal depth was k@xmath38 , although not all fields are covered this deep , but all have 5@xmath3 depths between k@xmath39 for point sources , measured in a 2 diameter aperture .
table 1 lists the deep2 fields , and the area we have imaged in each . for our purposes we will abbreviated the fields covered as : egs ( extended groth strip ) , field 2 , field 3 , and field 4 .
.the palomar fields and wirc pointings areas [ cols="<,^,^,^",options="header " , ] all of our k@xmath40-band data were acquired utilising the wirc camera on the palomar 5 meter telescope .
wirc has an effective field of view of @xmath41 , with a pixel scale of 0.25pixel@xmath34 .
our total survey contains 75 wirc pointings . during the k@xmath40-band observations we used 30 second integrations , with four exposures per pointing .
the j - band observations were taken with 120 second exposures per pointing .
typical total exposure times were between one and two hours for both bands .
our reduction procedure follows standard method for combining nir ground - based imaging , and is described in more detail in bundy et al .
the resulting seeing fwhm in the k@xmath40-band imaging ranges from 0.8 to 1.2 , and is typically 1.0 . photometric calibration was carried out by referencing persson et al .
( 1998 ) standard stars during photometric conditions .
the final images were made by combining individual mosaics obtained over several nights .
the k@xmath40-band mosaics are comprised of coadditions of @xmath42 second exposures dithered over a non - repeating 7.0 pattern .
the images were processed using a double - pass reduction pipeline we developed specifically for wirc . for galaxy detection and photometry
we utilised the sextractor package ( bertin & arnouts 1996 ) .
false artifacts are removed through sextractor flags which identify sources that do not have normal galaxy or stellar profiles . from this
we construct a k - selected sample , which is then cross - referenced with the deep2 redshift catalog .
other data used in this paper consists of : optical imaging from the cfht over all fields , mips imaging from the _ spitzer space telescope _ , imaging from the advanced camera for surveys ( acs ) on hubble , chandra x - ray imaging , and spectroscopy from the deimos spectrograph on the keck ii telescope ( davis et al .
a summary of these ancillary data sets , which are mostly within the extended groth strip , are presented in davis et al .
( 2006 ) .
the optical imaging of our fields comes from the cfht 3.6-m , and consists of data in the b , r and i bands taken with the cfh12k camera - a 12,288 @xmath43 8,192 pixel ccd mosaic with a pixel scale of 0.21 .
the integration times for these observations are 1 hour in @xmath44 and @xmath45 , and 2 hours in @xmath46 , per pointing , with a r - band 5 @xmath3 depth of @xmath47 , and similar depths at @xmath44 and @xmath46 ( coil et al .
the details of the data reduction for this data is described in coil et al .
( 2004b ) . from this imaging data
a r@xmath48 = 24.1 magnitude limit was used for determining targets for the deep2 spectroscopy .
the details for how these imaging data were acquired and reduced are covered in coil et al .
( 2004b ) .
the seeing for the optical imaging is roughly the same as that for the nir imaging , and we measure photometry consistently using a 2 diameter aperture . the keck spectra were acquired with the deimos spectrograph ( faber et al .
2003 ) as part of the deep2 redshift survey ( davis et al .
the selection for targets for the deep2 spectroscopy was based on the optical properties of the galaxies detected in the cfht photometry , with the basic selection criteria @xmath49 .
spectroscopy in the egs was acquired through this magnitude limit , with no strong colour cuts applied to the selection .
objects in fields 2 - 4 were selected for spectroscopy based on their position in @xmath50 vs. @xmath51 colour space to focus on galaxies at redshifts @xmath52 .
the total survey includes over 30,000 galaxies with a secure redshift , with about a third of these in the egs field . in all fields
the sampling rate for galaxies that meet the selection criteria is 60% .
the deimos spectroscopy was obtained using the 1200 line / mm grating , with a resolution r @xmath53 covering the wavelength range 6500 - 9100 .
redshifts were measured through an automatic method comparing templates to data , and we only utilise those redshifts measured when two or more lines were identified , providing very secure measurements .
roughly 70% of all targeted objects result in secure redshifts .
most of the redshift failures are galaxies at higher redshift , @xmath54 ( steidel et al .
2004 ) , where the [ oii ] @xmath553727 line leaves the optical window .
the acs imaging over the egs field covers a 10.1@xmath43 70.5 strip , for a coverage area of 0.2 deg@xmath2 .
this acs imaging is discussed in lotz et al .
( 2006 ) , and is briefly described here .
the imaging consists of 63 tiles imaged in both the f606w ( v ) and f814w ( i ) bands .
the 5-@xmath3 depths reached in these images are v = 26.23 ( ab ) and i = 27.52 ( ab ) for point sources , and about two magnitudes brighter for extended objects .
we further utilise spitzer mips 24@xmath56 m imaging , and chandra x - ray imaging of the egs field ( nandra et al .
2007 ) to determine the star forming and agn properties of our sample . the mips
24@xmath56 m imaging is part of the irac team gto program ( papovich et al .
our procedure was to use a catalog of 24@xmath56 m sources matched to our list of massive galaxies within 1 , and brighter than @xmath5760@xmath58jy .
this ensures that the source matching is reliable , and that the mips detections are significant to @xmath59 @xmath3 confidence .
we then convert these 24@xmath56 m fluxes into total infrared fluxes utilising the fact that there is a good correlation between fluxes at 15@xmath56 m and the total ir flux ( le floch et al . 2005 ) .
we use various templates to convert our observed 24@xmath56 m flux into a total ir flux , from which we compute star formation rates ( 4.4.2 ) .
our matching procedures for these catalogs progressed in the manner described in bundy et al .
( 2006 ) with the k - band catalog serving as our reference .
we then match the optical catalogs and spectroscopic catalogs to the k - band catalog , after correcting for astrometry by referencing all systems to 2mass stars . the mips catalogs and x - rays catalogs are likewise matched in a similar manner .
we estimate photometric errors , and the k - band detection limit of each k - band image by randomly inserting fake objects of known magnitude , surface brightness profile , and size into each image , and then recovering these simulated objects with the same detection parameters used for real objects .
the most basic simulations we perform are done by simply inserting objects with gaussian profiles with a fwhm of 13 to approximate the shape of slightly extended , distant galaxies .
we then retrieve these objects using the same sextractor parameters used for the original galaxy detections .
we use these simulated retrievals to determine the upper limit completeness of our sample , as well as constrain errors on our photometric measurements .
actual galaxy profiles however have extended envelopes that could be more difficult to measure accurately than compact gaussian profiles ( e.g. , graham et al . 2005 ) . to determine the detection and photometric fidelity of our images in more detail , we create two sets of 10,000 mock galaxies , each with an intrinsic exponential and de vaucouleurs profile .
these simulated galaxies have properties which are uniformly distributed as follows : k@xmath40 band total magnitudes between 15.5 and 20.5 mag , effective radius r@xmath60 between 0.0625 and 3.75 arcsec ( equivalent to 0.5 - 30 kpc at z@xmath131 ) , and ellipticities between 0 and 0.8 .
these simulated sources are placed randomly on our palomar images , and extracted in the same manner as the real source detections .
we construct from these simulations detection maps that reveal the fraction of input artificial sources detected per input magnitude and input log ( re ) bin ( see figure 1 ) .
as expected , galaxies with a de vaucouleurs profile , which are more centrally concentrated , are easier to detect at a given magnitude .
note that the detection fractions plotted in figure 1 are independent of output magnitude and size , and are plotted based on their input parameters .
we also estimated systematic errors in measuring magnitudes due to our detection method by using the same simulations ( see figure 2 ) . to do this
, we compute the magnitude difference between the recovered and input sources within the input magnitude and size bin . to estimate the output magnitudes , we use exactly the same methods used to find and measure photometry of our actual galaxies .
the decreased surface brightness of objects , at a given input magnitude , results in output magnitudes systematically fainter at larger effective radii .
in particular , due to the larger tail of the de vaucouleurs profiles , at a given magnitude and size , the difference in magnitude is larger for the de vaucouleurs profiles than for the exponential models . at the range of sizes for our sample objects , which range from 0.3-0.7 ( trujillo et al .
2007 , submitted ) , we find that the magnitude and recovery fraction are essential 100% of their simulated value .
depending on the sizes of our galaxies , we could be missing as much as 0.2 mag in the recovered k - band light .
we will address this in more detail in 4.2.1 .
we utilise several methods to study the properties of our k - band selected , stellar mass limited sample .
the outline of the procedure is described below .
the first step in this process , after reducing the k - band imaging , is to create k - selected catalogs .
we make these catalogs using sextractor , optimised for detection and splitting , with both topics discussed in bundy et al .
( 2006 , 2007 ) .
we then utilise the k - band imaging , combined with the optical imaging , to compute stellar masses from which our sample is selected .
we then match these massive galaxies to mips , x - ray and hubble sources from which we derive the physical features , and the evolution of these systems . within the total k - band area of our survey ( 1.53 deg@xmath2 )
we detect 61,489 sources , after removing false artifacts .
most of these objects ( 92% ) are at k @xmath61 , while 68% are at k @xmath62 , and 37% are at k @xmath63 . in total
there are 38,613 objects fainter than k @xmath64 in our sample . out of our total k - band population ,
10,693 objects have secure spectroscopic redshifts from the deep2 redshift survey ( davis et al .
we supplement these by 37,644 photometric redshifts within the range @xmath65 ( 3.2 ) . because our spectroscopy is r - band selected , while our stellar masses are based on k - band detections , we are required to divided our sample into different sub - samples .
we construct a `` primary sample '' consisting of those galaxies with both spectroscopic redshifts and k - band detections . by construction
, these galaxies also have optical magnitudes @xmath49 .
the matching between our k - band catalogue , and the deep2 catalogue was done with a 1 tolerance .
this results in a very low spurious rate of 1 - 2% due to the low surface densities in both catalogues .
most of the galaxies with spectroscopic redshifts are detected in the k - band , at a rate between 65 - 90% , depending on the k - band depth . because we are interested in the most massive galaxies , which appear bright in the nir ,
the present study is not biased by these non - matches .
the `` secondary '' sample consists of those galaxies with photometric redshifts . within the secondary sample ,
there are two kinds of photometric redshifts .
the first are those systems which have optical magnitudes at @xmath49 , the limit for spectroscopy .
the other types of photometric redshifts are for those galaxies which have optical magnitudes fainter than this optical limit , in which we use a different method to compute photometric redshifts ( 3.2 ) .
this second set includes all galaxies at @xmath66 , which have purely photometrically measured redshifts .
we calculate photometric redshifts for our k - selected galaxies which do not have deep2 spectroscopy .
this sample is referred to as the photometric - redshift or secondary sample .
these photometric redshifts are based on the optical+near infrared imaging , in the brijk ( or brik for half the sample ) bands , and are fit in two ways , depending on the brightness of a galaxy in the optical . for galaxies that meet the spectroscopic criteria , @xmath49 , we utilise a neural network photometric redshift technique to take advantage of the vast number of secure redshifts with similar photometric data .
most of the @xmath49 sources not targeted for spectroscopy should be within our redshift range of interest , at @xmath32 .
the neural network fitting is done through the use of the annz ( collister & lahav 2004 ) method and code . to train the code , we use the @xmath53 redshifts in the egs , which has galaxies spanning our entire redshift range .
the training of the photometric redshift fitting was in fact only done using the egs field , whose galaxies are nearly completed selected based on the magnitude limit of @xmath49 .
we then use this training to calculate the photometric redshifts for galaxies with @xmath49 in all fields .
the overall agreement between our photometric redshifts and our annz spectroscopic redshifts is very good using this technique , with @xmath67 out to @xmath15 .
the agreement is even better for the m@xmath5 galaxies where we find @xmath68 across all of our four fields .
the photometry we use for our photometric redshift measurements are done with a 2 diameter aperture . for galaxies which are fainter than @xmath69 we utilise photometric redshifts using bayesian techniques , and the software from benitez ( 2000 ) .
for an object to have a photometric redshift we require that it be detected at the 3-@xmath3 level in all optical and near - infrared ( brijk ) bands , which in the r - band reaches @xmath47 ( coil et al 2004b ) .
we optimised our results , and corrected for systematics , through the comparison with spectroscopic redshifts , resulting in a redshift accuracy of @xmath70 for @xmath71 systems .
these @xmath71 galaxies are however only a very small part of our sample . up to @xmath15 only 6 ( 2.6% ) of our m@xmath5 galaxies are in this regime ,
while 412 @xmath9 m@xmath10 ( 9% ) of our galaxies have an @xmath45-band magnitude this faint .
all of these systems are furthermore at @xmath27 . at @xmath66
all of our sample galaxies are measured through the benitez ( 2000 ) method due to a lack of training redshifts .
there are instances where not all galaxies are detected in the optical bands , and the way we deal with these galaxies is discussed in 4.2.1 in terms of our measured number and mass densities .
one concern with utilising the deep2 spectroscopy as the prior in the annz method is that any galaxies which are actually at higher redshifts , but with @xmath49 , could contain a very incorrect redshift .
this is particularly important in this paper , as we do not wish to have additional galaxies added to our sample which could produce false artifacts in the number and mass density evolution , as well as incorrect identifications of massive galaxies .
we know that about 30% of galaxies with @xmath49 will be at @xmath66 ( steidel et al .
these systems at @xmath66 are however nearly all fainter than @xmath72 ( reddy et al .
2006 ) , and are thus not likely contaminating our massive galaxy sample , which is nearly all at @xmath73 .
after our photometric redshifts were calculated , the cfht legacy survey published their own independent photometric redshifts in the egs , which are on average within @xmath74 = 0.2 of our photometric redshifts .
the poorer agreement between the cfht legacy survey photometric redshifts and ours is likely due to the lack of nir data in the cfht fits , and the more restrictive priors used in the cfht photometric redshift fitting ( cf .
bundy et al .
2006 ) . in any case
, we do not use individual photometric redshifts , and only utilise them in large ensembles , where errors due to photometric redshift mismatches can be properly accounted for .
we match our k - band selected catalogs to the cfht optical data to obtain spectral energy distributions ( seds ) for all of our sources , resulting in measured brijk magnitudes . from these
we compute stellar masses based on the methods and results outlined in bundy , ellis , conselice ( 2005 ) and bundy et al .
stellar masses computed in this way have a long established history ( e.g. , sawicki & yee 1998 ) .
about 40% of our sample do not contain j - band imaging , but the stellar masses computed do not vary significantly when the j - band data is included or excluded .
all our stellar masses are furthermore normalised by the observed rest - frame k - band light , which is roughly at rest - frame @xmath75 for most galaxies .
the basic mass fitting method consists of fitting a grid of model seds constructed from bruzual & charlot ( 2003 ) ( bc03 ) stellar population synthesis models , with different star formation histories .
we use an exponentially declining model to characterise the star formation history , with various ages , metallicities and dust contents included .
these models are parameterised by an age , and an e - folding time for parameterising the star formation history , where sfr @xmath76 .
the values of @xmath77 are randomly selected from a range between 0.01 and 10 gyr , while the age of the onset of star formation ranges from 0 to 10 gyr .
the metallicity ranges from 0.0001 to 0.05 ( bc03 ) , and the dust content is parametrised by @xmath78 , the effective v - band optical depth for which we use values @xmath79 .
although we vary several parameters , the resulting stellar masses from our fits do not depend strongly on the various selection criteria used to characterise the age and the metallicity of the stellar population .
it is however important to realise that these parameterisations are fairly simple , and it remains possible that stellar mass from older stars is missed under brighter , younger , populations . while the majority of our systems are passively evolving older stellar populations , it is possible that up to a factor of two in stellar mass is missed in any star bursting blue systems .
however , stellar masses measured through our technique are roughly the expected factor of 5 - 10 smaller than dynamical masses at @xmath24 using a sample of disk galaxies ( conselice et al .
2005b ) , demonstrating their inherent reliability .
we match magnitudes derived from these model star formation histories to the actual data to obtain a measurement of stellar mass using a bayesian approach .
we calculate the likely stellar mass , age , and absolute magnitudes for each galaxy at all star formation histories , and determine stellar masses based on this distribution .
distributions with larger ranges of stellar masses have larger resulting uncertainties .
it turns out that while parameters such as the age , e - folding time , metallicity , etc .
are not likely accurately fit through these calculations due to various degeneracies , the stellar mass is robust .
typical errors for our stellar masses are 0.2 dex from the width of the probability distributions .
there are also uncertainties from the choice of the imf .
our stellar masses utilise the chabrier ( 2003 ) imf , which can be converted to salpeter imf stellar masses by adding 0.25 dex
. there are additional random uncertainties due to photometric errors .
the resulting stellar masses thus have a total random error of 0.2 - 0.3 dex , roughly a factor of two .
the details behind these mass measurements and their uncertainties is also described in papers such as brinchmann & ellis ( 2000 ) , papovich et al .
( 2006 ) and bundy et al .
( 2006 ) .
there is furthermore the issue of whether or not our stellar masses are overestimated based on using the bruzual & charlot ( 2003 ) models .
it has recently been argued by maraston ( 2005 ) and bruzual ( 2007 ) that a refined treatment of thermal - pulsating agb stars in the bc03 models results in calculated stellar masses that can be too high by a factor of a few . while we consider an uncertainty of a factor of two in our stellar masses , it is worth investigating whether or not our sample is in the regime where the effects of a different treatement of tp - agb stars in e.g. , maraston ( 2007 ) will influence our mass measurements .
this has been investigated recently in maraston ( 2005 ) and bruzual ( 2007 ) who have both concluded that galaxy stellar masses computed with an improved treatment of tp - agb stars are roughly 50 - 60% lower .
this problem has also been recently investigated independently by kannappan & gawiser ( 2007 ) who come to similar conclusions , but do not advocate one model over another .
furthermore , the effect of tp - agb stars is less important at our rest - frame wavelengths probed than at longer wavelengths , especially in the rest - frame ir .
our survey is k - selected , and the observed k - band is used as the flux in which the masses are computed .
the rest - frame wavelength probed with the observed k - band ranges from 0.7@xmath56 m to 1.5@xmath56 m where the effects of tp - agb stars are minimised .
the ages of our galaxies are also older than the ages where tp - agb stars have their most effect ( maraston 2005 ; bruzual 2007 ) . to test this , after our analysis was finished , we utilised the newer bruzual and charlot ( 2007 , in prep ) models , which include a new tp - agb star prescription , on our massive galaxy sample . from this
we find on average a @xmath13 0.07 dex smaller stellar mass using the newer models . at most
, the influence of tp - agb stars will decrease our stellar masses by 20% .
the effect of this would decrease the number of galaxies within our sample , particularly those close to the m@xmath80 boundary .
this systematic error is however much smaller than both the stellar mass error we assume ( 0.3 dex ) , and the cosmic variance uncertainties , and thus we conclude it is not a significant factor .
our k - band selection allows us to measure the stellar masses of galaxies to @xmath81 .
the stellar mass function for galaxies within these limits is discussed in bundy et al .
( 2006 ) for our sample . in this paper
we discuss only those galaxies with m@xmath5 .
the k - band magnitude distribution with redshift and the stellar mass distribution with redshift are shown in figure 3 .
figure 3 displays our total sample - consisting of the primary spectroscopic , and secondary photometric redshift selected galaxies combined . while the selection limit of m@xmath80 is somewhat arbitrary , it is however the mass limit in which we are reasonably completed out to @xmath82 . figure 3a shows the k - band magnitude distribution for our sample , demonstrating that all of our sample is at @xmath83 , where we are 100% complete in k - band detections ( 2.1 ; conselice et al .
however , we are not complete in other bands at all redshifts for a @xmath83 , and thus a m@xmath5 selected sample . at @xmath84
we are complete at @xmath9 m@xmath10 up to @xmath24 , up to @xmath15 for @xmath9 m@xmath10 , and up to @xmath82 for m@xmath5 .
however , this completeness is not the same in the b - band .
this needs to be accounted for as we do not include photometric redshifts measurements for galaxies which are undetected in one of our bands ( 3.2 ) .
however , for the redshift and mass ranges plotted in figure 3 & 4 and listed in table 2 , we are 100% complete at @xmath85 and are 2% - 10% incomplete at higher redshifts .
the method we use to calculate , and correct for , this incompleteness in the number and mass densities is discussed in 4.2.1 . to 120 mm [ sample - figure ] utilising our stellar mass catalogs , the selection of massive galaxies out to @xmath82 is straightforward , and is simply a cut at stellar masses m@xmath5 and m@xmath5 .
we do not consider lower mass galaxies in this analysis , due to significant incompleteness , although these galaxies may show the most evolution within these redshifts , as can be seen in the star formation downsizing ( e.g. , bundy et al .
2006 ) . in this paper , galaxies with masses m@xmath86 @xmath9
are only discussed in terms of their relationship to higher mass galaxies .
note that our massive galaxies are much brighter than the survey 5 @xmath3 limit of @xmath87 . as figure 3 shows , we are unlikely missing galaxies within our mass selection which are fainter than this limit up to @xmath82 ( 4.2.1 ) .
for example , at @xmath88 galaxies with a maximum m / l ratio with m@xmath80 would still be several times brighter than our k - limit .
@ccccc mass & redshift & log ( @xmath89 ) ( h@xmath90 mpc@xmath91 dex@xmath34 ) & log ( @xmath92 ) ( h@xmath90 mpc@xmath91 ) & log @xmath93 ( h@xmath94 yr@xmath34 mpc@xmath91 ) + log m@xmath95 & 0.5 & -4.62@xmath96 & 6.83@xmath97 & -4.9@xmath98 + & 0.7 & -4.29@xmath99 & 7.17@xmath100 & -3.6@xmath98 + & 0.9 & -4.08@xmath101 & 7.34@xmath102 & -3.0@xmath98 + & 1.1 & -4.28@xmath103 & 7.10@xmath104 & -2.9@xmath98 + & 1.3 & -5.05@xmath105 & 6.73@xmath106 & -3.3@xmath98 + & 1.5 & -4.91@xmath107 & 6.83@xmath108 & ... + & 1.7 & -5.03@xmath109 & 6.66@xmath110 & ... + & 1.9 & -5.33@xmath111 & 6.36@xmath112 & ... + + @xmath113 log m@xmath114 & 0.5 & -3.32@xmath115 & 7.76@xmath116 & -2.6@xmath98 + & 0.7 & -3.21@xmath117 & 7.79@xmath115 & -2.2@xmath98 + & 0.9 & -3.08@xmath118 & 7.92@xmath117 & -2.0@xmath98 + & 1.1 & -3.26@xmath117 & 7.83@xmath117 & -1.9@xmath98 + & 1.3 & -3.66@xmath117 & 7.44@xmath119 & -2.1@xmath98 + & 1.5 & -3.80@xmath120 & 7.42@xmath115 & ... + & 1.7 & -4.10@xmath116 & 7.11@xmath116 & ... + + @xmath121 log m@xmath122 & 0.5 & -2.80@xmath123 & 7.88@xmath123 & ... + & 0.7 & -2.80@xmath123 & 7.88@xmath123 & ... + & 0.9 & -2.69@xmath124 & 7.79@xmath123 & ... + in this section we examine in detail the number density and mass density evolution for massive galaxies found between @xmath26 and @xmath82 . for the purposes of this paper we consider galaxies with m@xmath0 as massive systems .
these galaxies are in fact the most massive galaxies in the nearby universe , and are generally not found in smaller area nir surveys .
for example , in the hubble deep field - north ( hdf - n ) , there are only four galaxies with stellar masses m@xmath5 , and none with m@xmath5 ( dickinson et al .
2003 ) using a chabrier imf . in comparison , we have 4571 galaxies with @xmath125 , and 225 galaxies with @xmath125up to @xmath15 .
although the hdf - n might be slightly depleted of massive galaxies , it shows the importance of using a large area survey to determine the properties and evolution of massive galaxies .
roughly 30% of our m@xmath86 @xmath126galaxies have measured spectroscopic redshifts , while 36% of the m@xmath86 @xmath126galaxies have spectroscopic redshifts .
we show the redshift distribution of our sample of galaxies as a function of redshift up to @xmath82 as a function of mass and k - band magnitude in figure 3 , with the massive galaxies with masses @xmath9 m@xmath10 , and m@xmath5 labelled . to 120 mm [ sample - figure ]
the fact that such massive galaxies exist at high redshift is not completely surprising , as a large number of massive galaxies have been found at @xmath88 , albeit in much smaller fields ( glazebrook et al .
2004 ; saracco et al . 2005 ) .
cosmic variance is an issue in these previous studies , which have areas over a factor of ten smaller than ours .
this is especially a problem , even within our survey , for the most massive galaxies , which are the most clustered within the redshift ranges we examine ( e.g. , coil et al . 2004a ; foucaud et al . 2007 ) .
as a result , these massive m@xmath5 galaxies are nearly completely absent in previous high redshift stellar mass studies , although a few examples exist in some previous work using smaller area fields .
because any variations in number or mass densities suggest evolution , it is important to consider the various errors that can mimic real evolution in our analysis both in terms of evolution in number / mass densities , as well as for the selection of the most massive galaxy sample which we examine later in the paper . beyond random and shot noise errors ,
we also consider stellar mass measurement errors , both systematic and random , as well as errors from cosmic variance when examining the evolution of galaxy number and mass densities .
this includes considering in the stellar mass errors issues with photometry , both random and systematic .
we also consider in our stellar mass density and number density measurements the errors produced through using photometric redshifts .
we do this through monte - carlo simulations of how the stellar mass function would change given the known uncertainty on these measurements .
this also gives us some idea of the eddington bias which is affecting our selection near the m@xmath127 and m@xmath127 boundary selection limits .
to 120 mm [ sample - figure ] the random errors result from poisson statistics on the number of galaxies ( @xmath128 ) in each redshift and mass bin .
we thus include a component in our error budget to account for these counting uncertainties .
stellar mass uncertainties are considered in two ways .
the first method uses monte - carlo simulations of mass errors to calculate how a typical 0.3 dex error can affect the measurement of number densities .
this is the typical error due to mismeasuring the stellar masses for our sample ( e.g. , bundy et al .
this factor of two comes from several sources , which we conclude are all contributing random errors to our mass budget .
below we examine the reasoning for this , and why we do not include any systematic stellar mass uncertainties .
it is first important to note that our photometry originates from sextractor , and does not use the _ imcat _ software utilised by deep2 to obtain photometry for target selection and luminosity functions ( e.g , faber et al .
we investigated systematic issues in the photometry through our simulations of placing fake galaxies of known magnitude into our images , and then determining how the detected magnitude differs from the input .
we found that when the completeness was high , the retrieved magnitude was almost nearly identical to the input . the completeness at @xmath129 , the faintest magnitude where our galaxies are found , is @xmath130% ( 2.1 ) , and we are thus unlikely affected by a systematic photometry error .
we do not include any additional systematic uncertainty into our error budget due to our detection methods .
we also compare our total sextractor magauto magnitudes to magnitudes measured using 4 dimeter aperture photometry , and find a very slight change with redshift , such that we are not detecting all the light in the highest redshift galaxies using total magnitudes .
this however is a very slight effect , with at most a difference of 0.03 mag which corresponds to 0.01 dex in stellar mass .
again , however , this effect is less of an issue for the brightest galaxies , which are the dominate population in our sample .
we conclude that the factor of two , or 0.3 dex random uncertainty in the stellar mass is a representative estimate of the uncertainties in stellar masses . to determine how a 0.3 dex random error in stellar masses can change our densities , we simulate how the mass function and number densities of our sample would change after applying this error .
we do this by changing the measured stellar masses by a random amount , to within 0.3 dex , and then recalculate the number and mass densities .
the difference in stellar mass and number densities between the recomputed and original values is then added to the error budget .
another major source of error in our number densities originates from cosmic variance .
based on the likely bias ( b ) and dark matter variance ( @xmath131 ) , it is possible to calculate the likely cosmic variance ( somerville et al .
2004 ) , and to include this in our error budget .
for massive galaxies with m@xmath5 , and using the volume of our survey within each redshift bin , we calculate that the galaxy variance is @xmath132 .
this leads to a fractional variation of 20 - 30% for the most massive galaxies .
this variance drops by a factor of three for the lower mass galaxies in our sample , rendering cosmic variance less of an issue for these systems .
when we consider our four fields , which are widely separated , it reduces the estimates of our cosmic variance by a factor of two .
it turns out that our computed number and mass densities ( table 2 ) are often dominated by these cosmic variance errors .
another source of uncertainty is the fact that if a galaxy does not have a measured photometric redshift , it can not be included in our sample .
as discussed in 3.2 , we do not measure photometric redshifts for galaxies not detected in one of our optical bands . to investigate whether we could be missing significant numbers of massive galaxies , particularly at high redshift , where we find evolution
, we examine the number of galaxies which are not detected in one band , but which could be high redshift massive galaxies .
first , as figure 3 shows , the k - limit for our massive galaxy sample is roughly @xmath133 for m@xmath5 galaxies at redshifts @xmath32 .
there are only 16 objects in our k - selected sample that are within this k - band magnitude limit , but are not detected in the b - band , and thus do not have a measured photometric redshift .
most of these galaxies are near the k - limit with @xmath134 , and none are as bright as the m@xmath5 galaxies
. these galaxies with missing optical fluxes are unlikely influencing our measured number and mass densities at @xmath85 for the following reasons .
the total number of galaxies with m@xmath5 at @xmath15 is 45 .
if all 16 galaxies at our k - limit without a photometric redshift were placed into this bin , it would only increase densities by 30% , not enough to account for the non - evolution we see .
furthermore , a m@xmath5 galaxy with a maximum m / l ratio would still be brighter than @xmath135 at @xmath15 , and thus these 16 galaxies are not likely to be at @xmath32 with a mass m@xmath5 .
the @xmath136 @xmath9 m@xmath137 bin at @xmath15 contains 725 galaxies , and these 16 galaxies would contribute a neglectable amount to their number and mass densities at @xmath15 .
we therefore conclude that our criteria that galaxies be detected in all optical bands to be included in our sample is a very insignificant source of uncertainty in our analysis at @xmath32 .
we are however less complete for selecting galaxies within our mass ranges at @xmath138 .
we correct for this by adding the number of galaxies at these higher redshifts that are likely missed .
we following willmer et al .
( 2005 ) for this process , where we consider the likely sampling based on the k and i magnitudes , and the ( i - k ) colour for systems which did not have a measured photometric redshift , and are missing a flux in one of the two other optical bands ( generally the b - band ) .
we add into our number and mass densities those objects which have a optical - nir colour and magnitude in i and k similar to the distribution of real galaxies in the given stellar mass and redshift bin .
this will in fact give an upper limit on the density corrections , however in reality it makes little difference to the values .
one of the most basic methods for understanding the evolution of massive galaxies is determining how their number densities and integrated mass densities change as a function of redshift . in bundy
et al . ( 2006 ) we examined the overall change in the number densities of massive galaxies , finding little apparent evolution when comparing mass functions .
we reexamine this in more detail using our spectroscopic plus photometric redshift sample .
we furthermore examine this evolution in smaller redshift bins , and explicitly state how the evolution of the mass function is occurring , or not occurring , for the most massive galaxies , and with what certainly we can make such statements
. as we are counting galaxies , as opposed to using the v@xmath139 method to calculating mass functions , we differ in our approach from bundy et al .
it is important to therefore test that our methods give consistent results .
we find this is the case using the published schechter functions in bundy et al .
( 2006 ) , derived from v@xmath139 computed mass functions , and our method of empirically determining whether there is evolution by counting actual galaxies .
we find an agreement between the two methods using the bundy et al .
( 2006 ) redshift bins ( 0.4 - 0.7 , 0.7 - 1.0 , and 1.0 - 1.4 ) . in the m@xmath5 mass range
we find slightly more galaxies using our counting method .
this is due to the schechter function under - fitting the high mass end in bundy et al .
( 2006 ) , something that also occurs for the @xmath140 mass function fits ( see below ) .
figure 4 shows the number and mass density evolution of the combined primary spectroscopic , and secondary photometric - redshift selected , galaxies for systems with masses m@xmath22 and @xmath136 @xmath9 m@xmath137 out to @xmath82 .
also included is the evolution in number densities for galaxies with masses @xmath141
@xmath9 m@xmath142 . we show the number densities of galaxies within these mass ranges measured in the nearby universe out to @xmath143 by the 2mass/2df galaxy surveys ( cole et al .
2001 ) normalised using the same chabrier imf as we use for our higher redshift comparisons .
it is clear by looking at figure 4 , and analysing the number densities listed in table 2 , that there is very little evolution , statistically , at @xmath144 for the m@xmath5 systems .
we can conclude that all massive m@xmath5 galaxies are present by @xmath24 to within a factor of 2 - 3 ( see e.g. , cimatti et al .
2006 ; bundy et al . 2006 ; brown et al .
however , as can be seen in figure 4 there is some evolution for these massive galaxies at @xmath27 , with evolution occurring up to @xmath145 as well as up to @xmath82 .
this is the first time that a study has had a large enough area to make this claim , and as we argue below , at least a fraction of the increase in total stellar mass seen between @xmath24 and @xmath82 ( e.g. , dickinson et al .
2003 ) is produced in massive galaxies .
as can be seen by eye , within our observational errors , there is some evolution in number densities for m@xmath5 and perhaps m@xmath5 selected galaxies between @xmath146 , although it is not clear with a casual examination whether this is significant .
the evolution in the number and mass densities can be examined quantitatively in a number of ways .
first , when we consider evolution just within our sample from @xmath147 to @xmath148 we find significant increases at masses m@xmath10 , both in terms of number and mass densities .
we describe below a quantitative analysis of these changes .
galaxies with stellar masses m@xmath5 show an increase in number densities between @xmath147 to 0.4 of a factor of 2.7@xmath149 .
this is however significant only at the @xmath150 level , considering all uncertainties .
in fact , all of this apparent evolution occurs at @xmath27 .
furthermore , we find a factor of 1.3@xmath151 increase in the mass density associated with m@xmath5 galaxies at the same redshift range , although this is also at less than 3 @xmath3 significance .
when we investigate evolution from @xmath82 to @xmath24 we find an increase of 11.2@xmath152 in number densities , and a factor of 5.5@xmath153 increase in mass densities for systems with m@xmath5 .
this is also an insignificant increase , and it appears that for the most part galaxies with m@xmath5 are nearly all formed by @xmath82 . we can not however rule out a factor of 2 - 3 evolution in number densities for these galaxies , given our large uncertainties .
we calculate the evolution for m@xmath5 galaxies down to @xmath26 by comparing with nearby galaxy studies , such as cole et al .
it is important to note however that these local studies were done using different techniques than ours , and there is potentially significant systematic differences used to measure the total amount of light in these nearby galaxies , and in the way the stellar mass was computed .
the largest difference is that the 2mass data used to calculate the stellar masses in cole et al .
( 2001 ) may under - represent the total amount of light , given its shallow depth , by up to 0.5 mags . as we are potentially also missing light in our massive galaxies ( 2.1 ) , a direct comparison between our data and cole et al .
( 2001 ) is fair , although a systematic difference of a few 10%-s of percentage can not be ruled out . through a direct comparison with cole et al .
( 2001 ) there is a factor of 2.2@xmath154 increase in number density , and a factor of 0.74@xmath155 increase in integrated stellar mass between @xmath145 and @xmath26 for m@xmath5 systems .
these @xmath26 number and mass densities are taken from the schechter fits and data presented in cole et al .
note that the cole et al .
( 2001 ) schechter function under - fits the most massive systems .
we correct this by explicitly using the number densities for the most massive galaxies tabulated in cole et al .
( 2001 ) .
galaxies with stellar masses @xmath136 @xmath9 m@xmath137 show significant evolution in number density .
the number densities of systems with @xmath136 @xmath9 m@xmath21 increases by a factor of 2.2@xmath156 between @xmath157 and @xmath158 , a result significant at @xmath126 4 @xmath3 . just as for the most massive systems ,
this evolution occurs completely at @xmath27 .
similarly , we find a factor of 2.1@xmath159 increase in the integrated mass density for systems with @xmath136 @xmath9 m@xmath137 within the same redshift range , also at @xmath160 @xmath3 confidence .
we find furthermore , after correcting for incompleteness due to the r - band limit ( 4.2.1 ) , that there is a factor of 14.5@xmath161 evolution in the number densities , and a factor of 10.7@xmath162 in mass densities between @xmath82 and @xmath24 for galaxies with @xmath136 @xmath9 m@xmath137 . both of these are significant at the 5 @xmath3 level . when we consider evolution from @xmath145 to @xmath163 , using the comparison to cole et al .
( 2001 ) , we find a factor of 3.0@xmath164 increase in number densities , and an increase of 1.7@xmath165 in mass densities .
both of these increases are significant at the @xmath166 @xmath3 level .
these results show that there is some evolution in the number and mass densities of massive galaxies at @xmath167 .
this evolution is such that the most massive systems with @xmath9 m@xmath10 increase in number and mass densities by factors @xmath168 at a significance @xmath59 @xmath3 . taken as a whole , we calculate that the scenario whereby the stellar mass and number densities of galaxies does not evolve between @xmath145 to @xmath16 can be rejected at @xmath169 @xmath3 confidence
. therefore it does not appear that high mass galaxy formation , with the exception of m@xmath5 systems , is complete by @xmath15 , yet it is largely completed by @xmath24 . therefore , the redshift range @xmath170 is the final epoch for the build up of the majority of the stellar mass in the most massive galaxies . to 120 mm [ sample - figure ]
as can be seen in figure 4 , galaxies with m@xmath5 have perhaps a surprisingly higher galaxy number and mass density at @xmath24 than at @xmath16 .
this is unlikely a volume effect , as the total co - moving volume of our sample at @xmath171 is 2.4@xmath172 mpc@xmath173 , while at @xmath26 the 2df/2mass sample occupies 5 @xmath172 mpc@xmath173 ( cole et al .
the other redshift bins have similar co - moving volumes as the @xmath171 bin .
furthermore , we can see this over - density , to a much lesser degree , in the number and mass densities using other mass cuts .
due to the photometric selection there are more galaxies at @xmath52 with spectroscopic redshifts than at lower redshifts .
this should not however create such a difference , as we are filling in the missing galaxies using photometric redshifts . to test whether this effect is due to the spectroscopic redshifts picking up at @xmath52 , we redo figure 4 using just photometric redshifts .
when we do so , we obtain nearly the same result ( figure 4 ) .
we also see this peak to a limited degree when only examining objects with spectroscopic redshifts .
another issue is whether the photometric redshifts , most of which were trained with a neural network method ( annz ) , are somehow biased towards a @xmath171 redshift .
we tested this in a number of ways .
first , the spike at @xmath171 for the m@xmath5 galaxies appears to be dominated by fairly low number statistics , and the peak is most obvious in field 2 .
there is perhaps a slight excess in the egs , but there is no excess or peak seen in fields 3 or 4 when examining either the photometric or spectroscopic redshifts .
secondly , the annz method may bias the fitted redshifts to exist at only certain values .
this is likely not occurring in our sample for the following reasons .
we are training our method using only the egs field , which has a spectroscopic over - density at @xmath174 , while our over - density is found at @xmath171 .
when we examine the distribution of the number of m@xmath5 galaxies as a function of redshift in narrow @xmath175z = 0.005 bins , we do not find any trend or correlation between spectroscopic redshifts and photometric redshifts .
that is , the locations in redshift where photometric redshifts are found are not the same as the spectroscopic redshifts .
an increase in number density over the @xmath26 values for massive galaxies can be seen , often to an even greater level , in the mass functions plotted in bundy et al .
( 2006 ) , as well as in the independent munics ( drory et al . 2004 ) and combo-17 ( borch et al .
2006 ) fields .
there are a few possibilities , beyond cosmic variance , which can explain this result .
although we conclude that the following are not the causes of the excess , we investigate them to show they are unlikely playing a role .
the first is that the co - moving volumes we utilise to compute densities depends strongly on cosmological parameters .
the first is that if we use a cosmology without a cosmological constant , an open universe with @xmath176 , the density contrast becomes larger .
in fact , only values higher than @xmath177 = 0.7 helps alleviate the problem , but no reasonable values totally erase the excess .
another issue we examined is whether the measurements of stellar masses are incorrect due to using the bruzual & charlot models to compute our mass to light ratios .
models , such as maraston ( 2005 ) who include a newer treatement of thermally pulsing asymptotic giant branch ( tp - agb ) stars in their models , can lower computed mass to light ratios for certain populations ( see 3.3 ) .
as the tp - agb stars have their most effect at an age of 1 - 2 gyr , we investigate the fitted ages of our stellar populations to see if they are near these values .
the average fitted ages for galaxies at @xmath178 is roughly 4 gyr for systems with m@xmath5 , and thus it is not likely that the exclusion of tp - agb stars is influencing our mass measurements .
however , fitted ages using tp - agb stars might be lower , and in this case , it still remains possible that our masses are slightly overestimated .
however , this likely can not account for the entire increase , as the improved inclusion of the tp - agb stars would lover the stellar masses by only 20% ( 3.3 ) .
the real cause of the increase at @xmath171 is likely cosmic variance , and even surveys larger than ours will be needed to probe the very massive end of the mass function with higher certainty .
investigating the structures and morphologies of galaxies is becoming recognised as one of the most important methods for understanding galaxies ( e.g. , conselice 2003 ; trujillo et al .
2004 ; cassata et al .
2005 ; lotz et al .
since the morphologies and structures of galaxies have a direct relationship to their formation modes ( i.e. , disks , merging systems , etc ) , we study the morphological properties of our sample in some detail .
the overlap of our nir imaging and the acs imaging in the egs contains 506 galaxies with stellar masses m@xmath86 @xmath126 at @xmath32 . for nearly all of these systems ,
their magnitudes are bright enough such that effects due to redshifts will not affect our ability to classify these systems either by eye , or through quantitative methods ( e.g. , conselice et al .
2000c ; windhorst et al .
2002 ; papovich et al .
2003 , 2005 ; taylor - mager et al .
note that the following morphological and structural analyses are only within the acs coverage of the egs field .
we study the structures and morphologies of our sample within the egs using two different methods .
the first is a simple visual estimate of morphologies based on the appearance of our galaxies in the acs imaging .
while the acs imaging of the egs field includes both f606w and f814w imaging , we only use the f814w band imaging in our structural / morphological analyses .
the outline of our classification process is given in conselice et al .
( 2005a ) .
basically , we place each galaxy into one of nine categories : compact , elliptical , lenticular ( s0 ) , early - type disk , late - type disk , edge - on disk , irregular , merger / peculiar , and unknown / too - faint .
while classifying these galaxies by eye it became apparent that many of the early - types appear to have a slight distortion .
we therefore added a sub - class of distorted ellipticals to our classifications .
these classifications are very simple , and are only based on appearance .
no other information , such as colour or redshift , was used to determine these types .
a short explanation of these types is provided below , with the number we find in each class listed at the end of each description .
1 . ellipticals ( e ) : ellipticals are centrally concentrated galaxies with no evidence for lower surface brightness , outer structures .
( 263 systems , 68 of which are classified as disturbed es ) 2 .
lenticular ( s0 ) : a galaxy is classified as an s0 if it appears elliptical - like , but contained a disk - like outer structure with no evidence for spiral arms , or clumpy star forming knots , or other asymmetries .
( 23 systems ) 3 .
compact - a galaxy is classified as compact if its structure is resolved , but still appears compact without any substructure .
it is similar to the elliptical classification in that a system must be very smooth and symmetric .
a compact galaxy differs from an elliptical in that it contains no obvious features such as an extended light distribution or envelope .
( 65 systems ) 4 .
early - type disks : if a galaxy contains a central concentration with some evidence for lower surface brightness outer light in the form of spiral arms or a disk , it is classified as an early - type disk .
( 28 systems ) 5 .
late - type disks : late - type disks are galaxies that appear to have more outer low surface brightness disk light than inner concentrated light .
( 18 systems ) 6 . edge - on disks : disk systems seen edge - on and whose face - on morphology can not be determined , but is presumably an s0 or spiral .
( 12 systems ) 7 .
irregulars : irregulars are galaxies that appear to have no central light concentration , and a diffuse structure , sometimes with clumpy material present .
( 2 systems ) 8 .
peculiars : peculiars are systems that appear to be disturbed , or peculiar looking , including elongated / tailed sources .
these galaxies are possibly in some phase of a merger ( conselice et al .
2003a ) , and are more common at high redshifts .
( 92 systems ) 9 .
unknown / too - faint : if a galaxy is too faint for any reliable classification it was placed in this category . often these galaxies appear as smudges without any structure .
these could be disks or ellipticals , but their extreme faintness precludes a reliable classification .
( 3 systems ) we repeat our visual classifications for a fraction of our sample , finding an initial misclassification at the 5% level .
we also compared our eye - ball estimates to the cas quantitative values discussed in 4.3.4 , which allowed us to identify a significant fraction of the misclassified systems ( see appendix a ) , which were then corrected to their proper morphological type .
we only perform these classifications out to @xmath15 ( cf .
higher redshifts in conselice et al .
2007a ) , and only use the f814w band for the classifications .
this presents a potential k - correction issue , as the f814w band reveals bluer light at higher redshifts .
however , through tests with the f606w band at redshifts where f814w probes well into the rest - frame optial , as well as tests on classifications in the hdf and hubble ultra deep field , show that morphological typing within the f814w band gives very similar results , with at most a 5% difference between rest - frame b - band and the observed i - band at @xmath32 ( conselice et al .
2005a ) .
when we discuss classifications in terms of early / mid / late types , these refer to : early ( e / compact / s0 ) , mid ( sa - sb ) , late ( sc - irr ) . for the most part we find from our visual analysis that most of the @xmath32 , and m@xmath5 systems , in which we can see structure , have an early - type morphology .
there is however a small , but significant , morphological diversity among the m@xmath5 galaxies .
we find that over all redshifts , 69% of the m@xmath86 @xmath126 systems are early - types ( elliptical , s0 , compact ) , while 10% are disks , and 18% are peculiars .
this is perhaps a surprisingly high fraction of peculiars within our sample , and suggests that some of these systems are still undergoing some type of mass assembly , possibly through merging or star formation .
this however changes when we examine only the most massive systems with m@xmath5 .
these galaxies are @xmath179% early - type over all redshifts , with a roughly similar number of mergers and disks making up the remainder .
these results remain essentially the same to within 5% , if we consider the eddington bias bringing lower mass galaxies into our mass cuts due to observational uncertainty .
this confirms and expands with a larger sample , earlier work on massive galaxies which also concluded that more massive galaxies appear to develop an early - type morphology before lower mass systems ( brinchmann & ellis 2000 ; bundy et al . 2005 ) .
to 120 mm [ sample - figure ] in figure 5 we plot the visual estimates for how the morphological break - down for both the m@xmath86 @xmath126 and @xmath9 m@xmath10 galaxies evolves from @xmath15 to 0.1 .
the @xmath26 points are based on morphologies from the rc3 catalogue , as described in conselice ( 2006a ) .
the most remarkable aspect of this evolution is that the fraction of elliptical galaxies with masses greater than m@xmath80 is relatively constant out to @xmath15 , at about 70% .
there is perhaps a slight increase for the @xmath9 m@xmath10 systems at @xmath180 , as seen in previous work ( bundy et al .
remarkably , there seems to be a gradual trend for the remaining 30% to transition from mergers / peculiars into disks .
it is interesting that such massive disk galaxies exist at these redshifts , which must have evolved into the m@xmath5 bin through star formation , or through a morphological transformation from an early - type to a spiral ( 4.3.2 ) . in summary
, what we find is that the majority of massive galaxies up to @xmath15 have an early - type morphology ( see also bundy et al . 2005 ) .
this is consistent with the findings that most of these systems are already formed in terms of their stellar masses at similar epochs ( 4.2 ) .
however , it is clear that @xmath181% of galaxies with m@xmath5 are not early - types , which suggests that there is evolution in the massive galaxy population that can not be seen simply through the evolution in number and mass densities .
the disk galaxies we see suggest that there is some star formation occurring , and the peculiars suggest that there is merger activity .
we investigate both of these galaxy - types , and what they imply for the evolution of the most massive galaxies , in the next sections . as can be seen from figure 5
, there is a significant , @xmath182 , increase in the number , and relative fraction , of massive m@xmath183 disk galaxies at @xmath32 .
this result is robust to morphological k - corrections ( 4.3.1 ; conselice et al . 2005 ) , and shows a real increase in massive disk galaxies through time .
we investigate the possible reasons for this increase in this section .
as already noted , there is a gradual transition for the non - ellipticals with m@xmath5 to appear as peculiars at @xmath27 , and as disks at @xmath144 .
what is likely occurring is that the higher redshift peculiars are transforming into ellipticals after they dynamically relax .
this leaves the problem of why are there are significantly more disks at lower redshifts , and why the fraction of early - types does not change with redshift .
that is , if the highest redshift peculiars are becoming ellipticals , we would expect the early - type fraction to increase at lower redshift , but it remains roughly constant .
one possible scenario is that some of the already massive ellipticals acquire a disk through accretion of intergalactic gas , and the accretion of low - mass satellite galaxies ( e.g. , abadi et al .
although candidates for this disk formation are seen at slightly higher redshifts both morphologically and through kinematics ( e.g. , erb et al .
2003 ; conselice et al .
2004 ; labbe et al . 2004 ; elmegreen et al .
2005 ; forster - schreiber et al .
2006 ; genzel et al .
2006 ; bouche et al .
2007 ) , it is unlikely that these large disks without massive bulges are formed in this manner .
the rise of these massive disks are explainable by systems with disks masses m@xmath10 at @xmath27 being brought up into the m@xmath5 bin through star formation or minor mergers . as we have seen earlier , the number densities of galaxies with m@xmath5 increases slightly with time .
therefore these disk galaxies were likely lower mass ( m@xmath10 ) systems , and achieved m@xmath5 status by @xmath184 through star formation or minor merger accretion activity ( 5.1 ) .
for structural reasons this seems like the most likely scenario , and can explain the nearly constant elliptical fraction , despite the morphological transformation of peculiars , by simply increasing the total number of @xmath126 galaxies by adding new spirals .
this idea is backed up by the evolution in the number densities , and relative number fractions , of the different morphological types shown in figure 5 .
the total number densities of disk galaxies at @xmath144 is higher than the number densities of peculiar galaxies at @xmath27 .
this means that it is difficult for the peculiars , without significant new star formation , to produce all of the spirals seen at lower redshifts . at the same time
the peculiars are decreasing in number density at @xmath185 , the densities of the ellipticals are increasing .
this causal relationship is highly suggestive that these peculiars are transforming morphologically into early types by @xmath24 .
after @xmath24 there are very few peculiars in the mass range @xmath9 m@xmath10 , and no further growth in the number densities of early types .
another argument against a direct peculiar to spiral transition is the structures of these massive disks .
the nine most massive spirals / disks in our sample are shown in figure 6 . as can be seen , these systems are often dominated by disk light , and do not appear to have particularly large bulges . some systems also appear to have rings and bars . because these spirals have such large disks and small bulges , it is unlikely that the peculiar galaxies at @xmath186 are transforming into these systems .
if this were the case , we would expect the spirals to show more disturbances , or at least a large bulge formed through the merger .
massive disk galaxies in previous work are also found to be largely in place in terms of sizes , morphologies , and masses by @xmath24 ( ravindranath et al .
2004 ; jogee et al .
2004 ; conselice et al .
2005b ) , thus any evolution in disks must be gradual .
an interesting future study would be to study in detail these high mass disk galaxies at @xmath187 to decipher their origin in more detail . to 120 mm [ sample - figure ]
it is important to note that although early - types ( classified e / s0/compact ) dominate the galaxy population at both the m@xmath5 and m@xmath5 mass selection limits , these galaxies often contain evidence for morphological peculiarities .
usually these are in the form of outer low surface brightness features , or multiple cores .
these are most likely the result of recent past merger activity in these galaxies , either through major or minor mergers .
examples of these distributed ellipticals are shown in figure 7 .
a total of 68 out of 263 ( 26@xmath188% ) of the ellipticals with m@xmath5 show some internal substructure visible by eye ( figure 7 ) . as figure 8 shows
, the fraction of ellipticals with these distorted structures is relatively constant over time .
there is perhaps a slight increase in this fraction at lower redshifts , which if real can be accounted for by surface brightness dimming .
peculiar ellipticals have been seen in other ways previously , such as through colour gradients and colour structures in ellipticals ( e.g. , menanteau et al . 2005 ; stanford et al .
2004 ; teplitz et al .
2006 ) , resulting from star formation and which may be related to the features seen here .
these previous studies have generally found that the lower mass ellipticals contain these star formation signatures .
these morphological disturbances however do not appear more common in the lower mass ellipticals , and in fact , 36% of the m@xmath5 ellipticals show this signature - a higher fraction that for the @xmath9 m@xmath10 population .
if these features are formed from a merger of various types , then it would be an anti - downsizing signature , suggesting that many giant ellipticals are still forming at @xmath32 , perhaps through dry - merging .
we use the cas ( concentration , asymmetry , clumpiness ) parameters to probe the structures of our galaxies quantitatively .
the cas parameters are a non - parametric method for measuring the structures of galaxies on resolved ccd images ( e.g. , conselice et al .
2000a ; bershady et al .
2000 ; conselice et al .
2002 ; conselice 2003 ) .
our main purpose in using the cas system is to identify relaxed massive ellipticals , as well as any galaxies that are involved in recent merger activity .
the basic idea is that galaxies have light distributions that reveal their past and present formation modes ( conselice 2003 ) .
furthermore , well - known galaxy types in the nearby universe fall in well defined regions of the cas parameter space .
we can then automatically determine the structures of galaxies and classify them according to where they fall in this space ( conselice 2003 ) .
for example , the selection @xmath189 locates systems which are nearly all major galaxy mergers ( conselice et al .
2000b ; conselice 2003 ; hernandez - toledo et al . 2006
; conselice 2006b ) ; although ` dry ' mergers will not be detected easily through this method ( hernandez - toledo et al .
2006 ) .
we apply a revised cas system to our massive galaxy sample to determine their structural parameters .
there are two caveats to using the hst one orbit acs imaging of these galaxies .
the first is that there are morphological k - correction and surface brightness dimming effects which will change the measured parameters , such that the asymmetry and clumpiness indices will decrease ( conselice et al .
2000 ; conselice 2003 ) , and the concentration index will be less reliable ( conselice 2003 ) .
there is also the issue that systems at @xmath138 are viewed in their rest - frame ultraviolet using acs data , which means that there are complications when comparing their measured structures with the calibrated rest - frame optical indices for nearby galaxies .
figure 9 shows the concentration / asymmetry and asymmetry / clumpiness projections of the cas plane . as can be seen in the ca plane
, the early - type galaxies mostly fall into the @xmath26 calibrated early or mid - type region , and disks are found in the mid - type and late - type regions .
interestingly , the distorted ( as selected by - eye ) ellipticals are significantly more asymmetric than the normal ellipticals . to 120 mm [ sample - figure ] the cas approach for classifying these galaxies
is roughly similar to what the visual morphologies suggest , although there are important exceptions .
details of the differences between these two methods are further elaborated on in appendix a. for the most part , galaxy types , as determined by eye , are where they are expected to be found on this diagram .
one important exception is that many of the visually classified non - distorted early - types are not located in the corresponding @xmath26 part of the cas diagrams .
while these systems have concentration indices similar to nearby giant ellipticals ( conselice 2003 ) , they often have higher than expected asymmetry indices .
an examination of these galaxies , and the residuals from a rotation and subtraction reveals that many of these systems have internal structures , likely resulting from mergers of various forms , that can not be easily seen by eye .
furthermore , the sa plane ( figure 9 ) shows an interesting pattern , such that most of our galaxies appear to deviate from the @xmath26 relation , shown by the solid line .
this deviation is such that all galaxies are more asymmetric than their @xmath26 counterparts , although the slope of the relationship between asymmetry and clumpiness is similar to that found at @xmath26 .
this implies that galaxies of a given morphological type , particularly the early - types , have a higher degree of bulk asymmetry in their structures .
the concentration and clumpiness values , which trace to first order the stellar mass and star formation , remain similar to what we expect from samples in the nearby universe .
finally , it appears that the most asymmetric systems , and those that deviate the most from the a - s relationship are the peculiar / merging systems , or those ellipticals that show some evidence for structure due to recent tidal events .
this shows that our visual estimates of merging systems is borne out by this quantitative analysis .
one of the great benefits of using the cas system for finding mergers is that it allows us to quantify the merger fraction and merger rate , and the number of mergers occurring in a galaxy population ( conselice et al .
2003a ; conselice 2006 ) .
the first observation we can derive from our cas values is the evolution of the merger fraction .
we determine the merger fraction for the m@xmath5 galaxies using the criteria , outlined in conselice ( 2006 ) of , @xmath190 using these criteria , we determine the merger fraction for the m@xmath5 galaxies in our sample ( figure 8) . as can be seen , there is a slight increase with redshift in the merger fraction such that it increases as @xmath191 , similar to the evolution seen in lower mass galaxies ( conselice et al .
2003 ; bridge et al .
2007 ) up to @xmath24 . by using the number densities for our systems , and the merger time - scales for our cas method
, we can calculate the merger rate for our m@xmath5 galaxy sample .
the number densities for our systems are taken directly from table 2 in this paper , and the time - scales for merging are derived from equation 10 in conselice ( 2006 ) , based on n - body models analysed using the cas approach . from this , we derive a merger cas time - scale of @xmath192 gyrs for a galaxy with a mass of 3@xmath193 . note that this time - scale is not the total merger time , but the time - scale in which the cas system would identify an ongoing merger .
details for how this time - scale is computed are included in conselice ( 2006 ) . the merger rate for our systems can be calculated through the merger rate equation , to 120 mm to 0.4 of the merger fraction derived through the cas approach for galaxies with m@xmath5 .
we find that the merger fraction as derived with the cas parameters declines as @xmath191 .
also plotted is the fraction of early types which are distorted within the same mass range of m@xmath5 .
note that like all of the structural and morphological plots , these data are only taken from the egs.,width=340 ] [ sample - figure ] @xmath194 where n@xmath195 is the number densities of objects , and f@xmath195 is the merger fraction .
note that this is not the galaxy merger fraction , which is the fraction of galaxies merging , which is roughly double the merger fraction ( conselice 2006 ) .
we find that statistically the merger rate for these m@xmath5 galaxies is constant from @xmath196 , and is on average log @xmath197 .
we can furthermore calculate the total number of major mergers a galaxy with a stellar mass m@xmath5 undergoes from @xmath15 to @xmath16 using equation ( 11 ) in conselice ( 2006 ) .
we calculate that the average number of mergers a massive galaxy with m@xmath5 will undergo from @xmath15 to 0.4 is n@xmath198 .
thus , on average a massive galaxy will undergo about one major merger from @xmath15 to @xmath199 , roughly consistent with previous results ( conselice 2006 ; bell et al .
2006 ) .
to 150 mm [ sample - figure ] a major question concerning high - mass galaxies is whether or not these systems have ongoing star formation at high redshifts . while it is commonly thought that massive galaxies have finished their assembly and star formation by @xmath24 , our results suggest otherwise
this is not the first claim for this , as evidence has accumulated during the last few years that ellipticals , including massive ellipticals , tend to have some ongoing star formation at @xmath24 ( e.g. , stanford et al .
2004 ; teplitz et al .
2006 ) and even at @xmath26 ( e.g. , donas et al .
star formation , as measured through blue colours , found in morphologically selected early - types , tends to be found in lower mass systems ( bundy et al . 2005 ) .
we therefore investigate the fraction of massive galaxies undergoing star formation during this time , and the amount of stellar mass added due to this star formation . to 150 mm [ sample - figure ]
we begin our examination of the star forming properties of m@xmath0 galaxies by examining their location on the rest - frame b - band absolute magnitude ( m@xmath200 ) vs. rest - frame colour ( @xmath201 ) diagram ( figure 10 ) .
these quantities are inferred from the observed magnitudes through model fitting , utilising the redshift and a series of k - corrections ( willmer et al .
galaxies appear to separate into a red - sequence and a blue cloud in this parameter space ( e.g. , strateva et al .
2001 ; i m et al . 2002
; baldry et al . 2004 ; bell et al . 2004 ; faber et al . 2005 ) .
the evolution of galaxies on the colour - magnitude diagram has potential meaning for understanding how galaxy evolution occurs . in figure 10
we plot the @xmath201 vs. m@xmath200 diagram for galaxies in our fields , with the most massive galaxies with m@xmath22 and @xmath136 @xmath9 m@xmath21 labelled .
the dividing line between red - sequence and blue cloud galaxies is shown as the dotted line in figure 10 , taken from faber et al .
( 2005 ) ; @xmath202 we find at the highest redshift bin , @xmath203 , a significant number of massive galaxies that are not on the red - sequence . in general , the number of massive galaxies on the red - sequence increases at lower redshifts .
this is also true when we plot these galaxies in terms of their morphologies , where we find that many early - type galaxies are not on the red - sequence .
the quantification of the fractional evolution of galaxies on the red - sequence is displayed in figure 11 .
this shows that the most massive systems with m@xmath22 generally fall in the red - sequence region at all redshifts , but with a significant number of systems in the blue cloud region at @xmath12 . as figures 10 and 11 demonstrate , the fraction of galaxies on the red - sequence , with colours redder than the value given in eq .
( 3 ) , increases with time at all masses . at the highest redshifts we can probe star formation , at @xmath186 , about 60% of the m@xmath22 galaxies are on the red - sequence , yet this fraction increases to 100@xmath204% by @xmath16 .
galaxies with lower masses show a similar pattern , yet lower mass galaxies always have a lower fraction of galaxies on the red - sequence at all redshifts , up to @xmath15 .
this also leads to a very important conclusion regarding the red - sequence and high mass galaxies .
previous studies have examined the increase of the amount of stellar mass on the red - sequence , finding as much as a factor of two increase since @xmath24 ( bell et al . 2004 ; faber et al .
. however , this increase is due to galaxies appearing on the red - sequence , which were previously blue , and not due to in - situ growth on the red - sequence itself .
this can be clearly seen in figure 11 where the massive galaxies are gradually moving onto the red - sequence with time , which can also be seen in the decline in the number of blue massive galaxies found in the universe since @xmath24 .
this is not consistent with the idea that the red - sequence grows solely through the so - called ` dry mergers ' .
although merging may be present within the red - sequence , and within our massive galaxy sample , it does not appear to be the dominate method whereby the red - sequence grows . which are on the red - sequence ,
while the red short - dashed and blue long - dashed lines show the evolution for systems with @xmath205 m@xmath21 , and @xmath141 @xmath9 m@xmath142 .,width=317 ] we quantify the star formation in our sample of massive galaxies using [ oii ] line equivalent widths from deep2 spectroscopy , and through mips 24@xmath56 m fluxes from spitzer mips imaging . over half of our total sample ( spectroscopic and photometric redshift combined )
do not have measured redshifts or [ oii ] emission line measurements . as such ,
we are forced to utilise the 24@xmath56 m photometry to measure the star formation rate for the bulk of our systems .
as we only have mips imaging in the egs , we limit our analysis of star formation rates to galaxies in this field .
our star formation measures utilise the [ oii ] line , when available , added to the star formation measured from the 24@xmath56 m fluxes . when [ oii ] is not available , the star formation rate is measured just using the mips fluxes . to utilise the 24@xmath56 m fluxes , we first convert the 24@xmath56 m flux into a total ir luminosity l@xmath206 utilising templates from dale & helou ( 2002 ) , as parameterised by lefloch et al .
we then calculate the obscured star formation rate in these systems through the equation , @xmath207 which was derived in bell et al .
equation ( 4 ) uses a kroupa / chabrier initial mass function , the same as we have used to calculate our stellar masses . as is becoming clear , it is possible that both the [ oii ] and the mips fluxes could result from agn emission . in the following analysis
we have therefore removed galaxies which have been detected in x - rays , to avoid the most egregious cases of agn contamination from our sample . out of our sample of 1151 mips sources with m@xmath0 at @xmath19 , we find that 105 or @xmath208% are detected in x - rays .
the x - ray properties of our sample are discussed in 4.5 .
the star formation rate measured using the [ oii ] line is calculated through the equation , @xmath209 ) ( { \rm m_{*}\ , yr^{-1 } } ) = 10^{-11.6 - 0.4(m_{b}-m_{b \odot } ) } \times { \rm ew_{[oii]}},\ ] ] where we are forced to use the equivalent width of the line instead of the flux , as the deep2 spectroscopy is not flux calibrated .
although we do not include sources detected in x - rays in our star formation analysis , there is still some chance that the [ oii ] emission arises from agn ( yan et al .
this is mostly an issue for line emission from red galaxies , and usually these sources are liners .
we can not utilise line ratios to test this idea , as we do not have h@xmath210 or [ nii ] equivalent widths .
we can however examine the equivalent width ratios of the [ oiii ] and [ oii ] lines .
we find that this ratio , [ oiii ] @xmath211 / [ oii ] @xmath212 is nearly always in the region of star forming galaxies within our sample .
this is consistent with the idea that bluer galaxies , which dominate our sample at these higher redshifts , have line ratios consistent with star formation ( yan et al .
2006 ) .
perhaps surprisingly , about half of our massive galaxies are detected at 24@xmath56 m .
after matching the mips and [ oii ] star formation indicators , we find that @xmath1340% of the m@xmath5 systems at @xmath19 are detected to our 24@xmath56 m depth . a total of 37@xmath213% of the systems at m@xmath5 are detected at 24@xmath56 m , with an average star formation rate of 70 yr@xmath34 . for systems within the mass range @xmath9 m@xmath86 @xmath9 , we find a higher fraction of mips detected systems , with a fraction of 45@xmath2141% , at an average star formation rate of @xmath215 yr@xmath34 .
when we compare star formation rates derived from the [ oii ] line with those from the 24@xmath56 m flux , we find that the 24@xmath56 m derived star formation rates are always much higher .
furthermore , we find little correlation between the two indicators .
this is an indication that [ oii ] and ir star formation indicators are measuring two different aspects of the star formation in a galaxy , or that one , or both , of these methods is mis - measuring the star formation . a more detailed discussion of different star formation indicators using emission lines and mips fluxes is included in weiner et al .
( 2006 ) . for galaxies with stellar masses
@xmath9 m@xmath86 @xmath9 , we find that the fraction of galaxies undergoing star formation remains roughly similar at all redshifts .
circumstantially , this is consistent with the fact that the fraction of spirals+peculiars in this mass cut is roughly constant throughout this redshift range .
interestingly , the fraction of systems which are undergoing star formation is higher than the non - elliptical fraction , showing that some morphologically classified massive ellipticals must be undergoing star formation .
there is a slight decline with redshift in the fraction of m@xmath5 galaxies with a significant 24@xmath56 m detection .
the fraction declines from 33% at @xmath203 to 14% at @xmath216 , which is consistent with a drop in the morphological fraction of non - ellipticals .
this is however certainly a lower limit to the evolution , as the number of galaxies we can detect at 24@xmath56 m declines at higher redshifts .
the redshift distribution of star formation rate , as measured through the combined star formation seen in the infrared and [ oii ] line is shown in figure 12 , for systems with m@xmath5 .
as figure 12 shows , we are incomplete in measuring the star formation rate at the highest redshifts of our sample .
this is due to the mips imaging becoming incomplete at the highest redshifts .
when we derive the total star formation history , we correct for this by assuming that the relative star formation rate distribution of our sample , that is the relative number of galaxies at a given star formation rate , is the same at all redshifts .
this does not assume that the normalisation of the distribution is the same , simply that the shape of the star formation distribution function is similar at high and low redshift .
this correction is however small , and only accounts for a 30 - 40% increase in the total star formation rate density .
line equivalent widths , as a function of redshift for our sample .
the large open red triangles show the location of galaxies with m@xmath22 , while the solid dots show the location of galaxies with @xmath9 m@xmath86 @xmath9 .,width=317 ] we determine how the star formation rate has evolved within our sample by examining how the total integrated star formation density changes as a function of time and stellar mass ( figure 13 ) .
we find , similar to previous studies utilising ir star formation indicators ( e.g. , le floch et al . 2005 ) , a decline at lower redshift for our sample , as seen in the entire field galaxy population ( e.g. , lilly et al . 1995 ) out to @xmath24 .
the less massive galaxies have a higher average star formation rate per unit time , and have a softer more gradual decline at lower redshift , compared with the m@xmath5 galaxies .
when we fit these star formation histories up to their plateau ( i.e. , at @xmath24 ) as a power law @xmath217 we find differences between our two mass ranges . for systems with m@xmath5 , we fit @xmath218 , and for systems with @xmath9 m@xmath86 @xmath9 we fit @xmath219 .
the overall decline in the entire galaxy population s star formation history can be parameterised as @xmath220 ( hopkins 2004 ; le floch et al .
it appears that while the @xmath9 m@xmath86 @xmath9 galaxies have a similar decline as the overall field , the highest mass galaxies show a marginally faster decline .
we can use our measured star formation rate densities to determine how much stellar mass is added to our massive galaxy sample through the star formation process . as we are averaging over the entire sample , we do not need to worry about the particular time - scales of the star formation induced in these systems , although the star formation time - scales are likely long ( noeske et al .
2007a , b ) . in these calculations
we assume that the star formation rate density remains the same through a redshift bin .
thus , we can calculate the amount of new mass in stars created through the star formation process simply by integrating the star formation rate density throughout the redshift bin of interest .
this calculation is independent of whether the star formation is produced in multiple short bursts , or in stochastic star formation occurring through time , as both scenarios give the same result .
figure 14 shows the stellar mass evolution , and the amount of mass added to our sample due to star formation , as a function of redshift .
this figure is constructed with the mass change referenced to the stellar mass density at @xmath186 , in each bin .
the solid and short - dashed lines show the amount of stellar mass added due to star formation within the same bins , to galaxies with @xmath9 m@xmath86 @xmath9 and m@xmath5 , respectively , as a function of redshift . while the stellar mass increase with time due to star formation matches within 1 - 3 sigma the amount of mass at redshifts @xmath221 it is not clear that , when considered together , whether the star formation by itself can account for changes in the mass function with time .
we examine this statistically by determining the probability that the stellar mass increase due to star formation can account for the mass changes seen at all redshifts .
this test shows that the probability that star formation , within a bin , is alone responsible for the stellar mass increase is 0.13% , at 3 @xmath3 , for the @xmath9 m@xmath86 @xmath9 systems , and @xmath222% ( 3.5 @xmath3 ) for the m@xmath5 systems .
therefore , statistically star formation by itself , within a bin , can not account for the observed mass changes at the highest redshift bins .
we however do not consider here the effects of galaxy transfer between mass bins , although we do so in 5.1 .
note that although statistically the changes in the mass function are not significant at @xmath223 , there is a general trend such that the mass density increases , particularly at @xmath27 .
our goal here is to determine whether the stellar mass created from the star formation is consistent with the observed change .
the stellar mass added from star formation is therefore unlikely able to account for the changes in the stellar mass density seen as a function of stellar mass .
this result is however dominated by the large error bars on the mass density .
as the only other way to obtain stellar mass growth is through galaxies entering bins from lower masses and merging , we investigate these processes and derive statistically how much merging might be occurring from @xmath186 to @xmath16 in 5.1 . ) as a function of redshift up to @xmath15 .
plotted are the measured star formation rates for both the @xmath9 m@xmath86 @xmath9 , and the m@xmath5 samples .
, width=317 ] to determine the unobscured agn properties of our massive galaxy sample we match our galaxies with a preliminary catalog of x - ray fluxes from nandra et al .
( 2007 , in preparation ) .
the nandra et al .
( 2007 ) study is a major chandra program to cover an area of 0.6 deg@xmath2 in the egs with a depth per pointing of 200 ks in both the hard ( 3.1 , 3 kev ) and soft ( 12.4 , 1 kev ) bands .
a basic description of this x - ray data is given in nandra et al .
( 2007 ) , georgakakis et al .
( 2007 ) and davis et al .
( 2007 ) , with a full description to appear in nandra et al .
( 2007 ) .
we matched our stellar mass selected catalog to this x - ray catalog , with a valid match occurring for those within 1 - 3 .
there are 123 galaxies in our sample of m@xmath5 that have a matching x - ray detection down to a limit of 8.2 @xmath224 erg s@xmath34 in the hard band , and 1.1@xmath224 erg s@xmath34 in the soft band .
this accounts for a surprisingly high fraction ( 5% ) of the entire m@xmath5 sample .
to these limits , there are also eight galaxies with m@xmath5 with an x - ray detection . in
both ranges these x - ray sources are found in roughly 5% of the sample .
the properties of x - ray sources in deep extragalactic near infrared and optical surveys have been discussed in several previous papers ( e.g. , hornschemeier et al .
2003 ; grogin et al . 2005 ; lehmer et al . 2005 ; georgakakis et al . 2006 ; nandra et al .
these previous studies have generally found that x - ray sources are in evolved , concentrated galaxies up to @xmath225 ( grogin et al .
2005 ; lehmer et al .
this may imply that x - rays are tracing only one phase of the evolution of an active galaxy , yet for the purposes of this paper we examine these x - ray sources with the assumption that they probe a representative sample of galaxies with active galactic nuclei . and
@xmath9 m@xmath86 @xmath9 , respectively , referenced to their stellar mass densities at @xmath186 .
the solid blue line shows the evolution in the amount of stellar mass added from the @xmath186 bin as a function of redshift from galaxies with stellar masses of @xmath9 m@xmath86 @xmath9 .
the short - dashed line shows a similar increase in stellar mass due to star formation for the m@xmath5 systems .
the long - dashed red line shows the relative increase in the stellar mass for the m@xmath5 systems due to stellar mass brought up from the @xmath9 m@xmath86 @xmath9 bin due to star formation.,width=317 ] since we are interested in the spectral properties and the luminosities of the agn in our sample , we convert our measured x - ray fluxes from the nandra et al .
( 2007 ) catalog into x - ray luminosities utilising the formula : @xmath226 where d@xmath227 is the luminosity distance and @xmath228 is a quantification of the spectral shape . to obtain a flux
which is independent of intrinsic absorption we calculate our luminosities using @xmath229 , the intrinsic spectral shape of an agn , and always use the hard band flux ( 2 - 10 kev ) , as it is the least affected by absorption .
we also investigate the hardness ratios ( hr ) , @xmath230 of our x - ray sources , finding a range of values , with an average hardness ratio of @xmath2310.58 for the full sample at m@xmath5 .
however , we find that the most massive systems with m@xmath232 have an average hardness ratio of [email protected] , showing that , on average , the most massive galaxies have soft spectra , likely from unobscured agn
. this may be a further sign of downsizing , where the agn in the higher mass galaxies becomes less obscured due to the removal or dissipation of the obscuring mechanism earlier than in lower mass galaxies .
figure 15 shows our basic x - ray analysis results in a graphical format , where the stellar masses of our galaxies with an x - ray detection in the egs are plotted with their x - ray luminosities , as defined in equation ( 4 ) .
the morphological types and hardness ratios of these sources are also labelled .
the x - ray sources are also circled on the cas diagrams in figure 9 .
we generally find no correlation between x - ray source hardness ratio , x - ray luminosity or host galaxy stellar mass .
however , our range in stellar mass is limited , and this should not be interpreted as a complete study of agn , but simply as a study of the x - ray sources associated with massive galaxies .
there is one interesting trend , which is that the x - ray sources in massive galaxies tend to be found in peculiars and distorted early - types . for the overlapping sample of 29 galaxies with m@xmath183
that have both x - ray and hst imaging , we find that @xmath233% are either peculiar galaxies or distorted ellipticals .
the remainder are 34@xmath234% normal elliptical / s0/compact and 21@xmath235% disks .
this becomes even more significant when we determine what fraction of each morphological type is detected in the x - rays .
when we do this comparison , we find that @xmath1315% of the distorted ellipticals are x - ray detected , while only 3.5% of the e / s0/compact galaxies are detected
. this may appear to disagree with previous studies , such as grogin et al .
( 2005 ) who found that x - ray sources are largely in low asymmetry and highly concentrated galaxies .
we find that some of our x - ray sources have a large concentration index ( figure 9 ) , and some of the peculiar ellipticals have a modest asymmetry , but they are not in the merger region at @xmath189 .
it is likely that these earlier studies missed the fact that a significant fraction of the x - ray sources in the highest mass galaxies are found in early - type galaxies with peculiar structures .
since it is likely that these structures were produced in some type of a merger several gyr earlier , we are potentially witnessing the remnant agn activity produced in the last merger event .
these earlier studies of agn morphology also were not mass selected , as is our sample .
it is furthermore possible that we are not able to detect the ongoing peculiar galaxies merging , in x - rays , due to deep obscuration , although this idea needs to be tested in more detail in a future study .
are circled , while those with hard ratios @xmath236 are squared.,width=317 ]
the last few years have seen a number of studies examine similar issues as this paper , with many studies concluding that massive galaxies , typically those with m@xmath5 , are largely formed by @xmath24 .
this is statistically found to be the case in this paper , to within a factor of 2 - 3 , in terms of measured mass and number densities , for systems with m@xmath5 from @xmath24 , although there is some evolution from @xmath15 .
a scenario where the number and mass densities of m@xmath5 galaxies does not change at @xmath32 can be ruled out at @xmath169 @xmath3 confidence . while we , and others , have found that the number and mass densities for massive galaxies are similar at @xmath144 , this
does not necessarily imply that there is no evolution in this population .
as we have discussed , there can be as much as a factor of three in evolution which we could not identify due to the uncertainties in the process of measuring stellar masses , and computing their number densities .
it is likely that other methods besides densities , are need to conclusively argue whether massive galaxies are finished forming by @xmath24 . while we can rule out increases in densities that are a factor of three or greater
, we can not conclude that our that massive galaxies are not increasing in number and mass density by a factor of 2 or 3 .
the fact that we find a high fraction of massive galaxies that are forming stars ( 40% ) , and which are non - elliptical ( 30% ) , at @xmath24 alone suggests that there is some evolution .
one issue which is not clear is how much of the merger and star formation process , and especially the controversial and hard to find dry mergers , are responsible for the addition of mass in massive galaxies at these redshifts .
we can address this using just our mass functions , and the measured star forming histories for galaxies with m@xmath5 , and @xmath9 m@xmath237 . while the star formation history matches the observed increase in the stellar mass to within @xmath238 at any one redshift , the fact that the star formation history is consistently lower at all redshifts implies that star formation , within a bin , can not account for the total increase in stellar mass .
this shows that part of the mass growth in these systems must be accounted for by mergers , or as we examine below , galaxies with masses lower than the @xmath239 stellar mass limit evolving into the higher mass bin due to star formation and/or merging .
we find a higher probability that star formation can account for the increase in the mass density for systems with @xmath9 m@xmath10 , with a 0.13% chance ( @xmath13 3 @xmath3 ) .
it appears however that for all galaxies with m@xmath5 , additional mass is added to these redshift bins over that which is seen in the star formation . between the bins m@xmath5 and @xmath9
m@xmath237 we can determine the mass transfer due to star formation . we calculate this in the next section after making a few assumptions concerning the distribution of galaxies within these bins , and how the star formation seen within these bins is distributed . in this section
we examine ways in which galaxy mass functions can evolve due to star formation and merging .
we consider star formation within a given mass bin , star formation at a lower mass bin , and mergers within a lower mass bin .
note that mergers within a given bin will never increase the total mass within that bin , unless star formation is triggered with the merger .
it can however lower the number densities of galaxies within a bin .
consider the change in the total stellar mass within a mass range of @xmath240 at a mass of m@xmath241 .
we can express the evolution of the mass function within @xmath240 between redshifts @xmath242 as , @xmath243 where @xmath244 is the change in the total amount of mass density within the mass bin between @xmath245 and @xmath246 , and @xmath247 and @xmath248 .
@xmath249 is the amount of mass density within this bin added due to star formation within this mass bin , @xmath250 and @xmath251 are the amounts of stellar mass density added to the @xmath246 bin from mergers , and star formation which were initially in galaxies at @xmath252 .
likewise , these terms should include mass density which has left the bin due to mergers and star formation . if we consider the m@xmath5 bin , or any bin which is defined to be simply greater than a given mass , this additional term is zero .
these terms can be computed in the follow ways . for the star formation within the bin , where galaxies have an initial stellar mass between @xmath253 m@xmath10 @xmath246
, we can write the mass added from star formation as , @xmath254 where in equation ( 8) , @xmath255 is the star formation rate density and @xmath77 is the time interval between redshifts @xmath256 and @xmath257 .
this is the quantity plotted as the short - dashed and solid lines in figure 14 , and which falls short of accounting for the possible growth of stellar mass within these massive galaxies ( 4.4.3 ) . because the star formation within these bins can not account for the change in the number and mass densities for massive galaxies ( 4.4.3 ) , the increase in stellar mass
must be brought about from galaxies that were previously at lower masses .
this increase must be produced by mass entering the bin through star formation or merging .
since star formation is in principle easy to directly study and trace , we will examine this aspect first . to examine how much stellar mass is added to a bin of lower stellar mass , we must consider the star formation rate in galaxies in the lower mass bins .
if the width of a lower stellar mass bin is @xmath258 , then the amount of mass added per galaxy ( @xmath259 ) within a redshift and mass bin is given by @xmath260 where @xmath261 is the number density of galaxies in the lower mass bin .
if we assume that galaxies are distributed evenly within their mass bin , @xmath258 , then the fraction of galaxies within this bin which leave the bin due to star formation is simply @xmath262 the number density of galaxies that move into the higher mass bin due to star formation is then , @xmath263 since we assume that the original galaxies are distributed evenly throughout a bin , then the fraction of galaxies in the lower mass bin which move into the higher mass bin is similar to the fraction of mass that moves up . in this case
, the average mass of the galaxies brought up into the higher mass bin will be @xmath264 .
where @xmath265 is the upper mass limit of the lower mass bin from which these galaxies are taken from .
the amount of mass added to the higher mass bin due to star formation bringing up galaxies is therefore , @xmath266 where @xmath267 is the total amount of star formation occurring in galaxies in the lower mass bin . in this case ,
@xmath268 ( cf .
. 8) . to determine how star formation in lower mass bins can affect the mass and number densities in higher mass bins we take the example of galaxies in the 10@xmath269 bin moving into the m@xmath5 bin between @xmath186 and @xmath270 . in this case
@xmath268 = @xmath271 h@xmath94 mpc@xmath91 , and @xmath272 = 4.3@xmath273 .
since @xmath274 @xmath275 = 2.2 @xmath193 , then the fraction of galaxies in the lower mass bin at @xmath186 which are now in the upper mass bin at @xmath276 is @xmath277 .
the change in number density in the upper mass bin , in our example , m@xmath5 , is then 4.3@xmath278 mpc@xmath91 dex@xmath34 .
this is nearly entirely the difference between the number densities for galaxies at @xmath186 and @xmath276 for galaxies with masses m@xmath5 .
we calculate in the previous section that in the highest redshift bin there is no need for additional galaxies due to merging using the observed change in number densities .
however , a better test of the merger scenario is to examine how the total mass densities evolve .
the reason is that when mergers occur within a given mass bin the number density will decrease , but the mass density remains conserved , if no star formation is triggered . because star formation and perhaps merging is occurring within lower mass galaxies
, those systems will enter the higher mass bin and increase the number of galaxies , creating a static number number density , masking real evolution .
however , in this case the integrated mass density must increase within our observed bin .
we continue our examination of mass transfer between the m@xmath5 and @xmath9 m@xmath10 mass bins .
we calculate the amount of mass added to the m@xmath5 bin by subtracting the amount of mass added from star formation from the total mass density change . the amount of stellar mass added to the m@xmath5 bin from star formation in the @xmath9 m@xmath237 bin is given by eq .
( 11 ) . in this case , @xmath251 = 3.9@xmath172 h@xmath94 mpc@xmath91 . the total integrated stellar mass added to galaxies at m@xmath5 due to galaxies at @xmath9 m@xmath237 moving into the m@xmath5 limit from star formation
is shown in figure 14 as the long - dashed red line .
this additional mass accounts for roughly half of the mass increase in the m@xmath5 bin up to @xmath279 , with the remainder of the mass possibly originating from merging in the upper part of the @xmath9 m@xmath237 bin .
we however can not rule out that the scenario that mergers are not needed to account for the change in mass densities as the star formation from the lower mass bin is only lower than the change in mass densities by 1@xmath3 . by subtracting the star formation within the m@xmath5 bin , and
the mass added from star formation in the lower mass bin , we calculate @xmath250 = 3.3 @xmath172 h@xmath94 mpc@xmath91 using eq .
( 7 ) . to calculate the number of possible mergers , and the merger fraction within the @xmath9 m@xmath237 bin , we first must recognise which types of mergers , and of what masses ,
can enter the m@xmath5 bin .
first , to simplify things , we assume that galaxies within a given mass range and redshift range undergo at maximum one merger between @xmath256 and @xmath257 .
this is a very reasonable assumption within our redshift bins given that even the fastest mergers take around 0.5 gyr to complete .
we furthermore assume that all mergers within the lower mass bin are major mergers of equal mass . with these assumptions ,
it is clear that only galaxies with m@xmath5 @xmath265/2 will have a mass m@xmath5 @xmath265 after merging with a galaxy of similar mass .
therefore , only galaxies merging with masses between m@xmath280/2 , and m@xmath280 will enter the upper mass bin .
if we assume that galaxies are evenly distributed between these two masses , then the average mass of a merging galaxy is @xmath281 , or 3/4@xmath265 .
the total number density of galaxies merging , ( @xmath282 ) , is therefore equal to the stellar mass density of galaxies which have merged divided by the average mass of the merging galaxies , @xmath283 we can compute the galaxy merger fraction ( conselice 2006 ) , @xmath284 , as the total number of galaxies undergoing a merger divided by the total number of galaxies within a given bin .
this fraction can then be expressed as , @xmath285 plugging in numbers for our @xmath9 m@xmath10 and m@xmath5 mass bins , we find that 12% of the galaxies with @xmath9 m@xmath237 merge between @xmath286 , and thus enter the higher mass bin .
we find that at @xmath287 this merger fraction drops to 8% .
this is consistent with the cas results presented in 4.3.5 , and previous published results ( conselice et al .
2003 ; lin et al . 2004
; bundy et al .
2004 ; lotz et al . 2006 ; bridge et al .
this suggests that although number and mass densities do not statistically differ from each other , by examining the mass function in narrow redshift bins as well as the star formation rate and morphologies we conclude that there is additional mass added to these systems , by as much as a factor of two increase since @xmath288 , which is largely due to merging .
previously , we could not rule out no change in number densities , based only on counting galaxies .
the evolution is however clearly seen in the structures of these galaxies .
we also compare our stellar mass and number densities results to models from e.g. , de lucia et al .
( 2006 ) based on the millennium simulation ( lemson et al . 2006 ) as another approach towards understanding how well we can reproduce with simulations the evolution of these systems .
the millennium run is a @xmath289cdm simulation of the universe using 10@xmath290 particles and follows the dark matter and stellar assembly of galaxies in a 500@xmath291 mpc cube .
it is currently one of the most advanced simulations that are available for comparison to data , and have been used previously to trace the evolution of massive galaxies ( de lucia et al .
2006 ) .
while the @xmath55cdm model is very successful in explaining the large scale structure of the universe , there are inconsistencies when comparisons are done on the scale of galaxies . by testing this model in other ways , such as through the assembly of the most massive galaxies , we can perhaps determine the origin of some of the missing physics , or initial conditions used in these simulations .
the stellar mass calculation in the millennium simulation uses the same imf as our stellar mass measurements , and are based purely on the amount of stars produced in galaxies as a function of time .
the chabrier imf enters into these models through the computation of the fraction of gas in star formation which is returned to the cold gas , which in these models is 40% .
there is also a prescription for which stars explode as supernova , and when , which further reduces the stellar mass , but only for the stars at the highest masses .
when we compare our stellar mass number and mass densities to the millennium model , we find that they generally underestimate the number and mass densities of the most massive galaxies at @xmath20 ( figure 16 ) .
the agreement for both stellar mass number and mass densities at @xmath26 is however fairly good within the uncertainties .
the best agreement is between the number densities of systems with @xmath9 m@xmath86 @xmath9 .
however , as can be seen in figure 16 , this agreement breaks down when comparing the mass densities of these systems to the model . as there is generally a much worse agreement for the m@xmath5 systems in terms of mass densities than number densities
, we can interpret these disagreements in the sense that the models do not produce enough massive galaxies early enough .
this extends throughout the m@xmath5 range .
clearly , the models eventually produce these massive systems , as there is agreement at @xmath26 , but they are formed in the simulation at much later times , typically at @xmath144 through mergers ( de lucia et al .
2006 ) .
this disagreement suggests that the bulk formation of massive galaxy formation occurs at @xmath25 , especially for the massive systems .
this implies both an early star formation history , as well as an early mass assembly history .
when comparing with observations of the merger history , it is clear that massive galaxies are formed through mergers at @xmath25 ( conselice 2006 ) .
it is likely that these galaxies are much more biased than what the millennium simulation assumes , and thus they must have an early merger and star formation history . when examining the galaxy merger history ( conselice et al .
2003 ; conselice 2006 ) , it is clear that massive galaxies do not merge much at @xmath20 , but have a high merger fraction ( @xmath292% ) at higher redshifts , @xmath25 .
the millennium model produces most of the mass in these massive galaxies by @xmath82 , but they are not yet assembled in massive galaxies in the model until relatively late ( e.g. , de lucia et al .
it appears that the actual merger rate for massive galaxies is much higher in the past , than what is predicted , and likewise smaller at @xmath20 than the models . to 150 mm [ sample - figure ]
we utilise wide and deep near infrared imaging from the palomar telescope combined with deep2 spectroscopy to select and study the properties and evolution of m@xmath5 galaxies found at @xmath19 .
our total sample consists of 4571 galaxies with m@xmath5 , and 225 galaxies with m@xmath5 .
we investigate with deep2 spectroscopy , hubble space telescope ( hst ) imaging , chandra imaging , and spitzer mips imaging : the x - ray , morphological and star forming properties of our sample .
our major findings are : + 1 . the stellar mass and number densities of m@xmath5 galaxies does not change significantly at @xmath144 .
we however can not rule out factors of 2 - 3 in number and mass density evolution for these systems , based solely on densities , due to uncertainties in these measurements .
systems with @xmath9 m@xmath10 however show a significant increase between @xmath170 .
the increase in stellar mass density for systems with m@xmath5 is however marginally insignificant over our entire redshift range up to @xmath82 .
there is a diversity in the morphological properties of galaxies selected solely by stellar mass .
we find that most galaxies selected with m@xmath22 are classifiable as early - types at all redshifts , with only a small fraction classified as peculiars and spirals at @xmath186 .
systems with @xmath9 m@xmath10 have a lower fraction of galaxies consistent with being early - types , with a nearly constant fraction with redshift of 70% .
the remaining systems tend to be classified as peculiars / mergers at high redshift , and spirals at lower redshifts .
we further find that a significant number of the early - type galaxies contain a slight morphological disturbance , which is furthermore seen in the quantitative cas parameters .
we find that a significant fraction ( @xmath215% ) of massive galaxies at @xmath19 are undergoing star formation .
this is demonstrated through the blue colours of massive galaxies , and star formation rates as derived through spitzer mips imaging of these systems .
the fraction of galaxies with m@xmath22 which are on the red - sequence is only 60% at @xmath186 , but increases to @xmath13 100% by @xmath16 . likewise , the fraction of lower mass systems , with @xmath9 m@xmath10 , which are on the red - sequence increases with time , although at every redshift there is a higher fraction of systems with m@xmath22 on the red - sequence than at lower masses .
we investigate the star formation rate density for our massive galaxy sample , finding a steep decline in the star formation rate density for systems with m@xmath5 , and a more shallower decline for systems with @xmath9 m@xmath10 .
the rate of decline in star formation for m@xmath5 systems can be parameterised as @xmath293 , as opposed to @xmath294 for systems with masses @xmath9 m@xmath10 .
we match our catalog of massive galaxies to a catalog of x - ray sources discovered with chandra imaging in our largest field , the extended groth strip .
we find that a significant fraction ( 5% ) of the most massive galaxies are x - ray emitting agn .
we investigate in detail the properties of these agn and find that the most massive galaxies , with m@xmath5 , have soft hardness ratios , while the galaxies with @xmath9m@xmath10 have a mixture of hard and soft ratios .
furthermore , we find that nearly half of all the massive galaxies which have agn are either distorted ellipticals or peculiars .
we investigate how much stellar mass is added to galaxies due to star formation from @xmath15 to @xmath16 , and compare this to the observed changes in the stellar mass density .
we find to @xmath295 confidence that the star formation seen in individual massive galaxy bins at @xmath32 can not account for the changes in stellar mass seen in galaxies with m@xmath5 .
we however find that the amount of mass which transfers between bins due to star formation , bringing galaxies up into higher mass bins , can within 1 @xmath3 account for this increase .
the difference suggests that up to a single major merger is occurring for m@xmath5 galaxies at @xmath19 .
this is also found to be the case through an analysis of cas parameters , giving @xmath18 major mergers at @xmath19 .
our major conclusion in this paper is that the stellar mass assembly of massive galaxies is , within a factor of two , complete by @xmath24 .
there is however still significant morphological and colour evolution at @xmath144 for massive galaxies .
we also find significant evolution in the number and mass densities for galaxies with m@xmath5 at @xmath27 .
furthermore , we have shown that the study of ` early - types ' , defined through colour , morphology or mass , at high redshifts must be carefully done , and results of studies will vary significantly , depending on selection .
it is clear , particularly at high redshift , that red galaxies are not the equivalent of massive galaxies , or elliptical galaxies , and each of these populations must be studied individually .
the palomar and deep2 surveys would not have been completed without the active help of the staff at the palomar and keck observatories .
we thank niv drory , karl glazebrook , gabriella de lucia , gerard lemson & the virgo consortium for providing data in an electronic formation , and sandy faber and the anonymous referee for their comments on this work .
we thank stephen bevan for computing stellar masses using both the bruzual & charlot ( 2003 ) and ( 2007 ) models , and we thank g. bruzual and s charlot for providing us with their 2007 models before publication .
funding to support this effort came from a national science foundation astronomy & astrophysics fellowship , grants from the uk particle physics and astronomy research council ( pparc ) , support for the acs imaging of the egs in go program 10134 was provided by nasa through nasa grant hst - g0 - 10134.13-a from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
this work is based , in part , on observations made with the spitzer space telescope , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
support for this work was provided by nasa through contract 1255094 issued by jpl / caltech .
the authors wish to recognise and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
alc / jan is supported by nasa through hubble fellowship grant hf-01182.01-a / hf-011065.01-a .
support for this work was also provided by nasa to cjp through the spitzer space telescope fellowship program , through a contract issued by the jet propulsion laboratory , california institute of technology under a contract with nasa .
99 abadi , m.g . ,
navarro , j.f . , steinmetz , m. , eke , v.r .
2003 , apj , 591 , 499 baldry , i.k . ,
glazebrook , k , brinkmann , j. , ivezic , z. , lupton , r.h .
, nichol , r.c . ,
szalay , a.s .
2004 , apj , 600 , 681 bell , e.f .
, et al . 2004 , apj , 608 , 752 bell , e. , et al . 2005 , apj , 625 , 23 bell , e. , et al .
2006 , apj , 640 , 241 benitez , n. 2000 , apj , 536 , 571 bershady , m.a .
, jangren , j.a . , conselice , c.j .
2000 , aj , 119 , 2645 bertin , e. , & arnouts , s. 1996 , a&as , 117 , 393 borch , a. , et al . 2006 ,
a&a , 453 , 869 bouche , n. , et al .
2007 , eprint arxiv:0706.2656 bridge , c. , et al .
2007 , apj , 659 , 931 brinchmann , j. , ellis , r.s .
2000 , apj , 536 , 77l brown , m.j.i . ,
dey , a. , jannuzi , b.t . , brand , k. , benson , a.j . ,
brodwin , m. , croton , d.j .
, eisenhardt , p.r .
2007 , apj , 654 , 858 bruzual , g. , charlot , s. 2003 , mnras , 344 , 1000 bruzual , g. astro - ph/0703052 bundy , k. , fukugita , m. , ellis , r. , kodama , t. , conselice , c.j .
2004 , apj , 601 , 123l bundy , k. , et al .
2006 , apj , 651 , 120 bundy , k. , ellis , r.s . , conselice , c.j .
2005 , apj , 625 , 621 cassata , p. , et al .
2005 , mnras , 357 , 903 chabrier , g. 2003 , apj , 586 , 133l cimatti , a. , daddi , e. , renzini , a. 2006 , a&a , 453 , l29 coil , a.l . , et al . 2004a , apj , 609 , 525 coil , a.l . ,
et al . 2004b , apj , 617 , 765 collister , a.a . ,
lahav , o. 2004 , pasp , 116 , 345 cole , s. , et al .
2001 , mnras , 326 , 255 conselice , c.j .
1997 , pasp , 109 , 1251 conselice , c.j .
, bershady , m.a . ,
jangren , a. 2000a , apj , 529 , 886 conselice , c.j . ,
bershady , m.a . ,
gallagher , j.s .
2000b , a&a , 354 , 21l conselice , c.j . , gallagher , j.s . ,
calzetti , d. , homeier , n. , kinney , a. 2000c , aj , 119 , 79 conselice , c.j .
2003 , apjs , 147 , 1 conselice , c.j .
, bershady , m.a . ,
dickinson , m. , papovich , c. 2003a , aj , 126 , 1183 conselice , c.j . ,
chapman , s.c . ,
windhorst , r.a .
2003b , apj , 596 , 5l conselice , c.j . , gallagher , j.s . ,
wyse , r.f.g .
2002 , aj , 123 , 2246 conselice , c.j . ,
2004 , apj , 600 , 139l conselice , c.j . ,
blackburne , j. , papovich , c. 2005a , apj , 620 , 564 conselice , c.j . ,
bundy , k. , ellis , r. , brichmann , j. , vogt , n. , phillips , a. 2005b , apj , 628 , 160 conselice , c.j .
2006a , mnras , 373 , 1389 conselice , c.j .
2006b , apj , 638 , 686 conselice , c.j . , et al .
2007 , apj , 660 , 55l daddi , e. , et al .
2004 , apj , 600 , 127l dale , d.a . , helou , g. 2002 , apj , 576 , 159 davis , m. , et al .
2003 , spie , 4834 , 161 davis , m. , et al .
2007 , apj , 660 , 1l de lucia , g. , springel , v. , white , s.d.m . , croton , d. , kauffmann , g. 2006 , mnras , 366 , 499 dickinson , m. , papovich , c. , ferguson , h.c . ,
budavri , t. 2003 , apj , 587 , 25 donas , j. , et al .
2006 , astro - ph/0608594 drory , n. , bender , r. , feulner , g. , hopp , u. , maraston , c. , snigula , j. , hill , g.j .
2004 , apj , 608 , 742 drory , n. , salvato , m. , gabasch , a. , bender , r. , hopp , u. , feulner , g. , & pannella , m. 2005 , apj , 619 , l131 dunlop , j. , peacock , j. , spinrad , h. , dey , a. , jimenez , r. , stern , d. , windhorst , r. 1996 , nature , 381 , 581 elmegreen , d.m . ,
elmegreen , b.g . ,
rubin , d.s . , schaffer , m.a .
2005 , apj , 631 , 85 elston , r. , rieke , g.h . , rieke , m.j .
1988 , apj , 331 , 77l erb , d.k .
2003 , apj , 591 , 101 faber , s. , et al .
2003 , spie , 4841 , 1657 faber , s. , et al .
2005 , astro - ph/0506044 fontana , a. , et al .
2004 , a&a , 424 , 23 forster - schreiber , n.m . ,
et al . 2006 , apj , 645 , 1062 foucaud , s. , et al .
2007 , mnras , 376 , 20l genzel , r. , et al . 2006 ,
nature , 442 , 786 georgakakis , a. , et al .
2006 , astro - ph/0606139 giavalisco , m. , et al .
2004 , apj , 600 , 93l glazebrook , k. , et al .
2004 , nature , 430 , 181 graham , a.w . ,
driver , s.p . ,
petrosian , v. , conselice , c.j . ,
bershady , m.a . ,
crawford , s.m . , goto , t. 2005 , aj , 130 , 1535 graham , j.r .
, dey , a. 1996 , apj , 471 , 720 grazian , a. , et al .
2006 , a&a , 465 , 393 grogin , n.a . ,
2005 , apj , 627 , 97l hernandez - toledo , h.m .
, avila - reese , v. , conselice , c.j . , puerari , i. aj , 2005 , aj , 129 , 682 hernandez - toledo , h.m .
, avila - reese , v.,salazar - contreras , j.r . , conselice , c.j . 2006 ,
aj , 132 , 71 hopkins , a. 2004 , apj , 615 , 209 hornschemeier , a.e . ,
2003 , aj , 126 , 575 i m , m. , et al .
2002 , apj , 571 , 136 jogee , s. , et al .
2004 , apj , 615 , 105l juneau , s. , et al .
2005 , apj , 619 , 135 kannappan , s.j . , & gawiser , e. 2007 , apj , 657 , 5l kriek , m. , et al .
2006 , apj , 649 , l71 labbe , i. , et al .
2003 , apj , 591 , 95l labbe , i. , et al .
2005 , apj , 624 , 81l lane , k.p .
, et al . 2007 , mnras , ? ?
, 55l , arxiv:0704.2136 , larson , r.b .
1974 , mnras , 166 , 585l le fevre , o. , et al .
2000 , mnras , 311 , 565 le floch , e. , et al .
2005 , apj , 632 , 169 lehmer , b.d . ,
2005 , aj , 129 , 1 lemson , g. , et al .
2006 , astro - ph/0608019 lilly , s. , le fevre , o. , crampton , d. , hammer , f. , tresse , l. 1995 , apj , 455 , 50l lin , l. , et al .
2004 , apj , 617 , 9l lotz , j.m . ,
et al . 2006 , astro - ph/0602088 maraston , c. 2005 , mnras , 362 , 799 menanteau , f. , et al .
2005 , apj , 620 , 697 mobasher , b. , jogee , s. , dahlen , t. , de mello , d. , lucas , r.a . ,
conselice , c.j . ,
grogin , n.a .
, livio , m. 2004 , apj , 600 , 143l moustakas , l.a . , et al . 2004 , apj , 600 , 131l nandra , k. , et al . 2006 , astro - ph/0607270 noeske , k. , et al
. 2007a , apj , 660 , 47l noeske , k. , et al . 2007b , apj , 660 , 43l papovich , c. , dickinson , m. , giavalisco , m. , conselice , c.j . , ferguson , h.c .
2005 , apj , 631 , 101 papovich , c. , giavalisco , m.,dickinson , m. conselice , c.j . ,
ferguson , h.c .
2003 , apj , 598 , 827 papovich , c. , et al .
2004 , apjs , 154 , 70 papovich , c. , et al . 2006 ,
apj , 640 , 92 patton , d.r . , et al .
2002 , apj , 565 , 208 persson , s.e . ,
murphy , d.c .
, krzeminksi , w. , roth , m. , rieke , m.j .
1998 , aj , 116 , 2475 ravindranath , s. , et al .
2004 , apj , 604 , 9l reddy , n.a . ,
steidel , c.c . ,
erb , d.k . ,
shapley , a.e . ,
pettini , m. 2006 , apj , 653 , 1004 rudnick , g. , et al .
2006 , apj , 650 , 624 saracco , p. , et al .
2005 , mnras , 357 , 40l sawicki , m. , yee , h.k.c .
1998 , 115 , 1329 somerville , r.s .
, lee , k. , ferguson , h.c . ,
gardner , j.p . ,
moustakas , l.a . ,
giavalisco , m. 2004 , 600 , 171l spinrad , h. , dey , a. , stern , d. , dunlop , j. , peacock , j. , jimenez , r. , windhorst , r. 1997 , apj , 484 , 581 stanford , s.a . , dickinson , m.e . ,
postman , m. , ferguson , h.c . ,
lucas , r.a .
, conselice , c.j . ,
budavari , t. , somerville , r. 2004 , aj , 127 , 131 steidel , c.c . , shapley , a.e . ,
pettini , m. , adelberger , k.l . ,
erb , d.k . ,
reddy , n.a . , hunt , m.p .
2004 , apj , 604 , 534 strateva , i. , et al .
2001 , aj , 122 , 1861 taylor - mager , v. , conselice , c. , windhorst , r. , jansen , r. 2007 , apj , 659 , 162 teplitz , h.i . ,
2006 , aj , 132 , 853 treu , t. , et al .
2005 , apj , 633 , 174 trujillo , i. , et al .
2004 , apj , 604 , 521 trujillo , i. , et al .
2006 , mnras , 373 , 36l weiner , b.j .
2006 , apj , 660 , 39l white , s.d.m . ,
rees , m.j .
1978 , mnras , 183 , 341 willmer , c.n.a . ,
et al . 2006 , apj , 647 , 853 windhorst , r. , et al .
2002 , apjs , 143 , 113 worthey , g. 1994 , apjs , 95 , 107 yan , r. , newman , j.a . ,
faber , s.m . ,
konidaris , n. , koo , d. , davis , m. 2006 , apj , 648 , 281
in this appendix we describe in more detail the degree of agreement between the cas and visual estimates of morphology .
we are particularly interested in describing why some systems classified as a particular type by eye fall into a different region of cas space than what their apparent morphology would indicate .
a more limited discussion of this issue has been presented in conselice ( 1997 ) , bershady et al .
( 2000 ) , mobasher et al . (
2004 ) and conselice et al .
( 2005a ) .
before we discuss why there are differences between visual estimates of morphological type and the quantitative approach , it is important to describe the process of morphological classification by eye .
this process was carried out by one of us ( cjc ) by examining a galaxy s structure on a computer screen .
this was done through iraf and the viewing tool ds9 , although similar results can be had through any approach that lets the viewer change the contrast .
classification was done into the types described in 4.3 using the traditional approach of examining the entire galaxy at once , and judging which of the categories it belongs to .
when this process is done we do not take into account the smaller , more subtle , details of a galaxy s structure .
for example , although most of our sample of @xmath126 galaxies consists of ellipticals , about a third have some kind of morphological peculiarity . typically , this was either a diffuse outer envelope of material , or a lopsided centre with respect to the rest of the galaxy .
this is clearly seen the cas space for these systems - the peculiar ellipticals have higher asymmetry and clumpiness values than ellipticals , compacts and s0s . however , there are cases in which a galaxy has been classified as one type but appears in a region of the cas space where it ordinarily should not be .
examples of this include ellipticals with high asymmetries , or low concentrations , and peculiar galaxies with low asymmetries .
while on average these galaxy populations are found in the location expected within the cas space , there are some obvious exceptions . after creating the first version of figure 9
we went back and examined by eye the @xmath296% of galaxies whose visual morphology differed greatly from their measured cas values .
about half of the time it turned out that the galaxy was misclassified by eye or recorded incorrectly .
this always accounted for the most obvious cases where the cas values and eye estimates of morphology differed the most .
there are however systems that still differ in terms of their visual estimates of morphology and their measured cas values .
it turns out that the cas parameters are most successful at distinguishing disk galaxies , ellipticals and mergers from each other ( cassatta et al .
we therefore focus on these three populations .
first , perhaps the most obvious disagreement is the high - a ellipticals and the low - a peculiars .
the ellipticals classified by eye that contain a high asymmetry were nearly always systems that had some morphological peculiarity .
in other cases , it was determined that the high - a elliptical / s0 systems were nearly edge - on , or otherwise peculiar , s0s that often contained a prominent dust lane .
other examples were ellipticals that contained neighbouring galaxies that created a higher - a signal .
the peculiars that are at @xmath297 show a diversity of visual morphologies . often these systems were compact and concentrated with some diffuse material , or had blob like features nearby . in many cases
these systems appear to be ellipticals in assembly , and could have easily been classified as peculiar ellipticals .
other examples of lower - a peculiars are systems imaged with a low - s / n . as has been argued in conselice et al .
( 2000 ) and conselice ( 2003 ) a lower s / n will make it more difficult to measure quantitative indices for these systems .
several systems resembled spiral galaxies , or spirals in assembly , with what appears to be arms in development .
it thus appears that quantitative measures of structures are revealing information that eye - ball estimates are missing , namely that a gross morphology can be present while the galaxy structure still retains some signatures of recent formation .
this creates significant differences when comparing galaxies of the same type , as selected by eye - ball estimates , at low and high redshifts . | we present the results of a study on the properties and evolution of massive ( m@xmath0 ) galaxies at @xmath1 utilising keck spectroscopy , near - infrared palomar imaging , and hubble , chandra , and spitzer data covering fields targeted by the deep2 galaxy spectroscopic survey .
our sample is k - band selected and stellar mass limited , based on wide - area near - infrared imaging from the palomar observatory wide - field infrared survey , which covers 1.53 deg@xmath2 to a 5 @xmath3 depth of @xmath4 .
our primary goal is obtaining a broad census of massive galaxies through measuring how their number and mass densities , morphology , as well as their star formation and agn content evolve from @xmath1 .
our major findings include : ( i ) statistically the mass and number densities of m@xmath5 galaxies show little evolution between @xmath6 , and from @xmath7 for m@xmath5 galaxies .
we however find significant evolution between @xmath8 for @xmath9 m@xmath10 galaxies .
( ii ) after examining the structures of our galaxies using hubble acs imaging , we find that m@xmath5 selected galaxies show a nearly constant elliptical fraction of @xmath11 at all redshifts .
the remaining objects tend to be peculiars possibly undergoing mergers at @xmath12 , while spirals dominate the remainder at lower redshifts .
a significant fraction ( @xmath13 25% ) of these early - types contain minor structural anomalies .
( iii ) we find that only a fraction ( @xmath14% ) of massive galaxies with m@xmath5 are on the red - sequence at @xmath15 , while nearly 100% evolve onto it by @xmath16 .
( iv ) by utilising spitzer mips imaging and [ oii ] line fluxes we argue that m@xmath5 galaxies have a steeply declining star formation rate density @xmath17 . by examining the contribution of star formation to the evolution of the mass function , as well as the merger history through the cas parameters
, we determine that m@xmath5 galaxies undergo on average @xmath18 major mergers at @xmath19 .
( v ) we find that a high ( 5% ) fraction of all m@xmath5 galaxies are x - ray emitters .
roughly half of these are morphologically distorted ellipticals or peculiars . finally , we compare our mass growth with semi - analytical models from the millennium simulation , finding relative good agreement at @xmath20 for the m@xmath21 systems , but that the number and mass densities of m@xmath22 galaxies are under predicted by a factor of @xmath23 .
[ firstpage ] galaxies : evolution , formation , structure , morphology , classification |
at the ongoing and future b - physics experiments a realistic estimate of the uncertainties in the values of the bottom and charm quark masses will become increasingly important for the measurements of the ckm parameters and the search for new physics , particularly from inclusive b - decay rates .
@xcite moments of the respective @xmath3 r - ratios @xcite , @xmath7 where @xmath8 , are the most important instrument at present to determine the bottom and charm quark mass parameters .
( see also refs .
@xcite for recent bottom quark mass extractions from spectral moments in semileptonic b - decays . ) in general , one can distinguish between two regions in @xmath9 , which require a different theoretical treatment . for low values of @xmath9
the moments are dominated by relativistic dynamics and scales of order of the heavy quark mass @xmath10 .
this allows the use of the usual expansion in the number of loops for the theoretical computations , and the @xmath0 scheme is an appropriate choice for the heavy quark mass parameter .
however , the lack of data for @xmath11 in the continuum regions above the quarkonium resonances introduces model - dependent errors that have not been quantified in most of the previous analyses .
( see refs.@xcite for recent reviews . ) on the other hand , for large values of @xmath9 the continuum regions are suppressed and the moments become dominated by the quarkonium resonance region where good sets of data have been obtained in the past .
however , the theoretical predictions of moments for large values of @xmath9 is more complicated since the usual loop expansion breaks down and the size of non - perturbative effects increases . here
, summations of higher order contributions proportional to powers of @xmath12 need to be carried out in order to capture the relevant non - relativistic perturbative information@xcite , and so - called threshold masses @xcite are appropriate schemes for the heavy quark mass parameter .
moreover , there is an upper `` duality '' bound for the possible choices of @xmath9 since the energy range contributing to the moments , which is of order @xmath13 , needs to be larger than the typical hadronization scale @xmath14.@xcite in the case of @xmath15 and the determination of the bottom mass , this bound is around @xmath16 , and the two ranges in @xmath9 are believed to be well separated with their boundary being approximately at @xmath17 .
a good number of analyses exists for small and large values of @xmath9 and respecting the cancellation of the dominant renormalon contributions associated to the choice of the quark mass definition.@xcite ( see also ref.@xcite for an early analysis using the @xmath0 bottom quark mass for larger values of @xmath9 . ) for @xmath18 and the determination of the charm mass , the distinction between large- and low-@xmath9 moments is more delicate because @xmath19 is not much larger than @xmath20 . here
, the upper duality bound for @xmath9 is around @xmath21 or @xmath22 .
this leaves basically no space at all to carry out the non - relativistic summations that can be applied in the bottom quark case because the corresponding techniques are only valid for large @xmath9 and as long as @xmath13 is larger than @xmath20 . on the other hand , even for @xmath23 the non - relativistic region close to the @xmath24 threshold can have a considerable contribution to the moments , while the model - dependences from the experimentally unknown continuum region can still be significant . for a reliable ( error )
analysis these issues need to be taken into account . in this paper
we determine the @xmath0 charm quark mass @xmath25 from moments @xmath26 with @xmath23 using the usual loop expansion in powers of @xmath27 . in a previous work @xcite an error of less than 30 mev was obtained using results at order @xmath2 and a fixed renormalization scale @xmath28 between @xmath19 and @xmath29 .
this analysis assumed that the moments are dominated by scales of order @xmath19 . in our analysis
we focus on the theoretical uncertainties coming from the low - energy region close to the @xmath24 threshold .
it is well known that close to threshold , where the c.m .
velocity @xmath30 of the quarks is small , the dominant perturbative contributions to @xmath18 are in fact governed by renormalization scales of the order of the relative quark momentum , @xmath31.@xcite we believe that this effect should not be ignored in determinations of the charm quark mass since it accounts for important higher - order information .
it is the main aim of this work to study the impact of these higher - order contributions . in particular
, we find that they affect the estimate of the theoretical error .
we also carry out a careful study of all sources of uncertainties including the continuum regions where no data is available .
for the qcd parameters used in this work we adopt the @xmath0 renormalization scheme and the convention that the charm quark participates in the running ( @xmath32 ) .
the theoretical moments are directly parametrized in terms of @xmath33 , and the masses of the up , down and strange quarks are set to zero .
for the running of the strong coupling we use four - loop renormalization group equations , and three - loop matching conditions at the scale @xmath34 gev where we switch to @xmath32 flavors .
we use three different methods to implement the renormalization scale for the theoretical predictions of the moments and apply order @xmath2 results obtained earlier in refs.@xcite . the first method is simply based on a fixed energy - independent renormalization scale , @xmath35 , in analogy to previous analyses . in the ope including the first non - perturbative contribution from the dimension - four gluon condensate , the moments have the form @xmath36 where @xmath37\bigg\}\ ] ] and @xmath38 \,.\end{aligned}\ ] ] here , @xmath39 .
the order @xmath27 terms were given in refs.@xcite and the coefficients for the condensate contribution were taken from refs.@xcite adopting the renormalization - group - invariant normalization of the gluon condensate .
the numerical results for the coefficients in eq.([mn1 ] ) for @xmath40 are collected in tab.[tabcoeff1 ] .
.[tabcoeff1 ] coefficients of the theoretical expressions for the moments @xmath26 at order @xmath2 for massless light quarks based on a fixed energy - independent renormalization scale .
[ cols="^,>,>,>,>",options="header " , ] for the determination of @xmath41 we fit single moments using the three different methods for the theoretical predictions described above .
the contribution from the gluon condensate is taken from the first method for all cases .
we use @xmath42 ( @xmath43 ) and @xmath44 as theoretical input .
our final results for @xmath40 , for the three different theoretical methods and for varying the renormalization scale parameter in the ranges @xmath45 and @xmath46 are given in tab.[tabmcfinal ] .
the central values are obtained using the respective central values for all experimental data and the theoretical input parameters , and by taking the median of the range in @xmath41 covered by varying @xmath47 in the respective ranges .
the uncertainties are obtained by combining all individual error sources as described below and shown in tab.[tabmcerrors ] . in tab.[tabmcfinal ] we have also given combined results representing the total range in @xmath48 covered by the three theoretical methods .
@xmath9 & 1 & 2 & 3 & 4 & 1 & 2 & 3 & 4 + + central & 1280 & 1272 & 1267 & 1264 & 1287 & 1296 & 1298 & 1298 + @xmath49 & 12/12 & 9/9 & 7/7 & 6/6 & 11/11 & 9/9 & 7/7 & 6/6 + @xmath50 & 3/3 & 1/1 & 1/1 & 0/0 & 3/3 & 1/1 & 1/1 & 0/0 + @xmath51 gev & 2/10 & 1/4 & 0/2 & 0/1 & 2/10 & 1/4 & 0/2 & 0/1 + @xmath52 gev & 12 & 3 & 1 & 0 & 11 & 3 & 1 & 0 + @xmath53 gev & 1/3 & 0/0 & 0/0 & 0/0 & 1/3 & 0/0 & 0/0 & 0/0 + @xmath54 gev @xmath55 & 4 & 0 & 0 & 0 & 4 & 0 & 0 & 0 + @xmath56 & 4 & 8 & 11 & 14 & 4 & 6 & 8 & 9 + @xmath57 & 14 & 8 & 5 & 3 & 17 & 17 & 17 & 18 + @xmath58 & 28 & 18 & 12 & 9 & 25 & 19 & 15 & 12 + combined error & 101 & 61 & 47 & 41 & 100 & 69 & 58 & 53 + + central & 1268 & 1264 & 1262 & 1260 & 1278 & 1291 & 1294 & 1295 + @xmath49 & 12/12 & 9/9 & 7/7 & 6/6 & 12/12 & 9/9 & 7/7 & 6/6 + @xmath50 & 3/3 & 1/1 & 1/1 & 0/0 & 3/3 & 1/1 & 1/1 & 0/0 + @xmath51 gev & 2/10 & 1/4 & 0/2 & 0/1 & 2/10 & 1/4 & 0/2 & 0/1 + @xmath52 gev & 12 & 3 & 1 & 0 & 12 & 3 & 1 & 0 + @xmath53 gev & 1/3 & 0/0 & 0/0 & 0/0 & 1/3 & 0/0 & 0/0 & 0/0 + @xmath54 gev @xmath55 & 4 & 0 & 0 & 0 & 4 & 0 & 0 & 0 + @xmath56 & 5 & 9 & 14 & 20 & 5 & 7 & 10 & 14 + @xmath57 & 11 & 5 & 3 & 2 & 11 & 10 & 9 & 9 + @xmath58 & 24 & 15 & 10 & 7 & 24 & 20 & 18 & 14 + combined error & 92 & 56 & 45 & 43 & 94 & 65 & 55 & 52 + a detailed account of individual errors is presented in tab.[tabmcerrors ] . for experimental data
the first error is statistical and the second systematical .
the results are only given for the second and the third theoretical method , as the first and second methods yield very similar results , particularly for @xmath59 .
this is expected since the high - energy range , @xmath60 , where the renormalization scales of the two methods differ , is suppressed .
the uncertainties from experimental data and from the treatment of the non - perturbative corrections are similar for the three theoretical methods . as mentioned before
, the model - dependent error coming from the lack of experimental data in the continuum region is negligible for @xmath59 and amounts to only about 10 mev for @xmath61 .
this is in contrast to determinations of @xmath62 where the model - dependent error is significantly larger , particularly for @xmath61 and @xmath63 .
@xcite the uncertainties coming from the error in @xmath64 for @xmath59 are larger for the third method than for the first and the second one , since variations in the value of @xmath64 have a larger impact at lower scales .
to obtain combined errors we have added all statistical uncertainties in quadrature , and all other uncertainties linearly .
since the statistical errors are smaller than the other sources of uncertainties , the reader should note that our total errors can not be interpreted statistically .
it is instructive to have a closer look onto the perturbative uncertainties obtained by the different theoretical methods .
for all methods the uncertainties obtained by varying @xmath47 either in the range @xmath65 or @xmath46 become smaller for larger values of @xmath9 .
this behavior is , however , less pronounced for the third method .
thus , each of the methods yields a behavior in contrast to the expectation that the scaling uncertainties grow like the uncertainties from the non - perturbative corrections , since for larger @xmath9 the moments become dominated by dynamics at lower energies . on the other hand , comparing the results for the central values obtained by the three theoretical methods it is found that they tend to decrease for larger values of @xmath9 for the first and second method , while there is the opposite effect for the third method .
in fact , the difference between the central values for @xmath59 is larger than the uncertainties based on varying @xmath47 .
this feature is related to the different treatment of higher - order contributions and appears to be similar to the discrepancies found between the `` contour - improved '' and the `` fixed - scale '' approaches for predictions of the hadronic @xmath66-decay rate .
( see e.g. refs.@xcite for discussions on this issue ) .
interestingly , the convergence behavior of the moments for all three methods is quite good .
for example , for @xmath67 , @xmath68 gev , @xmath69 and @xmath70 one obtains ( in units of @xmath71 gev@xmath72 ) @xmath73 , @xmath74 and @xmath75 for the perturbative expansion of the moment for the first , second and the third method , respectively .
the discrepancy between the first two methods and the third one is considerably larger than an error estimate based on the size of the order @xmath2 terms obtained from each of the methods .
these discrepancies represent theoretical uncertainties that are not captured by conventional variations in the renormalization scale .
however , they should be included in the estimate of the theoretical error .
we include this type of theoretical uncertainty in our procedure to determine the combined results shown in tab.[tabmcfinal ] . as our final result
we adopt @xmath76 obtained from our fits using the moments for @xmath77 .
they are sufficiently below the duality bound and , at the same time , have only small sensitivity to the continuum region above the @xmath50 resonance .
our result is compatible with earlier analyses based on moments of @xmath18 @xcite , and with recent lattice determinations @xcite as well as with the finite energy sum rule analysis of ref.@xcite .
however , our error is much larger than in refs.@xcite due to our more pertinent treatment of theoretical and experimental uncertainties . on the other hand , the size of our error is comparable to refs.@xcite where summations of terms proportional to powers of @xmath12 @xcite were partially included to account for non - relativistic effects in the low - energy region @xmath78 .
however , the resulting central values are systematically lower than ours by up to @xmath79 mev .
this effect apparently arises from the summation of the higher order terms just mentioned and from the use of moments for @xmath80 .
both issues are problematic from the conceptual point of view .
the size of our error is also comparable to the analysis of ref.@xcite , but our central value is about @xmath79 mev lower .
this discrepancy seems to indicate that there is a systematic difference between the sum rules based on eq.([momdefcomplex ] ) and the finite energy moments employed in ref.@xcite .
we note , however , that the strong influence of systematic errors in the derivation of the non - charm r ratio @xmath81 and possible ambiguities in the choice of the upper energy cut - off represent sources of uncertainties for the finite energy sum rule approach that are not easy to quantify .
based on the numbers given in tabs.[tabmcfinal ] and [ tabmcerrors ] it is straightforward to discuss how the uncertainty in @xmath41 might be further reduced in the future .
one improvement could be achieved by the computation of the order @xmath82 corrections to the moments if they indeed lead to a reduction of the discrepancies between `` contour - improved '' and `` fixed - scale '' predictions .
at present the theoretical uncertainty coming from the different ways to implement the renormalization scale amounts to about @xmath83 to @xmath84 mev .
the uncertainties coming from the @xmath49 @xmath3-partial width , from the error in @xmath27 and from the treatment of non - perturbative effects are each at the level of @xmath85 mev . for the combination of these uncertainties a reduction of up to @xmath86 mev appears possible for an improved measurement of the @xmath3-partial width and if smaller uncertainties in the determinations of the strong coupling are achieved . on the other hand , improvements in the data for @xmath50 and for the @xmath24 continuum will not improve the situation significantly .
we have also applied our method to estimate the theoretical uncertainties for determinations of @xmath87 from low-@xmath9 moments of @xmath15 . using the approach described above we find @xmath88 gev .
the error is only @xmath89 mev larger than the error obtained from an earlier analysis by one of the authors which was based only the first method with an energy - independent renormalization scale .
this shows that the scale choice @xmath90 has a smaller impact in the bottom quark case due to its larger mass .
10 m. battaglia _ et al .
_ , arxiv : hep - ph/0304132 . p. gambino and m. misiak , nucl
. phys .
b * 611 * , 338 ( 2001 ) [ arxiv : hep - ph/0104034 ] .
v. a. novikov _ et al .
_ , phys .
lett . * 38 * , 626 ( 1977 ) [ erratum - ibid . *
38 * , 791 ( 1977 ) ] .
l. j. reinders , h. r. rubinstein and s. yazaki , nucl .
b * 186 * , 109 ( 1981 ) . c. w. bauer , z. ligeti , m. luke and a. v. manohar , phys .
d * 67 * , 054012 ( 2003 ) [ arxiv : hep - ph/0210027 ] .
a. h. mahmood _ et al .
_ , phys .
d * 67 * , 072001 ( 2003 ) [ arxiv : hep - ex/0212051 ] .
a. h. hoang , arxiv : hep - ph/0204299 ; + a. x. el - khadra and m. luke , arxiv : hep - ph/0208114 . m. b. voloshin , int .
j. mod .
phys . a * 10 * , 2865 ( 1995 ) [ arxiv : hep - ph/9502224 ] .
a. h. hoang , phys .
d * 59 * , 014039 ( 1999 ) [ arxiv : hep - ph/9803454 ] .
i. i. y. bigi , m. a. shifman , n. uraltsev and a. i. vainshtein , phys .
d * 56 * , 4017 ( 1997 ) [ arxiv : hep - ph/9704245 ] ; a. czarnecki , k. melnikov and n. uraltsev , phys .
lett . * 80 * , 3189 ( 1998 ) [ arxiv : hep - ph/9708372 ] .
a. h. hoang , z. ligeti and a. v. manohar , phys .
lett . * 82 * , 277 ( 1999 ) [ arxiv : hep - ph/9809423 ] ; a. h. hoang , z. ligeti and a. v. manohar , phys . rev .
d * 59 * , 074017 ( 1999 ) [ arxiv : hep - ph/9811239 ] .
m. beneke , phys .
b * 434 * , 115 ( 1998 ) [ arxiv : hep - ph/9804241 ] .
e. c. poggio , h. r. quinn and s. weinberg , phys .
d * 13 * , 1958 ( 1976 ) .
j. h. kuhn and m. steinhauser , nucl .
b * 619 * , 588 ( 2001 ) [ erratum - ibid .
b * 640 * , 415 ( 2002 ) ] [ arxiv : hep - ph/0109084 ] .
g. corcella and a. h. hoang , phys .
b * 554 * , 133 ( 2003 ) [ arxiv : hep - ph/0212297 ] .
k. melnikov and a. yelkhovsky , phys .
d * 59 * , 114009 ( 1999 ) [ arxiv : hep - ph/9805270 ] .
a. h. hoang , phys .
d * 61 * , 034005 ( 2000 ) [ arxiv : hep - ph/9905550 ] ; a. h. hoang , arxiv : hep - ph/0008102 .
m. beneke and a. signer , phys .
b * 471 * , 233 ( 1999 ) [ arxiv : hep - ph/9906475 ] .
m. jamin and a. pich , nucl .
b * 507 * , 334 ( 1997 ) [ arxiv : hep - ph/9702276 ] ; m. jamin and a. pich , nucl .
suppl . * 74 * , 300 ( 1999 ) [ arxiv : hep - ph/9810259 ] .
a. h. hoang , j. h. kuhn and t. teubner , nucl .
b * 452 * , 173 ( 1995 ) [ arxiv : hep - ph/9505262 ] .
a. czarnecki and k. melnikov , phys .
* 80 * , 2531 ( 1998 ) [ arxiv : hep - ph/9712222 ] ; m. beneke , a. signer and v. a. smirnov , phys .
lett . * 80 * , 2535 ( 1998 ) [ arxiv : hep - ph/9712302 ] .
a. h. hoang , m. jezabek , j. h. kuhn and t. teubner , phys .
b * 338 * , 330 ( 1994 ) [ arxiv : hep - ph/9407338 ] ; k. g. chetyrkin , j. h. kuhn and m. steinhauser , nucl .
b * 505 * , 40 ( 1997 ) [ arxiv : hep - ph/9705254 ] .
k. g. chetyrkin , a. h. hoang , j. h. kuhn , m. steinhauser and t. teubner , eur .
j. c * 2 * , 137 ( 1998 ) [ arxiv : hep - ph/9711327 ] .
d. j. broadhurst _ et al .
_ , phys .
b * 329 * , 103 ( 1994 ) [ arxiv : hep - ph/9403274 ] .
a. baikov , v. a. ilyin and v. a. smirnov , phys .
* 56 * ( 1993 )
1527 [ yad . fiz . * 56n11 * ( 1993 ) 130 ] .
f. le diberder and a. pich , phys .
b * 286 * , 147 ( 1992 ) .
a. a. pivovarov , z. phys .
c * 53 * , 461 ( 1992 ) [ sov .
j. nucl .
phys . * 54 * , 676 ( 1991 yafia,54,1114.1991 ) ] [ arxiv : hep - ph/0302003 ] .
k. hagiwara _ et al .
_ [ particle data group collaboration ] , phys . rev .
d * 66 * , 010001 ( 2002 ) .
j. z. bai _ et al .
_ [ bes collaboration ] , phys . rev .
lett . * 88 * , 101802 ( 2002 ) [ arxiv : hep - ex/0102003 ] .
a. e. blinov _ et al .
_ , z. phys .
c * 70 * , 31 ( 1996 ) .
r. ammar _ et al . _
[ cleo collaboration ] , phys .
d * 57 * , 1350 ( 1998 ) [ arxiv : hep - ex/9707018 ] .
the lep electroweak working group et al .
, arxiv : hep - ex/0212036 . s. groote , j. g. korner and a. a. pivovarov , phys .
b * 407 * , 66 ( 1997 ) [ arxiv : hep - ph/9704396 ] .
a. baikov , k. g. chetyrkin and j. h. kuhn , phys .
d * 67 * , 074026 ( 2003 ) [ arxiv : hep - ph/0212299 ] .
m. eidemuller and m. jamin , phys .
b * 498 * , 203 ( 2001 ) [ arxiv : hep - ph/0010334 ] .
j. erler and m. x. luo , phys .
b * 558 * , 125 ( 2003 ) [ arxiv : hep - ph/0207114 ] .
b. l. ioffe and k. n. zyablyuk , eur .
j. c * 27 * , 229 ( 2003 ) [ arxiv : hep - ph/0207183 ] .
m. eidemuller , phys .
d * 67 * , 113002 ( 2003 ) [ arxiv : hep - ph/0207237 ] .
d. becirevic , v. lubicz and g. martinelli , phys .
b * 524 * , 115 ( 2002 ) [ arxiv : hep - ph/0107124 ] .
k. j. juge , nucl .
suppl . * 106 * , 847 ( 2002 ) [ arxiv : hep - lat/0110131 ] .
j. rolf and s. sint [ alpha collaboration ] , jhep * 0212 * , 007 ( 2002 ) [ arxiv : hep - ph/0209255 ] . j. penarrocha and k. schilcher , phys .
b * 515 * , 291 ( 2001 ) [ arxiv : hep - ph/0105222 ] . | a detailed error analysis is carried out for the determination of the @xmath0 charm quark mass @xmath1 from moments at order @xmath2 of the charm cross section in @xmath3 annihilation . to estimate the theoretical uncertainties the renormalization scale
is implemented in various ways including energy - dependent functions , which lead to `` contour - improved '' predictions .
we obtain @xmath4 gev which contains a substantial theoretical uncertainty .
mpi - pht 2004 - 29 + hd - thep 04 - 6 + march 2004 a. h. hoang@xmath5 and m. jamin@xmath6 _ @xmath5max - planck - institut fr physik + werner - heisenberg - institut + fhringer ring 6 , 80805 mnchen + germany _ _ @xmath6institut fr theoretische physik + universitt heidelberg + philosophenweg 16 , 69120 heidelberg + germany _ |
in most plasmas , the forces acting on the ions are different from those acting on the electrons , so naively one would expect one species to be transported faster than the other , by diffusion or convection or some other process .
if such differential transport has a divergence , then it will result in a change of the charge density , which will in turn create an electric field that will alter the transport of one or both species in such a way that they become equal . in astrophysics , _ ambipolar diffusion _ refers specifically to the decoupling of neutral particles from plasma .
the neutral particles in this case are mostly hydrogen molecules in a cloud , and the plasma is composed of ions ( mostly protons ) and electrons , which are tied to the interstellar magnetic field .
nejad - asghar ( 2007 ) has recently made the assumption that the molecular cloud is initially an uniform ensemble which then fragments due to thermal instability .
he finds that ambipolar drift heating is inversely proportional to density and its value , in outer parts of the cloud , can be significantly larger than the average heating rates of cosmic rays and turbulent motions .
his results show that the isobaric thermal instability can occur in outer regions of the cloud . the rapid growth of thermal instability results a strong density imbalance between the cloud and the low - density surroundings ; therefore it may produce the cloud fragmentation and formation of the condensations .
many authors have developed computer codes that attempt to model ambipolar diffusion .
black & scott ( 1982 ) used a two - dimensional , deformable - grid algorithm to follow the collapse of isothermal , non - rotating magnetized cloud .
the three - dimensional work of maclow et al .
( 1995 ) treats the two - fluid model in a version of the zeus magnetohydrodynamic code .
an algorithm capable of using the smoothed particle hydrodynamics ( sph ) to implement the ambipolar diffusion in a fully three - dimensional , self - gravitating system was developed by hosking & whitworth ( 2004 ) .
they described the sph implementation of two - fluid technique that was tested by modeling the evolution of a dense core , which is initially thermally supercritical but magnetically subcritical .
nejad - asghar & molteni ( 2007 ) have recently optimized the two - fluid sph implementation to test the pioneer works on the behavior of the ambipolar diffusion in an isothermal self - gravitating molecular layer . in this paper
, we include the sph equivalent of the energy equation , which follows from the first law of thermodynamics . here , we include the ambipolar drift heating rate in the net cooling function , and we investigate the thermal phases in the self - gravitating magnetized molecular layer . for this purpose ,
the two - fluid sph technique and its algorithm are given in section 2 .
section 3 devotes to the basic equations of the one dimensional molecular layer and its evolution by sph .
the summary and conclusion are presented in section 4 .
in two - fluid sph technique of hosking & whitworth ( 2004 ) , the initial sph particles represent by two sets of _ molecular particles _ : magnetized ion sph particles and non - magnetized neutral sph particles . in this method , for each sph particle we must create two separate neighbor lists : one for neighbors of the same species and another for those of different species .
consequently , each particle has two different smoothing lengths . in the following sections we refer to neutral particles as @xmath0 and @xmath1 , and ion particles as @xmath2 and @xmath3
; the subscripts @xmath4 and @xmath5 refer to both ions and neutral particles .
we adopt the usual smoothing spline - based kernel ( monaghan & lattanzio 1985 ) and apply the symmetrized form proposed by hernquist & katz ( 1989 ) @xmath6 where @xmath7 , @xmath8 are positions of the particles @xmath4 and @xmath5 , respectively .
@xmath9 is the smoothing length of particle @xmath4 when considering neighbors of kind @xmath5 .
we update the adaptive smoothing length according to the octal - tree based nearest neighbor search algorithm ( nns ) .
the exact fraction of the total fluid that is ionized depends upon many factors ( e.g. the neutral density , the cosmic ray ionization rate , how efficiently ionized metals are depleted on to dust grains ) .
here , we use the expression employed by fiedler & mouschovias ( 1992 ) , which states that for @xmath10 , @xmath11 where in standard ionized equilibrium state , @xmath12 and @xmath13 are valid . in reality , the gas in this case is very weakly ionized , thus , we adopt the approximation @xmath14 in fluid equations . in this case , the molecular cloud is considered as global neutral which consists of a mixture of atomic and molecular hydrogen ( with mass fraction @xmath15 ) , helium ( with mass fraction @xmath16 ) , and traces of @xmath17 and other rare molecules , thus , the mean molecular weight is given by @xmath18 .
the neutral density in place of neutral particles is estimated via usual summation over neighboring neutral particles @xmath19 while in place of ions , @xmath20 , is given by interpolation technique from the values of the nearest neighbors .
the ion density is evaluated via equation ( [ ionden ] ) as follows @xmath21 for both places of ions and neutral particles . according to this new ion density ,
we update the mass of ion @xmath2 as @xmath22 in each time step so that the usual summation law @xmath23 for ions might being appropriate .
the sph form of the drift velocity of ion particle @xmath2 is given by hosking & whitworth ( 2004 ) as @xmath24.\end{aligned}\ ] ] where @xmath25 is the artificial viscosity between ion particles @xmath2 and @xmath3 @xmath26 where @xmath27 is an average density , @xmath28 and @xmath29 are the coefficients , @xmath30 is the mean sound speed , and @xmath31 is defined as its usual form @xmath32 with @xmath33 .
nejad - asghar , khesali & soltani ( 2008 ) has recently considered the coefficients in the monaghan s standard artificial viscosity as time variable , and a restriction on them is proposed such that avoiding the undesired effects in the subsonic regions . here
, we use the monaghan s standard artificial viscosity , since the cloud contraction is quasi - hydrostatic and there is not supersonic motions and shock formation during this contraction .
it is better to keep in mind the second golden rule of sph which is to rewrite formulae with the density inside operators ( monaghan 1992 ) . in this case , the drift velocity is interpolated as @xmath34,\end{aligned}\ ] ] where two extra density terms are introduced , one outside and one inside the summation sign .
this comes as a result of the approximation to the volume integral needed to perform function interpolation ( nejad - asghar & molteni 2007 )
. there is no analytical expression that allows to calculate the value of drift velocity of the neutral particles . here
, we use the interpolation technique that starts at the nearest neighbor , then add a sequence of decreasing corrections , as information from other neighbors is incorporated ( e.g. , press et al .
these drift velocities at neutral places are used to estimate the drag acceleration @xmath35 instead the method of hosking & whitworth ( 2004 ) who used the expression of monaghan & kocharayan ( 1995 ) . in the usual symmetric form , the self - gravitating sph acceleration equation for neutral particle
@xmath0 is @xmath36 where @xmath37 is the gravitational acceleration of particle @xmath0 .
the ion momentum equation assuming instantaneous velocity update so that we have @xmath38 where the first term on the right - hand side gives the neutral velocity field at the ion particle @xmath2 , calculated using a standard sph approximation .
finally , the magnetic induction equation in sph is @xmath39 where the usual notation is used . in ambipolar diffusion process , the ion particles are physically diffused through the neutral fluid , thus , the ions will be bared in the boundary regions of the cloud ( i.e. without any neutral particles in their neighbors ) .
we check the position of ions before making a tree and nearest neighbor search , so that we send out the bared ion particles in the boundary regions of the simulation at next time - step . both the cloud and boundary regions contain ion and neutral particles .
we set up boundary particles ( @xmath40 up and down in @xmath41 ) using the linear extrapolation approach ( from the values of the inner particles ) to attribute the appropriate drift velocity , drag acceleration , pressure acceleration , and the magnetic induction rate to the boundary particles .
the algorithm of the two - fluid sph for simulation of thermal phases in a one dimensional molecular cloud is shown in fig .
[ algor ] .
we write the continuity equation of neutral part as its common form @xmath42 while , the relation ( [ ionden ] ) is used to determine the ion density wherever it is required in the two - fluid equations .
the momentum equation of neutral part then becomes @xmath43 where the gravitational acceleration @xmath44 obeys the poisson s equation @xmath45 and the pressure is given by the ideal gas equation of state @xmath46 where @xmath47 is the mass of hydrogen atom and the temperature @xmath48 is approximated the same as for both neutral and ion fluids ( @xmath49 ) .
the thermal energy per unit mass is @xmath50 where the mean internal ( rotation and vibration ) energy of an @xmath51 molecule is included .
the energy equation follows from the first law of thermodynamics , that is @xmath52 where @xmath53 is the net cooling function @xmath54 where @xmath55 and @xmath56 are the heating rates due to cosmic rays and ambipolar diffusion , respectively , and @xmath57 and @xmath58 are the parameters for the gas cooling function that here we use the polynomial fitting functions , outlined by nejad - asghar ( 2007 ) , as follows @xmath59 @xmath60 where @xmath61 . the magnetic fields are directly evolved by charged fluid component , as follows : @xmath62 where the last term outlines the ambipolar diffusion effect with drift velocity , @xmath63 which is obtained by assumption that the pressure and gravitational force on the charged fluid component are negligible compared to the lorentz force because of the low ionization fraction .
the notation @xmath64 in drift velocity represents the collision drag coefficient .
the chosen physical scales for length and time are @xmath65=200 \mathrm{au}$ ] , and @xmath66=10 ^ 3 \mathrm{yr}$ ] , respectively , so that velocity unit is approximately @xmath67=1 \mathrm{km.s^{-1}}$ ] .
the newtonian constant of gravitation is set @xmath68^{-1 } [ l]^3 [ t]^{-2}$ ] for which the calculated mass unit is @xmath69= 4.5 \times 10^{29 } \mathrm{kg}$ ] .
consequently , the derived physical scale for density , energy per unit mass , and drag coefficient are @xmath70= 1.7\times 10^{-11 } \mathrm{kg.m^{-3}}$ ] , @xmath71= 10^{6 } \mathrm{j.kg^{-1}}$ ] , and @xmath72 ^ 3 [ m]^{-1 } [ t]^{-1}$ ] , respectively . in this manner ,
the numerical values of @xmath73 and @xmath74 are @xmath75^{-3/2 } [ m]^{1/2}$ ] and @xmath76^{-15/2 } [ m]^{5/2}$ ] , respectively .
the magnetic field is scaled in units such that the constant @xmath77 is unity .
since the magnetic flux density has dimensions @xmath78=\frac{[m]}{[t][charge]},\ ] ] while @xmath77 has dimensions @xmath79=\frac{[m][l]}{[charge]^2},\ ] ] specifying @xmath80 therefore scales the magnetic field equal to @xmath81=5.1 \mathrm{nt}$ ] . with aforementioned units ,
the thermal energy per unit mass ( [ ut ] ) is represented by @xmath82 , the heating rates due to cosmic rays and ambipolar diffusion are @xmath83/[t]$ ] and @xmath84 respectively , and the parameters for the gas cooling function are @xmath85/[t]}\right ) = -4.48 - 0.87 ( \log \frac{\rho_\alpha}{2.24\times10^{-4 } } ) \\ - 0.14 ( \log \frac{\rho_\alpha}{2.24\times10^{-4}})^2,\end{aligned}\ ] ] @xmath86 the initial conditions for this simulation are a parallel magnetic field directed perpendicular to the @xmath41-axis so that the initial ratio of magnetic to gas pressure is everywhere a constant ( @xmath87 ) , and a density profile given by @xmath88 where @xmath89 is the initial central density and @xmath90 is a length - scale parameter according to the initial sound speed @xmath91 ( see fig . 1 of nejad - asghar & molteni 2007 ) . the magnetic field is assumed to be frozen in the fluid of charged particles and the central density is assumed to be @xmath92 $ ] .
we choose a molecular cloud with the mass fraction of molecular hydrogen and helium are @xmath93 and @xmath94 , respectively , and have initial uniform temperature of @xmath95 .
we assume that the cloud layer is spread from @xmath96 $ ] to @xmath97 $ ] ( according to fig . 1 of nejad - asghar & molteni 2007 ) .
the initial values of the cooling and heating functions are shown in figure [ coolheat]a . as presented in this figure , the isobaric thermal instability criterion , @xmath98 is satisfied in the outer parts of the cloud , thus , these regions are thermally unstable while the inner part is stable as outlined by nejad - asghar ( 2007 ) .
the present sph code has the main features of the treesph class so that the nearest neighbors searching are calculated by means of this procedure .
we integrate the sph equations using the one order simple integration scheme .
the selection of time - step , @xmath99 , is of great importance .
there are several time - scales that can be defined locally in the system . for each particle
@xmath4 , we calculate the smallest of these time - scales using its smallest smoothing length , @xmath100 , i.e. @xmath101,\ ] ] where @xmath102 is the alfvn speed of ion fluid and @xmath103 is the courant number which in this paper is adopted equal to @xmath104 ( for numerical stability ) . the evolution were carried out to time @xmath105 $ ] so that the stability of the outer parts of the cloud may be revealed .
the values of the cooling and heating rates versus position ( and neutral density ) are shown in fig .
[ coolheat]b - d , at times @xmath106 $ ] , @xmath107 $ ] and @xmath108 $ ] , respectively .
in the preceding work of nejad - asghar & molteni ( 2007 ) , two - fluid sph implementation was used to check the pioneer works on behavior of the ambipolar diffusion in an isothermal self - gravitating layer . in this paper
, we include the sph equivalent of the energy equation that can be obtained from the first law of thermodynamics . here
, we incorporate the ambipolar drift heating in the net cooling function , and we investigate the thermal phases in the self - gravitating magnetized molecular layer .
the values of the cooling and heating functions are shown in figure [ coolheat ] . according to this figure ,
the isobaric thermal instability criterion is satisfied in the outer parts of the cloud , thus , these regions are thermally unstable while the inner part is stable .
the sph equations were integrated , and the evolution were carried out to time @xmath109 $ ] so that the stability of the outer parts of the cloud is revealed . it is obvious that the instability of the cloud at the outer parts , causes the formation two relative cool regions in that area . the rapid growth of thermal instability results a strong density imbalance between the cloud and the low - density surroundings .
this feature may be responsible for the planet formation in the outer parts of a collapsing molecular cloud and/or may also be conscientious for the formation of star forming dense cores in the clumps .
the mna would like to thank the research council of the damghan university of basic sciences . | we present an investigation on effect of the ion - neutral ( or ambipolar ) diffusion heating rate on thermal phases of a molecular cloud .
we use the modeling of ambipolar diffusion with two - fluid smoothed particle hydrodynamics , as discussed by nejad - asghar & molteni .
we take into account the ambipolar drift heating rate on the net cooling function of the molecular clouds , and we investigate the thermal phases in a self - gravitating magnetized one dimensional slab .
the results show that the isobaric thermal instability criterion is satisfied in the outer parts of the cloud , thus , these regions are thermally unstable while the inner part is stable
. this feature may be responsible for the planet formation in the outer parts of a collapsing molecular cloud and/or may also be relevant for the formation of star forming dense cores in the clumps . |
connes , gromov and moscovici @xcite developed and used the concepts of almost flat bundle , almost flat k - theory class and group quasi - representation as tools for proving the novikov conjecture for large classes of groups . for a compact manifold
@xmath4 it was shown in @xcite that the signature with coefficients in a ( sufficiently ) almost flat bundle is a homotopy invariant . moreover , the authors indicate that they have a reformulation of the notion of almost flatness to bundles with infinite dimensional fibers which allows them to show that if @xmath0 is a countable discrete group such that all classes of @xmath5 are almost flat ( up to torsion ) , then @xmath0 satisfies the novikov conjecture ( * ? ? ?
the problem of constructing nontrivial almost flat k - theory classes is interesting in itself .
suppose that the classifying space @xmath1 of a countable discrete group @xmath0 admits a realization as a finite simplicial complex .
using results of kasparov @xcite , yu @xcite and tu @xcite , the second - named author showed in @xcite that if @xmath0 is coarsely embeddable in a hilbert space and the full group c*-algebra @xmath3 is quasidiagonal , then all classes in @xmath5 are almost flat .
inspired by @xcite , in this paper we investigate the correspondence between between the almost flat classes in @xmath5 and the group homomorphisms @xmath6 that are implemented by pairs of discrete asymptotic homomorphisms @xmath7 , in the sense that @xmath8 for @xmath9 ; see definition [ def : pushforward ] .
it turns out that this correspondence is one - to - one ( modulo torsion ) if the full assembly map @xmath10 is bijective .
see theorem [ thm_cor : almost - flat - surj ] and its generalization with coefficients theorem [ thm : main - correspondence ] .
we take this opportunity to give a self - contained presentation of the correspondence ( and its continuity properties ) between almost flat bundles on a connected triangularizable compact space and the quasi - representations of its fundamental group ; see theorems [ thm : main - correspondence ] , [ thm : approx - inverse ] .
as far as we can tell , while this correspondence was more or less known to the experts , it has not been well documented in the literature .
we rely on work of phillips and stone @xcite .
these authors studied topological invariants associated with lattice gauge fields via a construction which associates an almost flat bundle to a lattice gauge field with controlled distortion and small modulus of continuity .
the following terminology is useful for further discussion of our results .
if @xmath11 is a sequence of natural numbers , we write @xmath12 . recall that a separable c*-algebra @xmath13 is mf
is it embeds as a c*-subalgebra of @xmath14 for some @xmath15 .
in other words , there are sufficiently many @xmath16-homomorphisms @xmath17 to capture the norm of the elements in @xmath13 , @xcite . by analogy ,
let us say that a separable c*-algebra @xmath13 is `` k - theoretically mf '' if there exist sufficiently many @xmath16-homomorphisms @xmath17 to capture the k - theory of @xmath13 in the following sense : for any homomorphism @xmath18 there exist @xmath15 and two @xmath16-homomorphisms @xmath19 such that @xmath20 for all @xmath21 .
here we identify @xmath22 with a subgroup of @xmath23 , and @xmath24 is the coset of element @xmath25 . with this terminology ,
theorem [ thm_cor : almost - flat - surj ] reads as follows : let @xmath0 be a discrete countable group whose classifying space @xmath1 is a finite simplicial complex .
if the full assembly map @xmath26 is bijective , then the following conditions are equivalent : 1 .
all elements of @xmath5 are almost flat modulo torsion ; 2 .
@xmath3 is k - theoretically mf . by a result of higson and kasparov @xcite ,
the assumptions of this theorem are satisfied by groups @xmath0 with the haagerup property and finite classifying spaces .
we note that a separable quasidiagonal c*-algebra that satisfies the uct is k - theoretically mf by ( * ? ? ?
we suspect that a similar result holds for mf algebras satisfying the uct .
let @xmath27 be a connected compact metric space that admits a finite triangulation .
this means that @xmath27 is the geometric realization of some connected finite simplicial complex @xmath28 , written @xmath29 .
let @xmath30 be the fundamental group of @xmath27 .
let @xmath13 be a unital @xmath31-algebra .
we establish a correspondence between quasi - representations of @xmath0 into @xmath32 and almost flat bundles over @xmath27 with fiber @xmath13 and structure group @xmath32 .
to be more accurate , this correspondence is between almost unitary quasi - representations and almost unitary principal bundles .
it is convenient to work in a purely combinatorial context .
following @xcite , we call the combinatorial version of an almost flat bundle an _ almost flat coordinate bundle_. see definition [ def : af - bundle ] .
we prove the equivalence of these concepts in proposition [ prop : correspondence_af ] .
for our combinatorial approach we begin with a finite simplicial complex @xmath28 endowed with some additional structure , as follows .
the vertices of @xmath28 are denoted by @xmath33 with possible indices .
the set of @xmath34-simplices of @xmath28 is denoted by @xmath35 we fix a maximal tree @xmath36 and a root vertex @xmath37 .
it will be convenient to view the fundamental group of the geometric realization of @xmath28 as the _ edge - path group _ @xmath38 of @xmath28 .
the groups @xmath39 and @xmath38 are isomorphic and we will simply write @xmath0 for either group .
once @xmath28 and @xmath40 are fixed so is the following standard presentation of @xmath0 in terms of the edges of @xmath28 ( see e.g. ( * ? ? ?
* sections 3.63.7 ) ) : @xmath0 is isomorphic to the group generated by the collection of all edges @xmath41 of @xmath28 subject to the following relations : * if @xmath41 is an edge of @xmath40 , then @xmath42 ; * if @xmath33 are vertices of a simplex of @xmath28 , then @xmath43 . we should point out that in the second relation @xmath44 , @xmath45 , and @xmath34 are not necessarily distinct ; consequently @xmath46 .
since @xmath47 belongs to @xmath40 one has @xmath48 as expected .
[ not : edge - path - grp ] let @xmath49 be the free group generated by the edge set of @xmath28 and let @xmath50 be the group epimorphism corresponding the presentation of @xmath0 just described .
write @xmath51 for the image under @xmath52 of the edge set of @xmath28 ; this is a symmetric generating set for @xmath0 .
write @xmath53 for the elements of @xmath51 .
let @xmath54 be the collection of all relators : @xmath55 choose a set - theoretic section @xmath56 of @xmath52 that takes the neutral element of @xmath0 to the neutral element of @xmath57 .
this section @xmath58 will remain fixed for the rest of the paper .
we introduce one last notation before the definition .
if @xmath13 is a unital c*-algebra and @xmath59 , set @xmath60 note that @xmath61 if @xmath62 .
[ def : q - rep ]
let @xmath13 be a unital c*-algebra . 1 .
let @xmath63 be finite and let @xmath64 .
a function @xmath65 is an _
@xmath66-representation _ of @xmath0 if 1 .
@xmath67 for all @xmath68 ; 2 .
@xmath69 for all @xmath70 ; 3 .
@xmath71 , where @xmath72 is the neutral element of @xmath73 2 .
define a pseudometric @xmath74 on the set of all bounded maps @xmath75 by @xmath76 we may sometimes refer to an @xmath66-representation as a `` quasi - representation '' without specifying @xmath77 or @xmath78 . in most of the paper
we will take @xmath79 .
once @xmath28 is fixed , so is a cover @xmath80 of @xmath81 , called the dual cover .
we recall its definition ( borrowing heavily from the appendix of @xcite ) .
let @xmath82 be a simplex of @xmath28 . for @xmath83 ,
the _ dual cell block _ @xmath84 , dual to @xmath44 in @xmath85 , is defined in terms of the barycentric coordinates @xmath86 by @xmath87 the _ dual cell _ @xmath88 ,
dual to the vertex @xmath44 , is the union of cell blocks dual to @xmath44 : @xmath89 the _ dual cover _
@xmath80 is the collection of all dual cells .
( see figure [ fig : dual - cells ] . )
we usually write @xmath90 for the intersection @xmath91 , @xmath92 for @xmath93 etc .
the barycenter of a simplex @xmath85 is denoted @xmath94 .
note that @xmath95 .
[ def : af - bundle ] recall that we have fixed a unital @xmath31-algebra @xmath13 . 1 .
an _ @xmath96-flat @xmath32-coordinate bundle on @xmath28 _ is a collection of continuous functions @xmath97 satisfying : 1 .
@xmath98 for all @xmath99 and all @xmath100 ; 2 .
@xmath101 for all @xmath99 and all @xmath100 ; 3 .
@xmath102 for all @xmath103 and all @xmath104 ; and 4 .
@xmath105 for all @xmath106 and all @xmath107 .
2 . define a metric @xmath74 on the set of all @xmath32-coordinate bundles on @xmath28 by @xmath108 we may sometimes refer to an @xmath96-flat coordinate bundle as an `` almost flat coordinate bundle '' without specifying @xmath96 .
we think of an almost flat coordinate bundle @xmath109 as a collection of transition functions defining a bundle over @xmath81 , with fiber @xmath13 , that has `` small '' curvature .
we substantiate this point of view in section [ sec:7 ] where we show that there are positive numbers @xmath110 that depend only on @xmath28 such that for any @xmath96-flat @xmath32-coordinate bundle @xmath111 on @xmath28 , with @xmath112 , there is a 1-ech cocycle @xmath113 that extends @xmath111 to prescribed open sets and @xmath114 is @xmath115-flat in the sense that @xmath116 for all @xmath117 . here
since the sets @xmath119 are open , the usual gluing construction based on @xmath114 defines a locally trivial ( almost flat ) bundle .
these objects are closely related to almost flat @xmath120-theory classes ; see section [ sec : an - appl ] for details .
we state the main results on this topic . the proofs
are given in subsequent sections .
[ thm : main - correspondence ] let @xmath28 be a finite connected simplicial complex with fundamental group @xmath0 .
there exist positive numbers @xmath121 , @xmath122 , and @xmath123 such that the following holds .
if @xmath13 is a unital @xmath31-algebra , then there are functions @xmath124
\xrightarrow{\hspace{20pt}}\\ \xleftarrow{\hspace{20pt}}\\[-5pt ] \beta \end{array } { \mathopen{}\mathclose\bgroup\originalleft}\ { \begin{array}{c } ( \mathcal{f}_\lambda , \delta_0)\text{-representations}\\ \text{of } \gamma \text { to } \operatorname{gl}(a ) \end{array } { \aftergroup\egroup\originalright}\}\ ] ] such that : 1 . if @xmath125 and @xmath109 is an @xmath96-flat @xmath32-coordinate bundle on @xmath28 , then @xmath126 is an @xmath127-representation of @xmath0 to @xmath32 ; and 2 .
if @xmath128 and @xmath65 is an @xmath129-representation , then @xmath130 is a @xmath131-flat @xmath32-coordinate bundle on @xmath28 .
moreover : 1 . if @xmath125 and @xmath109 and @xmath132 are @xmath96-flat @xmath32-coordinate bundles , then @xmath133 .
2 . if @xmath128 and @xmath134 are @xmath129-representations , then @xmath135 .
[ def : normalized - coord - bundle ] a @xmath32-coordinate bundle @xmath136 is _ normalized _ if @xmath137 for every edge @xmath138 of @xmath40 . by proposition [ prop : t - normalized ] , if a vector bundle over @xmath81 may be represented by an @xmath96-flat @xmath32-coordinate bundle , then it can be represented by a normalized @xmath139-flat @xmath32-coordinate bundle ( where @xmath140 depends only on @xmath28 ) .
[ thm : approx - inverse ] let @xmath28 be a finite connected simplicial complex with fundamental group @xmath0 .
there exist positive numbers @xmath141 and @xmath142 such that the following holds for any unital @xmath31-algebra @xmath13 . 1 .
if @xmath143 and @xmath109 is a normalized @xmath96-flat @xmath32-coordinate bundle on @xmath28 , then @xmath144 2 .
if @xmath145 and @xmath65 is an @xmath129-representation , then @xmath146
in this section we construct the map @xmath147 announced in section [ sec : results ] . it is a combinatorial version of a construction due to connes - gromov - moscovici @xcite involving parallel transport on a smooth manifold .
let @xmath148 be an @xmath96-flat coordinate bundle on @xmath28 .
we will define a quasi - representation @xmath149 with properties described in proposition [ prop : qrep - existence ] .
[ not : baryv ] recall that the barycenter of a 1-simplex @xmath150 is written @xmath151 .
for such a 1-simplex , let @xmath152 .
for a path @xmath153 of vertices in @xmath28 , let @xmath154 [ def : quasi - rep - constr ] define a group homomorphism @xmath155 as follows . if @xmath100 , let @xmath156 be the unique path along @xmath40 from @xmath37 to @xmath44 and @xmath157 be the unique path from @xmath37 to @xmath45 .
set @xmath158 finally , set @xmath159 where @xmath58 is the set theoretic section of @xmath160 that was fixed in notation [ not : edge - path - grp ] .
[ lem : u - epsilon - prod ] let @xmath161 and @xmath162 . if @xmath163 , @xmath164 and @xmath165 for all @xmath44 , then @xmath166 .
in particular , if @xmath167 , then @xmath168 . for @xmath169 ,
let @xmath170 be such that @xmath171 .
one checks that @xmath172 [ not :
ell - and - l ] for @xmath173 , let @xmath174 be the word length of @xmath175 with respect to the generating set @xmath176 .
we denote by @xmath177 the length ( number of edges ) of a longest path in @xmath28 that starts at the root @xmath37 and does not repeat any edge . [
lem : pi - tilde - u - epsilon ] if @xmath178 , then @xmath179 .
equation implies that @xmath180 is a product of at most @xmath181 elements of @xmath182 for any edge @xmath41 of @xmath28 .
it follows from lemma [ lem : u - epsilon - prod ] that @xmath183 .
another application of lemma [ lem : u - epsilon - prod ] ends the proof .
[ prop : qrep - existence ] there is a constant @xmath184 , depending only on @xmath28 , @xmath40 , @xmath37 , and @xmath58 , such that if @xmath185 is an @xmath96-flat @xmath32-coordinate bundle on @xmath28 , then @xmath126 is an @xmath186-representation of @xmath0 on @xmath32 .
write @xmath187 .
first we define a few constants so the proof will run more smoothly .
let @xmath188 .
if @xmath189 , then @xmath190 belongs to the kernel of @xmath50 , that is , to the normal subgroup generated by the set of relators @xmath191 .
( see notation [ not : edge - path - grp ] . ) for each pair @xmath192 choose and fix a representation @xmath193 with @xmath194 and @xmath195 .
let @xmath196 and let @xmath197 be the maximum of the lengths @xmath198 of all the elements @xmath199 that appear in equation for all pairs @xmath200 .
finally , let @xmath201 . for any @xmath202 , we show that @xmath203 .
using lemma [ lem : pi - tilde - u - epsilon ] we note that @xmath204 and hence @xmath205 now , @xmath206 implies that either @xmath207 is an edge of @xmath40 in which case @xmath208 or @xmath209 for some vertices @xmath210 , and hence @xmath211
let @xmath212 be the barycenter @xmath213 .
since @xmath111 , @xmath214 , and @xmath215 are @xmath96-constant with norm @xmath216 , we get that @xmath217 by lemma [ lem : pi - tilde - u - epsilon ] @xmath218 .
therefore @xmath219 because @xmath220 applying lemma [ lem : pi - tilde - u - epsilon ] again we get @xmath221 combined with , this proves that for all @xmath222 @xmath223 we must also prove that @xmath224 if @xmath225 .
this follows immediately from lemma [ lem : pi - tilde - u - epsilon ] since @xmath226 and @xmath227 .
let us single out an estimate from the proof for later use ( cf . ) .
[ lem : bound - on - r ] there exists @xmath228 , depending only on @xmath28 , @xmath40 , @xmath37 , and @xmath58 , such that the following holds .
suppose @xmath109 is an @xmath96-flat @xmath32-coordinate bundle on @xmath28 .
let @xmath229 be as in definition [ def : quasi - rep - constr ] . if @xmath230 , then @xmath231 and @xmath232 [ prop : t - normalized ] suppose @xmath136 is an @xmath96-flat @xmath32-coordinate bundle on @xmath28 . then there exist a constant @xmath140 , depending only on @xmath28 , and elements @xmath233 such that the coordinate bundle @xmath234 defined by @xmath235 is normalized and @xmath139-flat ; @xmath109 and @xmath236 yield isomorphic bundles on @xmath81 .
recall that @xmath177 is the length of the longest path in @xmath40 that starts at the root @xmath37 and has no backtracking .
we show that @xmath237 verifies the statement . for each vertex @xmath44 of @xmath28 ,
let @xmath238 be the unique path along @xmath40 from @xmath37 to @xmath44 and using notation as in [ not : baryv ] and [ def : quasi - rep - constr ] set @xmath239 .
thus @xmath240 then clearly @xmath241 for all @xmath242 .
moreover @xmath243 since @xmath244 by hypothesis .
if @xmath245 is a 1-simplex of @xmath28 and @xmath246 , then @xmath247 moreover , since @xmath248 by lemma [ lem : u - epsilon - prod ] and @xmath249 it follows immediately that @xmath250 .
therefore , @xmath236 is @xmath251-flat . by ( * ?
* theorem 3.2 ) @xmath109 and @xmath236 yield isomorphic bundles on @xmath81 .
in this section we describe the map @xmath252 announced in section [ sec : results ] .
the idea is to extend a quasi - representation of @xmath0 to a quasi - representation of the fundamental groupoid of @xmath28 and reinterpret the latter as a lattice gauge field .
this enables us to invoke a construction of phillips and stone @xcite that associates an almost flat coordinate bundle to a lattice gauge field with controlled distortion and small modulus of continuity . for the sake of completeness we give a full account of this construction .
let @xmath253 , with @xmath177 as in [ not : ell - and - l ] , and let an @xmath129-representation @xmath254 of @xmath0 be given .
we will define an almost flat coordinate bundle @xmath130 on @xmath28 .
the following notation will be used in the definition .
we fix a partial order * o * on the vertices of @xmath28 such that the set of vertices of any simplex of @xmath28 is a totally ordered set under * o*. one may always assume that such an order exists by passing to the first barycentric subdivision of @xmath28 : if @xmath255 and @xmath256 are the barycenters of simplices @xmath257 and @xmath258 of @xmath28 , define @xmath259 if @xmath257 is a face of @xmath258 ( cf .
@xcite ) . when we write @xmath260 it is implicit that the vertices of @xmath85 are written in increasing * o*-order .
[ not : mod - bary - coords ] following @xcite , we re - parametrize the dual cell blocks @xmath84 using `` modified barycentric coordinates '' @xmath261 .
these are defined in terms of the barycentric coordinates by @xmath262 . in these
coordinates @xmath84 is identified with the cube @xmath263 see figure [ fig : barycentric - vs - modified ] .
the construction of @xmath130 is inspired by ( and borrows heavily from ) the work of phillips and stone @xcite .
it is somewhat involved , but we outline the procedure in the following definition before going into the details .
[ def : cocycle ] let @xmath44 and @xmath45 be adjacent vertices of @xmath28 .
we will define @xmath264 by defining @xmath111 on all the dual cell blocks @xmath265 such that @xmath85 contains @xmath44 and @xmath45 . 1 .
let @xmath266 where @xmath267 is the perturbation of @xmath254 provided by proposition [ prop : perturb - qrep ] .
2 . suppose @xmath85 is a simplex in @xmath28 containing @xmath44 and @xmath45 and that @xmath268 .
write @xmath85 ( in increasing * o*-order ) as @xmath269 . + for an * o*-ordered subset of vertices @xmath270 , set @xmath271 ( where it is understood that if @xmath272 , then @xmath273 ) .
3 . define @xmath274 , using modified barycentric coordinates on @xmath265 ( see [ not : mod - bary - coords ] ) , as follows . for @xmath275 let @xmath276 where @xmath277 with @xmath278 if @xmath279 and @xmath280 if @xmath281 .
+ the sum above is over the subsets @xmath282 of @xmath283 that contain both @xmath44 and @xmath45 as above .
one can identify the subsets @xmath282 with ascending paths from @xmath44 to @xmath45 that are contained in @xmath85 .
let us note that @xmath284 for @xmath285 since @xmath286 .
we will see that the range of @xmath287 is actually contained in @xmath32 . if @xmath288 , let @xmath287 be the pointwise inverse of @xmath289 .
4 . corollary [ cor : cocycle - well - defined ] below shows that for each @xmath100 the collection of all @xmath287 above determines a function @xmath290 .
we define @xmath291 [ prop : almost - flat - existence ] there exist positive numbers @xmath292 and @xmath122 , depending only on @xmath28 , such that if @xmath128 and @xmath293 is an @xmath294-representation of @xmath0 , then @xmath130 is a @xmath295-flat @xmath32-coordinate bundle on @xmath28 .
the rest of the section is devoted to the proof of proposition [ prop : almost - flat - existence ] .
[ rem : bundle - constr ] the construction described in definition [ def : cocycle ] is an attempt to have @xmath111 be `` as constant as possible '' and equal to @xmath296 at the barycenter of @xmath138 ( cf .
2 ) ) . the cocycle condition one might hope for would force relations of the form @xmath297 , which do not necessarily hold since @xmath254 ( and therefore @xmath298 ) is only approximately multiplicative .
the definition of @xmath287 uses successive linear interpolation to account for this .
for example : 1 . if @xmath299 , then @xmath300 and @xmath301 , but @xmath302 2 . if @xmath303 , then @xmath304 , @xmath305 , @xmath306 @xmath307 @xmath308 and @xmath309 notice that in the modified barycentric coordinates @xmath310 the barycenter of @xmath311 is given by @xmath312 and @xmath313 for all @xmath314 .
therefore , @xmath315 as desired . to start the construction , we first perturb @xmath254 slightly so that we can deal with unitary elements instead of just invertible ones when convenient .
[ prop : perturb - qrep ] given an @xmath129-representation @xmath293 , @xmath316 , there exists a function @xmath317 such that 1 .
@xmath318 ; 2 .
@xmath319 for all @xmath320 ; 3 .
@xmath321 for all @xmath322 ; 4 .
@xmath323 for all @xmath324 with @xmath325 ; and 5 .
@xmath326 for all @xmath327 . the next lemma will be used in the proof .
[ lem : polar - decomp - estimate ] let @xmath328 be given by @xmath329 . if @xmath330 , @xmath316 , then @xmath331 .
observe that for @xmath332 @xmath333 using the neumann series .
assume @xmath330 .
then there is @xmath334 such that @xmath335 , so @xmath336 .
let @xmath337 .
then @xmath338 and thus @xmath339 .
now @xmath340 , so @xmath341 . therefore , by , @xmath342 .
finally , @xmath343 the idea is simple : define @xmath344 where @xmath345 is as in lemma [ lem : polar - decomp - estimate ] and @xmath346 is the function @xmath347 we check the required properties .
let @xmath225 .
first we prove that @xmath348 and @xmath349 . because @xmath67 , we can write @xmath350 for some @xmath334 and @xmath351 with @xmath352
. then @xmath353 by and hence @xmath354 .
therefore , @xmath355 it is just as plain to see that @xmath356 as claimed . using these bounds
we see that @xmath357 thus @xmath358 in particular @xmath359 .
items ( 1 ) and ( 2 ) in the statement of the proposition are immediate . for ( 3 )
observe that if @xmath360 , then @xmath361 .
it follows that @xmath362 . we deal with ( 5 ) . from and
lemma [ lem : polar - decomp - estimate ] we obtain @xmath363 . together with this
gives @xmath364 we are left with ( 4 ) .
suppose @xmath365 are such that @xmath366 .
then @xmath367 the next proposition allows us to use induction on the number of vertices of @xmath85 in the proofs that follow .
[ prop_restriction ] if @xmath368 are simplices of @xmath28 and @xmath369 are vertices of @xmath85 , then the restriction of @xmath370 to @xmath265 is equal to @xmath371 .
we may assume that @xmath372 is a simplex of @xmath28 that has @xmath85 as one of its faces and @xmath373 .
if @xmath374 , then @xmath375 and moreover @xmath376 precisely when @xmath377 .
let @xmath282 be a subset of @xmath378 that contains both @xmath44 and @xmath45 as above . if either @xmath379 or @xmath380 , then @xmath85 and @xmath381 have exactly the same set of increasing paths from @xmath44 to @xmath45 and hence @xmath382 for @xmath376 by definition [ def : cocycle ] .
( in fact , @xmath383 , again by definition 5.3 . ) suppose now that @xmath384 .
let @xmath385 be an increasing path in @xmath381 . if @xmath386 and @xmath387 then @xmath388 on the other hand if @xmath389 , then @xmath390 since @xmath391 is one of its factors .
the statement follows now immediately from by definition [ def : cocycle ] since @xmath392 [ prop_cocycle ] if @xmath384 are vertices of a simplex @xmath85 of @xmath28 , then @xmath393 for all @xmath394 .
let @xmath395 be an increasing path in @xmath396 .
if @xmath397 , then @xmath398 since @xmath399 is one of its factors .
on the other hand , if @xmath389 , say @xmath400 then letting @xmath401 we see that @xmath402 the statement now follows from definition [ def : cocycle ] since @xmath403 [ cor : cocycle - well - defined ] the family of functions @xmath404 yields a continous function @xmath405 such that @xmath406 for all @xmath407 .
proposition [ prop_restriction ] shows that if two simplices @xmath85 and @xmath408 contain @xmath409 , then @xmath410 on @xmath411 , so that @xmath111 is well - defined .
the cocycle condition follows from proposition [ prop_cocycle ] .
it remains to show that @xmath111 takes values in @xmath32 .
this will follow from the estimate @xmath412 that we now verify .
if @xmath413 and @xmath414 are as in definition [ def : cocycle ] , then show that @xmath415 by induction on @xmath416 .
this is trivial if @xmath417 , since in that case @xmath418 .
for the inductive step , we use the estimate @xmath419 proved in proposition [ prop : perturb - qrep](4 ) . because @xmath420 , the estimate follows since @xmath421 for @xmath285 .
let @xmath422 .
corollary [ cor : cocycle - well - defined ] all but implies proposition [ prop : almost - flat - existence ] . to complete the proof ,
we need to verify the almost flatness condition .
assume @xmath423 are vertices as in the proof of corollary [ cor : cocycle - well - defined ] .
we have seen that @xmath424 for all @xmath425 .
since @xmath296 is a unitary and since @xmath426 by hypothesis , we can apply to see that @xmath427 we conclude that @xmath428 is @xmath295-flat where @xmath429 , completing the proof of proposition [ prop : almost - flat - existence ] .
most of the work needed to prove theorems [ thm : main - correspondence ] and [ thm : approx - inverse ] was done in sections [ sec : bundle - to - qrep ] and [ sec : qrep - to - bundle ] .
what is left is basically bookkeeping related the various constants defined so far , but it is somewhat technical due to the nature of the definitions of @xmath147 and @xmath252 .
the definitions of @xmath147 and @xmath252 are given in sections [ sec : bundle - to - qrep ] and [ sec : qrep - to - bundle ] .
propositions [ prop : qrep - existence ] and [ prop : almost - flat - existence ] show the existence of @xmath430 and a constant @xmath431 satisfying parts ( 1 ) and ( 2 ) of the theorem .
we will actually set @xmath432 where @xmath120 is provided by lemma [ lem : bound - on - r ] .
we prove ( 3 ) .
let @xmath433 , @xmath434 .
let @xmath435 be such that @xmath436 as in definition [ def : quasi - rep - constr ] , let @xmath437 be the unique path along @xmath40 from @xmath37 to @xmath44 and @xmath438 be the unique path from @xmath37 to @xmath45 .
then @xmath439 from lemma [ lem : bound - on - r ] we get that @xmath440 , where @xmath228 depends only on @xmath28 , @xmath40 , @xmath37 , and @xmath58 .
the same bound holds with @xmath441 instead of @xmath442 . using this and the estimates @xmath443 , @xmath444 for @xmath445
, we see that @xmath446 since @xmath447 and @xmath448 we have @xmath449 thus @xmath450 . for part ( 4 ) , recall that @xmath451 . let @xmath452 and @xmath453 ( these are @xmath295-flat @xmath32-coordinate bundles by proposition [ prop : almost - flat - existence ] ) . recall that their definition ( see definition [ def : cocycle ] ) makes use of the maps @xmath298 and @xmath454 ( given by proposition [ prop : perturb - qrep ] ) respectively , and that @xmath455 etc . for @xmath456 and @xmath425
we estimate @xmath457 it follows that @xmath458 .
we prove ( 1 ) from the statement of the theorem first .
let @xmath459 , @xmath460 , and @xmath461 .
( @xmath121 is provided by theorem [ thm : main - correspondence ] , @xmath120 by lemma [ lem : bound - on - r ] and @xmath177 by notation [ not : ell - and - l ] ) .
let @xmath143 and suppose @xmath462 is an @xmath96-flat @xmath32-coordinate bundle on @xmath28 .
let @xmath463 and @xmath464 .
observe that @xmath254 is an @xmath465-representation and @xmath466 so that the construction of @xmath130 from section [ sec : qrep - to - bundle ] may be used .
we want to prove that @xmath467 recall the notation @xmath468 from [ not : baryv ] .
since @xmath109 is @xmath96-flat and @xmath132 is @xmath469-flat , it follows that @xmath470 let @xmath456 and set @xmath471 as in lemma [ lem : bound - on - r ] . applying the definition of @xmath442 ( see equation ) and the fact that @xmath109 is normalized ( definition [ def : normalized - coord - bundle ] )
, we obtain @xmath472 the definition of @xmath130 shows that @xmath473 and proposition [ prop : perturb - qrep ] implies @xmath474 .
thus @xmath475 lemma [ lem : bound - on - r ] guarantees that @xmath476 . in combination with and
this proves that @xmath477 we prove ( 2 ) from the statement of the theorem .
let @xmath478 and suppose @xmath65 is an @xmath129-representation
. let @xmath479 ( this is a @xmath480-flat @xmath32-coordinate bundle ) and let @xmath317 be given by proposition [ prop : perturb - qrep ] .
let also @xmath481 ( this is an @xmath482-representation of @xmath0 to @xmath32 ) .
we want to prove that @xmath483 suppose @xmath456 . as above , we may write @xmath484 where by lemma [ lem : bound - on - r ] , @xmath485 .
first notice that @xmath486 by the definition of @xmath487 ( definition [ def : cocycle ] ) .
let @xmath282 be the unique path along @xmath40 from @xmath37 to @xmath44 and @xmath488 be the unique path from @xmath37 to @xmath45 .
observe that @xmath109 is normalized since @xmath489 ; hence @xmath490 because @xmath282 and @xmath488 are paths in the tree @xmath40 .
then , by the definition of @xmath491 , @xmath492 therefore @xmath493
the goal of this section is to connect the notion of almost flat coordinate bundle from definition [ def : af - bundle ] , which is defined using simplicial structure and involves cocycles defined on closed sets , with the notion of almost flat bundle over a compact space from definition [ definition : to_alm_flat ] below .
almost flat bundles and @xmath120-theory classes appeared in the work gromov and lawson @xcite , of connes , gromov , and moscovici @xcite . in these references a vector bundle over a riemannian manifold is called @xmath96-flat if there is a metric - preserving connection with curvature of norm less than @xmath96 .
almost flat @xmath120-theory classes have been studied in different contexts in @xcite .
we adapt the definition to bundles over topological spaces as in @xcite and connect this with the version for simplicial complexes by proving proposition [ prop : bundle_extensions ] .
let @xmath27 be a compact space and let @xmath494 be a finite open cover of @xmath27 .
a ech 1-cocycle @xmath495 satisfies @xmath496 for all @xmath497 and @xmath102 for all @xmath498 .
[ definition : to_alm_flat ] let @xmath499 . 1 .
a ech 1-cocycle @xmath500 is _ @xmath96-flat _ , if 1 .
@xmath98 for all @xmath501 ; and 2 .
@xmath105 for all @xmath502 .
a principal @xmath32-bundle @xmath503 over @xmath27 is @xmath504-_flat _ if its isomorphism class is represented by an @xmath96-flat cocycle @xmath505 .
it is clear that if @xmath506 is an open cover that refines @xmath507 , then the restriction of @xmath508 to @xmath506 is also @xmath96-flat .
we now establish a result that connects the notion of @xmath96-flat ech 1-cocycles from definition [ definition : to_alm_flat ] with the notion of @xmath96-flat coordinate bundle in the simplicial sense as given in definition [ def : af - bundle ] .
suppose that @xmath29 is the geometric realization of a finite simplicial complex @xmath28 .
recall that @xmath27 has a ( closed ) cover @xmath80 given by dual cells @xmath88 ; see section [ sec : basic - def - not ] .
let @xmath74 be the canonical metric for the topology of @xmath27 obtained using barycentric coordinates .
fix a sufficiently small number @xmath161 such that if we set @xmath509 then for any finite intersection @xmath510 note that if @xmath511 is as in definition [ definition : to_alm_flat ] and the cover @xmath512 satisfies , then the restriction of @xmath511 to @xmath513 is an @xmath96-flat coordinate bundle .
proposition [ prop : correspondence_af ] below allows us reverse this operation .
[ prop : correspondence_af ] there are numbers @xmath514 and @xmath515 , depending only on @xmath28 , such that for any @xmath516 , any @xmath96-flat @xmath32-coordinate bundle @xmath517 on @xmath28 , and any @xmath161 satisfying , there is an @xmath518-flat cocycle @xmath519 that extends @xmath520 proposition [ prop : correspondence_af ] is a direct consequence of proposition [ prop : bundle_extensions ] below , whose content is of independent interest in connection with extension properties of principal bundles
. let @xmath521 be a closed subspace of a compact metric space @xmath27 and let @xmath522 be a closed cover of @xmath521
. for @xmath161 and @xmath523 let @xmath524 and set @xmath525 .
fix @xmath161 small enough such that for any finite intersection @xmath526 [ prop : bundle_extensions ] 1 . for any cocycle @xmath527 on @xmath521 there exist @xmath161 satisfying and a cocycle @xmath528 that extends @xmath111 , i.e. @xmath529 on @xmath530 .
2 . there exist @xmath531 and a universal constant @xmath532 that depends only on @xmath533 such that for any @xmath516 , any @xmath534-flat cocycle @xmath535 on @xmath521 , and any @xmath161 satisfying , there is an @xmath518-flat cocycle @xmath536 on @xmath537 which extends @xmath520 we begin with the proof of ( 1 ) and will explain subsequently how to adapt the argument to prove ( 2 ) as well .
we prove ( 1 ) by induction on the cardinality @xmath533 of the cover .
suppose that the statement is true for any integer @xmath538 .
let @xmath539 be given with @xmath540 .
set @xmath541 . by the inductive hypothesis ,
there exist @xmath161 satisfying and a cocycle @xmath528 , @xmath542 , which extends @xmath111 .
thus the following condition , labelled as @xmath543 , is satisfied : @xmath544 to pass from @xmath545 to @xmath533 we proceed again by induction on increasing @xmath546 , where @xmath547 is the set of those integers @xmath548 with the property that @xmath549 .
the inductive hypothesis that we make is that the functions @xmath550 extend to functions @xmath551 such that the following conditions ( depending on @xmath34 ) are satisfied : @xmath552 if @xmath547 reduces to @xmath553 , then we simply define @xmath554 and we are done .
assume @xmath547 contains more than one element .
let @xmath555 be the smallest element of @xmath547 ( so @xmath556 ) . to construct @xmath557 we first define an extension @xmath558 of @xmath559 on suitable closed subsets of @xmath560 as follows : @xmath561 let us observe that @xmath562 is well - defined since if @xmath563 , @xmath564 then @xmath565 on @xmath566 if and only if @xmath567 on the same set .
in view of condition @xmath543 this reduces to the equality @xmath568 on @xmath569 , which holds true since @xmath114 extends @xmath111 . by tietze s theorem
we can now extend the function @xmath562 defined by ( @xmath570 ) to a continuous function @xmath571 .
since @xmath32 is open in @xmath13 we will have that @xmath572 for all @xmath573 provided that @xmath574 is sufficiently small .
we need to very that @xmath557 satisfies @xmath575 and @xmath576 .
condition @xmath575 amounts to @xmath577 on @xmath578 which holds since @xmath579 .
condition @xmath576 reduces to @xmath580 on @xmath581 for @xmath582 with @xmath583 .
this holds true in view of ( @xmath570 ) and so the base case for the induction is complete .
fix @xmath546 , @xmath584 and suppose now that we have constructed @xmath585 for all @xmath586 with @xmath587 such that the conditions ( @xmath588 ) and ( @xmath589 ) are satisfied .
let @xmath590 be the successor of @xmath34 in @xmath547 .
we may assume that @xmath556 for otherwise there is nothing to prove .
we construct a map @xmath557 on @xmath578 that satisfies the corresponding conditions @xmath575 and @xmath576 as follows .
the first step is to define an extension @xmath562 of @xmath591 on suitable closed subsets of @xmath592 as follows : @xmath593 we need to observe that the conditions @xmath594 and @xmath595 are compatible so that @xmath596 is well - defined and continuous .
there are three cases to verify .
first we check that @xmath597 on @xmath598 for @xmath599 , @xmath600 .
this is a consequence of conditions @xmath543 and @xmath601 .
second , we verify that @xmath565 on @xmath602 for @xmath603 , @xmath604 .
note that this holds if and only if @xmath567 on the same set . in view of condition @xmath543
this reduces to the equality @xmath568 on @xmath569 , which holds true since @xmath114 extends @xmath111 .
finally we need to verify that @xmath605 on @xmath606 for @xmath607 , @xmath608 .
by @xmath543 , this equality holds if and only if @xmath609 on @xmath610 .
the latter equality holds due to condition @xmath611 which is satisfied by the inductive hypothesis . by tietze s theorem
we can now extend the function @xmath562 defined by ( @xmath612 ) and ( @xmath613 ) to a continuous function @xmath614 .
since @xmath32 is open in @xmath13 we will have that @xmath572 for all @xmath615 provided that @xmath574 is sufficiently small .
it is clear that the functions @xmath616 satisfy the conditions @xmath617 as a consequence of ( @xmath612 ) , ( @xmath588 ) , ( @xmath613 ) and ( @xmath589 ) .
this completes the inductive step from @xmath34 to @xmath555 and hence from @xmath545 to @xmath533 . during this step
we had to pass to a possibly smaller @xmath574 but this does not affect the conclusion .
the proof follows the pattern of the proof of ( 1 ) with one important modification .
namely we use the following strengthened version of tietze s theorem due to dugunji @xcite .
let @xmath27 be an arbitrary metric space , @xmath521 a closed subset of @xmath27 , @xmath13 a locally convex linear space and @xmath618 a continuous map .
then there exists an extension @xmath619 of @xmath620 such that @xmath621 is contained in the convex hull of @xmath622 .
fix a point @xmath623 in each nonempty intersection @xmath530 and set @xmath624 .
since the cocycle is @xmath534-flat , we have that @xmath625 .
let us define positive numbers @xmath626 for @xmath627 as follows .
if @xmath628 , then @xmath629 . if @xmath369 , @xmath626 is defined by the following recurrence formula .
set @xmath630 and @xmath631 . if @xmath632 then we define @xmath633 with the convention that @xmath634 .
we only need to consider the maps @xmath557 with @xmath635 and @xmath636 .
we proceed as in proof of ( 1 ) by induction on @xmath533 and @xmath546 with the additional provision that * @xmath637 , for all @xmath638 and for all @xmath639 with @xmath640 , @xmath586 and @xmath641 . the basic idea of the proof is to observe that it follows from the equations ( @xmath570 ) , ( @xmath612 ) and ( @xmath613 ) that @xmath562 is close to @xmath642 ( if @xmath643 ) or @xmath644 ( if @xmath645 ) both of which are near @xmath646 and hence @xmath596 is close to @xmath647 .
it will follow that the image of @xmath558 is contained in a ball @xmath648 where @xmath649 is a universal constant computed recursively from previously determined @xmath626 .
therefore we can invoke the strengthened version of tietze s theorem of @xcite to extend @xmath596 to a continuous map @xmath650 with values in convex open ball @xmath648 .
fix @xmath546 , @xmath584 . by the inductive hypothesis ,
suppose that we have constructed @xmath114 and they satisfy ( @xmath651 ) .
we need to consider two cases .
the first is the case when @xmath652 .
letting @xmath653 and @xmath654 , then from condition ( @xmath570 ) , for each @xmath655 with @xmath656 , @xmath657 @xmath658 let @xmath555 be the successor of @xmath34 in @xmath547 .
we may assume that @xmath556 otherwise we are done .
if @xmath659 with @xmath656 , @xmath660 then using ( @xmath613 ) it follows just as above that @xmath661 on the other hand , if @xmath662 with @xmath663 , @xmath664 then using ( @xmath612 ) we have @xmath665 in view of this estimates we can extend @xmath558 to @xmath666 using the strengthened version of tietze s theorem so that @xmath667 .
it follows that @xmath668 for all @xmath669 .
this completes the proof .
one of the motivations for this paper is the detection of nontrivial k - theory elements of a group c*-algebra , via lifting of homomorphisms @xmath670 to quasi - representations @xmath671 .
suppose that the full assembly map is a bijection for a discrete group @xmath0 .
roughly speaking , our main result states that the quasi - representations @xmath672 which induce interesting partial maps on k - theory are as abundant as the non - trivial almost flat k - theory classes of the classifying space @xmath1 .
more generally , for a c*-algebra @xmath673 we consider the connection between almost flat k - theory classes in @xmath674 and quasi - representations @xmath675 that implement a given homomorphism @xmath676 .
let @xmath13 be a unital c*-algebra .
a quasi - representation @xmath677 extends to a unital linear contraction @xmath678 in the obvious way .
we like to think of @xmath254 as `` inducing '' a partially defined map @xmath679 ( cf .
we briefly recall the definition of @xmath680 . in the definition
we write @xmath681 for the function @xmath682 where @xmath683 .
[ def : pushforward ] let @xmath684 , @xmath673 be banach algebras and let @xmath685 be a unital contractive map .
let @xmath686 be an idempotent and let @xmath687 .
define @xmath688 \in k_0(b ) $ ] whenever @xmath689 . in a similar manner
, one defines the pushforward @xmath690 of an invertible element @xmath691 as the class of @xmath692,$ ] under the assumption that @xmath254 is ( sufficiently ) approximately multiplicative on a suitable finite subset of @xmath684 that depends on @xmath693 . a discrete asymptotic homomorphism from an involutive banach algebra @xmath684 to c*-algebras @xmath698 consists of a sequence @xmath699 of maps such that @xmath700 for all @xmath701 and @xmath702 .
the sequence @xmath703 induces a @xmath16-homomorphism @xmath704 if each @xmath698 is a matrix algebra over some fixed c*-algebra @xmath673 , then this further induces a group homomorphism @xmath705 a discrete asymptotic homomorphism gives a canonical way to push forward an element @xmath706 to a sequence @xmath707 of elements of @xmath708 , which is well - defined up to tail equivalence : two sequences are tail equivalent , written @xmath709 , if there is @xmath416 such that @xmath710 for all @xmath711 .
note that one can adapt definition [ def : pushforward ] to maps which are approximately contractive ( in addition to being approximately multiplicative ) . 1 .
a sequence of @xmath712-representations @xmath713 , with @xmath714 as @xmath715 , induces a discrete asymptotic homomorphism ( still written @xmath716 ) from the involutive banach algebra @xmath717 to the c*-algebras @xmath698 .
2 . a discrete asymptotic homomorphism @xmath718 as above induces a @xmath16-homomorphism @xmath719 and hence a @xmath16-homomorphism @xmath720 such as the following diagram is commutative .
@xmath721 ^ -{\pi_\infty } \ar[d]_{j } & b_\infty \\ c^*(\gamma)\ar[ur]_-{\bar{\pi}_\infty } & } \ ] ] ( where @xmath45 is the canonical map ) .
3 . let @xmath722 be a discrete asymptotic homomorphism given by some set - theoretic lift of @xmath723 .
if @xmath724 , then we have the following tail equivalence : @xmath725 the results of @xcite will allow us to relate the push - forward of elements in the image of the full baum - connes assembly map with the almost flat bundles we have constructed .
the relationshp involves the mishchenko line bundle and its push - forward , which we now discuss .
let @xmath726 be the universal cover of @xmath29 .
consider the dual cover @xmath727 of @xmath27 and the associated open cover @xmath728 where @xmath729 .
the mishchenko line bundle is the bundle @xmath730 , obtained from @xmath731 by passing to the quotient with respect to the diagonal action of @xmath0 .
it is isomorphic to the bundle @xmath503 obtained from the disjoint union @xmath732 by identifying @xmath733 with @xmath734 whenever @xmath735 , where @xmath736 are as in notation [ not : edge - path - grp ] ; see for example ( * ? ? ?
* lemma 3.3 ) .
let @xmath737 be a partition of unity subordinate to @xmath512 .
it follows that the mishchenko line bundle corresponds to the class of the projection @xmath738,\ ] ] where @xmath739 are the canonical matrix units of @xmath740 and @xmath741 is the number of vertices in @xmath28 .
we have inclusions of rings @xmath742\subset \ell^1(\gamma)\subset c^*(\gamma)$ ] . the class of the idempotent @xmath72 in @xmath743 or @xmath744 is denoted by @xmath555 . from definition [ def : cocycle ]
we get the coordinate bundle @xmath130 associated with @xmath254 . applying proposition [ prop : bundle_extensions ] to @xmath130
we obtain an almost flat cocycle @xmath748 .
let @xmath749 be the bundle constructed from the disjoint union @xmath750 by identifying @xmath733 with @xmath751 for @xmath752 in @xmath753 .
let @xmath757 and let @xmath746 be an @xmath745-representation . because the bundle @xmath749 is represented by the idempotent @xmath758 , it follows that @xmath759 now , by the construction of @xmath511 , there is a constant @xmath760 depending only on @xmath28 such that @xmath761 .
therefore , @xmath762 if @xmath123 is chosen to be sufficiently small .
let @xmath27 be a compact connected space and let @xmath673 be a unital c*-algebra .
we consider locally trivial bundles @xmath503 over @xmath27 with fiber finitely generated projective hilbert - modules @xmath763 over @xmath673 and structure group @xmath32 , where @xmath764 , the c*-algebra of @xmath673-linear adjointable endomorphisms of @xmath763 .
the k - theory group @xmath765 consists of formal differences of isomorphism classes of such bundles .
let @xmath507 be a finite open cover of @xmath27 .
a bundle @xmath503 as above is @xmath504-flat if it admits an @xmath504-flat associated @xmath32-principal bundle in the sense of definition [ definition : to_alm_flat](2 ) .
an element @xmath766 is _ almost flat _ if there is a finite open cover @xmath507 of @xmath27 such that for every @xmath757 there are @xmath504-flat bundles @xmath767 over @xmath27 such that @xmath768 - [ e^-]$ ] .
we say that @xmath769 is _ almost flat modulo torsion _ if there is a torsion element @xmath770 such that @xmath771 is almost flat . by the uct given in ( *
* lemma 3.4 ) , the kasparov product @xmath772 induces an exact sequence @xmath773 if @xmath708 is finitely generated and torsion free , then the torsion subgroup of @xmath674 coincides with the image of @xmath774 .
[ thm : almost - flat - surj ] let @xmath0 be a discrete countable group whose classifying space @xmath1 is a finite simplicial complex and let @xmath673 be a unital c*-algebra . consider the following conditions : 1 .
for any @xmath775 there is @xmath776 such that @xmath771 is almost flat .
2 . for any group homomorphism @xmath777 there exist discrete asymptotic homomorphisms @xmath778 such that @xmath779 for every @xmath780 in the image of the full assembly map @xmath781 \(1 ) @xmath782 ( 2 ) .
let @xmath784 be given
. then @xmath785 . by the uct
, there is @xmath775 such that @xmath786 note that if @xmath787 , then @xmath788 .
thus without any loss of generality we may assume that @xmath752 is almost flat . therefore there exist a finite open cover @xmath507 of @xmath1 , a decreasing sequence @xmath789 of positive numbers converging to @xmath790 and two sequences @xmath791 of bundles over @xmath1 such that @xmath792 are @xmath793-flat and satisfies @xmath794 - [ e_n^-]$ ] for all @xmath533 . by passing to barycentric subdivisions of the simplicial structure @xmath28 of @xmath1
we may assume that the dual cover @xmath795 refines the open cover @xmath507 .
by proposition [ prop : t - normalized ] we may arrange that the coordinate bundles underlying the @xmath791 are normalized .
write @xmath796 for the fibers of @xmath792 ; these are finitely generated projective hilbert @xmath673-modules and therefore embed as direct summands of some @xmath797 .
this gives full - corner embeddings @xmath798 .
using proposition [ prop : qrep - existence ] and proposition [ prop : perturb - qrep ] we associate with @xmath792 quasi - representations @xmath799 such that @xmath800 for all @xmath801 and @xmath802 for all @xmath803 . the sequences @xmath804 induce morphisms of groups @xmath805 and hence @xmath16-homomorphisms @xmath806 and @xmath807 where @xmath808 .
let @xmath809 be a set - theoretic lifting of @xmath810 .
write @xmath811 .
for a sufficiently multiplicative quasi - representation @xmath812 and a sufficiently small @xmath59 , we will denote by @xmath749 the corresponding almost flat bundle constructed using the cocycle @xmath130 constructed in proposition [ prop : almost - flat - existence ] ( see notation [ not : e - pi ] ) . for @xmath533 sufficiently large we have that @xmath813 = [ e_n^\pm]$ ] .
this follows from theorem [ thm : approx - inverse ] and proposition [ prop : correspondence_af ] since bundles whose cocycles are sufficiently close to each other are isomorphic .
we have seen earlier that one can represent @xmath555 by a projection @xmath72 in matrices over @xmath818 $ ] .
so long as @xmath533 is sufficiently large , proposition [ prop : push - mish - flat ] guarantees that @xmath813 $ ] equals @xmath819 , the push - forward of @xmath72 by @xmath820 in @xmath821 .
the latter isomorphism is induced by the full - corner embeddings @xmath822 .
it follows that @xmath823 , z \rangle - \langle [ e_n^- ] , z \rangle = \langle \ell_{\pi_{n}^{+ } } , z \rangle - \langle \ell_{\pi_{n}^{- } } , z \rangle\ ] ] for all @xmath824 .
let @xmath825 be lafforgue s @xmath826-version of the assembly map .
it is known that @xmath827 where @xmath828 is the canonical map @xcite . by (
* theorem 3.2 and corollary 3.5 ) we have that @xmath829 for each @xmath830 so long as @xmath533 is sufficiently large .
( for @xmath831 we interpret @xmath832 as @xmath833 where @xmath834 are projections with @xmath835-[q_z]$ ] .
there is a similar interpretation of @xmath836 for @xmath837 obtained by replacing idempotents by invertible elements . ) therefore , @xmath838 for all @xmath824 . from remark [ rem : disc - asymp - obs](3 )
we deduce that @xmath839 @xmath840 for all @xmath824 .
the discrete asymptotic homomorphisms @xmath841 have the desired properties .
\(2 ) @xmath782 ( 1 ) .
let us assume now that @xmath708 is finitely generated and that @xmath783 is split injective .
let @xmath842 be given .
we will find an almost flat element @xmath843 such that @xmath844 . since @xmath783 is split - injective by hypothesis ( i.e. the image of @xmath783 is a direct summand of @xmath845 it follows from the exactness of the sequence that there is a homomorphism @xmath846 such that @xmath847 for all @xmath824 . by the assumptions in ( 2 )
there are two discrete asymptotic homomorphisms @xmath848 such that @xmath849 for all @xmath850 . by a standard perturbation argument
we may assume that @xmath851 for all @xmath533 and @xmath852 . invoking ( * ? ? ?
3.2 and cor . 3.5 ) we obtain that @xmath853 by proposition [ prop : push - mish - flat ] , we have that @xmath854 $ ] where the bundles @xmath855 are @xmath793-flat and @xmath856 .
if we set @xmath857-[e_{\pi_n^-}]$ ] , it follows from that for any @xmath824 , there is @xmath858 such that @xmath859 for @xmath860 . since @xmath861 is finitely generated there exists @xmath862 such that @xmath863 for all @xmath864 .
since @xmath708 is finitely generated , the group @xmath865 is finite .
therefore after passing to a subsequence of @xmath866 we may arrange that the sequence @xmath867 is constant and so there is @xmath868 such that @xmath869 for all @xmath533 .
it follows that @xmath870 is is almost flat and @xmath871 .
if @xmath1 is a finite complex and the group @xmath708 is finitely generated and torsion - free , it follows that the group @xmath774 is finite and by it coincides with the torsion subgroup of @xmath674 . by taking @xmath872 in theorem [ thm : almost - flat - surj ]
we obtain the following .
[ thm_cor : almost - flat - surj ] let @xmath0 be a discrete countable group whose classifying space @xmath1 is a finite simplicial complex .
if the full assembly map is bijective , then the following conditions are equivalent gromov indicates in @xcite how one constructs nontrivial almost flat @xmath120-theory classes for residually finite groups @xmath0 that are fundamental groups of even dimensional non - positively curved compact manifolds .
mikhael gromov , _ geometric reflections on the novikov conjecture _ , novikov conjectures , index theorems and rigidity , vol . 1 ( oberwolfach , 1993 ) , london math
lecture note ser .
226 , cambridge univ .
press , cambridge , 1995 , pp .
mr 1388301 to3em , _ positive curvature , macroscopic dimension , spectral gaps and higher signatures _ , functional analysis on the eve of the 21st century , vol .
ii ( new brunswick , nj , 1993 ) , progr .
132 , birkhuser boston , boston , ma , 1996 , pp .
mr 1389019 mikhael gromov and h. blaine lawson , jr .
, _ positive scalar curvature and the dirac operator on complete riemannian manifolds _ , inst .
hautes tudes sci .
publ . math .
( 1983 ) , no . 58 , 83196 ( 1984 ) .
mr 720933 vladimir m. manuilov and aleksandr s. mishchenko , _ almost , asymptotic and fredholm representations of discrete groups _ , acta appl
* 68 * ( 2001 ) , no . 1 - 3 , 159210 , noncommutative geometry and operator @xmath120-theory .
mr 1865957 henri moscovici , _ cyclic cohomology and invariants of multiply connected manifolds _ , proceedings of the international congress of mathematicians , vol .
i , ii ( kyoto , 1990 ) ( tokyo ) , math . soc .
japan , 1991 , pp .
mr 1159254 georges skandalis , _ approche de la conjecture de novikov par la cohomologie cyclique ( daprs a. connes , m. gromov et h. moscovici ) _ , astrisque ( 1991 ) , no . 201 - 203 , exp .
739 , 299320 ( 1992 ) , sminaire bourbaki , vol .
mr 1157846 jean - louis tu , _ the gamma element for groups which admit a uniform embedding into hilbert space _ ,
recent advances in operator theory , operator algebras , and their applications , oper .
theory adv .
153 , birkhuser , basel , 2005 , pp . | we give a rigorous account and prove continuity properties for the correspondence between almost flat bundles on a triangularizable compact connected space and the quasi - representations of its fundamental group . for a discrete countable group @xmath0 with finite classifying space @xmath1 , we study a correspondence between between almost flat k - theory classes on @xmath1 and group homomorphism @xmath2 that are implemented by pairs of discrete asymptotic homomorphisms from @xmath3 to matrix algebras . |
non - critical string theory @xcite provides a simple and tractable toy model to understand various aspects of string theory .
this string theory is solvable not only in the world sheet cft approach @xcite , but also in the non - perturbative formulation based on matrix models @xcite .
the solvability of the theory has uncovered many non - perturbative behaviors of string theory beyond string - coupling perturbative expansions @xcite which played an important role in the discovery of d - branes @xcite .
this theory has also provided a simple toy model of gauge / string correspondence @xcite and led to a new correspondence between two - cut critical points @xcite of the matrix models and type 0 superstrings @xcite . along this line ,
multi - cut extension of the correspondence has also been proposed in @xcite : the infinite sequence of correspondences between @xmath1-cut matrix models @xcite and @xmath1-fractional superstring theory @xcite .- th fractional superstring theory is called @xmath1-fractional superstring theory . therefore , on the worldsheets of the @xmath1-fractional superstrings , there are @xmath0 zamolodchikov - fateev parafermions @xcite with bosonic fields as their fractional superpartners . ] existence of new string - theory dual descriptions in non - critical string theory is especially attractive because this would provide a new experimental laboratory to extract non - trivial information about string theory .
therefore , it would be interesting to ask _ what kind of phenomena appear beyond this infinite variety of the correspondence?_. in this paper , as a first step toward this study , we quantitatively analyze the multi - cut two - matrix models and their geometries in the fractional - superstring critical points .
as is reviewed in section [ fscriticalpointssection ] , the fractional - superstring critical points are special critical points in the @xmath1-cut two - matrix models .
the multi - cut matrix models are generally shown @xcite to be controlled by multi - component kp hierarchy @xcite .
the fractional - superstring critical points are characterized by the following form of the @xmath1-component kp lax pairs @xmath9 @xcite : @xmath10 with the two different choices of the leading matrix @xmath11 @xcite : @xmath12 these choices of the leading matrices generally break the intrinsic @xmath0 symmetry of the multi - cut two - matrix models ( @xmath13 and @xmath14 are the two matrices ) , symmetry . ] @xmath15 and the system requires rather non - trivial analysis as compared with the @xmath0 symmetric critical points @xcite . despite of this fact , we show that these systems turn out to be still solvable even in the presence of _ an arbitrary number of cuts_. the actual quantities investigated here are _ macroscopic loop amplitudes _
@xcite , and its scaling parameter @xmath16 , although they are distinguished in the main text . ] @xmath17,\end{aligned}\ ] ] and they are known to play important roles in extracting various information about the matrix models : eigenvalue distributions @xcite , generating functions of local ( on - shell ) operators on worldsheets , and also effective potentials of a single ( pair of ) matrix - model eigenvalues @xmath18 @xcite . furthermore , since these amplitudes directly correspond to boundary states of fzzt branes @xcite , it is through the study of these amplitudes that we can perform a direct comparison with the amplitudes in the liouville theory @xcite .
since the worldsheet cft calculation of fractional super liouville field theory @xcite is far from completion , the study of macroscopic loop amplitudes is important not only for an understanding of the actual dynamics of this system , but also it could provide an important clue to the analysis of this not - yet solved liouville theory . in this paper , the macroscopic loop amplitudes are studied within the daul - kazakov - kostov prescription @xcite and its multi - cut extension @xcite . in the dkk prescription ( reviewed in the begining of section [ sectioncoshsinh ] ) , the macroscopic loop amplitude @xmath19 is identified with the eigenvalues of the lax operator @xmath20 .
it turns out that there are generally two kinds of solutions in the fractional - superstring critical points : one solution is given by hyperbolic cosine and the other is by hyperbolic sine functions : @xmath21 in the background with the bulk cosmological constant @xmath22 .
the integer @xmath23 is the label of eigenvalues of the lax operator @xmath20 .
the sinh solution only appear when @xmath1 is even , and the cosh solutions appear generally .
therefore , the odd @xmath1 cases are natural extension of the bosonic string amplitudes @xcite ; the even @xmath1 cases are natural extension of type 0 superstring amplitudes @xcite .
interestingly , these general formulae naturally include bosonic @xmath24 and superstring @xmath25 cases , and the properties of algebraic curves and their singular points constitute natural generalizations of these special cases .
this continuous extrapolation makes our current study distinct from the @xmath0 symmetric critical points @xcite .
the phase shift @xmath26 in the solutions in eq .
( [ cssolintroduction ] ) is given by some proper rational number , and only this part of the amplitudes depends on the models we consider : the number of cuts , @xmath1 , and the hermiticity of the two - matrix models , i.e. the leading gamma matrices ( [ gammaintroduction ] ) : @xmath27 with @xmath28 .
the first cases are realized in @xmath29-rotated ( complex ) potentials and the second are in real potentials .
these two kinds of critical points are different when @xmath30 .
these are the quantitative results obtained in this paper which enable us to extract the algebraic geometry realized in the weak - coupling regime of the multi - cut two - matrix models / fractional superstring theories .
in addition to these quantitative results , we also discuss several implications based on our analysis as follows : from the first sight , it might seem strange that the single macroscopic loop amplitude @xmath19 or resolvent operator corresponds to the @xmath1 different amplitudes @xmath31 ( @xmath23 ) which are realized as eigenvalues of the lax operator @xmath20 .
however we argue in section [ multicutgeometrystokessection ] that they play an important role to make up the _ @xmath1-cut geometry _ anticipated from the definition of the @xmath1-cut two - matrix models .
the idea is based on the study about _ stokes phenomena _ in the baker - akhiezer functions @xcite .
that is , although each amplitude @xmath31 has maximally _ two cuts _ on their branch , the resolvent operator @xmath19 can have @xmath1 semi - infinite cuts by being patched with @xmath1 different amplitudes @xmath31 in _ different stokes sectors _ of the multi - cut baker - akhiezer function .
the patching rule is also proposed in section [ multicutgeometrystokessection ] .
the reduction of the number of cuts on each branch is also related to the highly reducibility of our resulting algebraic curve of the macroscopic loop amplitude .
that is , we show that the algebraic curve is factorized into @xmath32 irreducible curves : is written as @xmath33 , e.g. @xmath34 and @xmath35 . ]
@xmath36 each irreducible curve corresponds to a pair of eigenvalues of @xmath20 , @xmath37 as a general belief in matrix models , if the algebraic curve is factorized into irreducible pieces , each irreducible sector does not communicate with other sectors in all order perturbation theory . ] therefore , these considerations indicate that in the spacetime @xmath38 of @xmath2 minimal @xmath1-fractional superstring theory , which has been given by @xcite @xcite . ] or in the weak - coupling _ landscape _ of this fractional superstring theory , there are @xmath32 perturbatively isolated vacua labeled by @xmath39 .
therefore , this non - perturbative string theory includes various perturbative string theories like its super - selection sectors , although these sectors can interact with each other in the non - perturbative formulation . a typical landscape of minimal @xmath40-fractional superstring theory
is shown in fig .
[ 12cuts ] .
each asymptotic regime around the arrow ( denoted as @xmath41 ) is a perturbatively isolated sector , which corresponds to the irreducible algebraic curve @xmath42 .
this string theory includes six perturbative string theories which are also realized in @xmath1-fractional superstring theory of @xmath43 . in particular , @xmath44 are identified with bosonic strings , and @xmath45 are with ( two - cut phase of ) type 0 superstring theory . around the center
is the strong coupling regime . ] furthermore , if one takes @xmath46 , the theory becomes @xmath2 minimal @xmath3-fractional superstring theory and this theory includes all the perturbative vacua realized in @xmath2 minimal @xmath1-fractional superstring theory ( with arbitrary number of @xmath47 , and @xmath2 is fixed ) . in this sense
, @xmath3-fractional superstring theory can be viewed as a mother theory of perturbative fractional superstring theory . interestingly in this limit ,
the labeling of perturbative vacua @xmath48 or @xmath26 forms a continuum @xmath49 _ angular _ direction which can only be observed by non - perturbative correction of string theory . since the @xmath0 charge conjugation of d - branes
is obtained by a translation of this angular coordinate @xmath50 , the fractional superstring theory can be understood as a kaluza - klein reduction along this non - perturbative direction @xmath51 in a three - dimensional gravity theory .
this situation resembles the _ non - critical m theory _ proposed by p. hoava and c. a. keeler @xcite as an extension of @xmath52 type 0a/0b superstring theories . a possible connection to non - critical m theory is also disccused in section [ secconclusion ] .
organization of this paper is as follows : in section [ macroscopicloopamplitudessection ] , after the basics of the fractional - superstring critical points are reviewed in section [ fscriticalpointssection ] , macroscopic loop amplitudes in these critical points are investigated in section [ sectioncoshsinh ] .
section [ complexpotentialsection ] is for the @xmath29-rotated - potential critical points , and section [ realpotentialmodels ] is for the real - potential critical points and the comparison between @xmath29-rotated - potential critical points and real - potential critical points is shown . in section [ matrixrealizationsection ] , the actual expression of the lax operators for the solutions is shown . in section [ algebraicstructuressection ] , the algebraic properties are studied .
the algebraic equations are derived in section [ algebraicequationssection ] , branch points are studied in section [ branchcurvessection ] and singular points are studied in section [ algsingularpointssection ] . in section [ multicutgeometrystokessection ] , the multi - cut geometry of macroscopic loop amplitudes is discussed . in section [ sectionfzztandzzamp ] , the fzzt- and zz - brane amplitudes are shown .
section [ secconclusion ] is devoted to discussion , which includes a possible connection to non - critical m theory .
here we review some basic facts about the multi - cut two - matrix models in the fractional - superstring critical points .
one can find more details and references on fractional superstring theory in @xcite and on the multi - cut matrix models in @xcite .
the @xmath1-cut two - matrix models @xmath53 are an extension of the usual two - matrix models @xcite , and defined by integrating over the set of @xmath54 normal matrices @xmath55 whose eigenvalues are along the @xmath0 symmetric radial lines @xmath56 : @xmath57 the two - matrix potentials @xmath58 are generally chosen to be polynomials and there are two possible definitions of hermiticity : @xmath59 here @xmath60 means complex conjugation of the function @xmath61 .. ] these two definitions of hermiticities are the same if the potentials are @xmath0 symmetric : @xmath62 ; otherwise , these two definitions will give rise to different kinds of potentials . each kind of potentials
is shown to admit critical points , i.e. critical potentials @xcite , and these critical potentials of different kinds can be related by analytic continuation of a parameter in the potentials : for example , the @xmath63 fractional - superstring critical potential @xcite is given as @xmath64 with a non - zero real number , @xmath65 , and the potentials , @xmath66 , is given by @xmath67 with @xmath68 .
, the parameter @xmath69 in eq .
( [ fractionalsuperstringcriticalpoints ] ) becomes @xmath70 .
see @xcite for detail . ]
that is , they are related by choosing a proper complex phase of each term in the potentials .
critical points of each case are characterised by the recursive relation of the orthonormal polynomials , @xmath71 with @xmath72 .
this can be translated in terms of the baker - akhiezer functions of @xmath1-component kp hierarchy as follows : @xmath73 here @xmath74 is a @xmath1-vector baker - akhiezer function which can be derived and @xmath75 was first proposed by @xcite in one - matrix models .
the case of the multi - cut two - matrix models is given in @xcite .
] from the orthonormal polynomials @xmath75 , and @xmath76 is its dual which comes from the dual polynomials @xmath77 .
the scaling parameters are defined as @xmath78 and the boundary cosmological constants @xmath16 as @xmath79 the lax operators in eq .
( [ smoothorthpoly ] ) are related by @xmath80 and in particular , in the fractional - superstring critical points @xmath81 @xcite , the pair @xmath82 is shown to be expressed as follows @xcite : @xmath83 with the matrix @xmath11 which is given by the @xmath84 _ shift matrix _ , @xmath85 the matrix @xmath86 is the @xmath84 unit matrix . @xmath87 and @xmath88 are non - zero real constants .
is a critical value of @xmath89 with @xmath90 .
see the definition in @xcite .
] an important consequence from the matrix models is that _ all the coefficient functions @xmath91 and @xmath92 in eq .
( [ fractionalsuperstringcriticalpoints ] ) are real functions _ @xcite . in the case of real - potential critical points ,
what we need to do is just replacing the shift matrix @xmath11 by @xmath93 : @xmath94 that is , @xmath95 here all the coefficient functions @xmath96 and @xmath97 in eq .
( [ fractionalsuperstringcriticalpointsreal ] ) are also real functions @xcite . from these lax operators , one can conclude that the fractional - superstring critical points ( except for two - cut cases ) do not have the @xmath0 symmetry even right at the critical points , @xmath98 since the @xmath0 symmetry requires that the leading matrix of operator @xmath99 should be @xmath100 @xcite .
note that the critical points of even - number - cut models admit the following residual @xmath101 symmetry : @xmath102 and the lax operators @xmath103 satisfy the following constraint : @xmath104 with the @xmath101 grading matrix @xmath105 which is defined by @xmath106 in particular , the two - cut ( @xmath107 ) critical potentials preserve the @xmath101 symmetry at the critical points .
in the two - cut cases . in this sense , the @xmath101-breaking critical points in the two - cut matrix models are missing in our analysis .
since they correspond to string theories flowed by ramond - ramond perturbation of the type 0
superstring theory @xcite , it is interesting system to study . ] in the section below , we consider the following special pair of the lax operators : @xmath108 which satisfies the douglas equation @xcite : @xmath109= g_{\rm str } i_k$ ] .
note that the operators on the dual side @xmath110 are simply related by transpose : @xmath111 which also satisfy the douglas equation : @xmath112= g_{\rm str } i_k$ ] .
consequently , the recursive relation of orthonormal polynomials ( [ smoothorthpoly ] ) can be reexpressed as @xmath113 with @xmath114,\quad \widetilde \psi(t;\tilde \zeta ) = \widetilde \psi(t;\tilde \zeta ) \ ,
\exp\bigl [ \frac{-a^{-\frac{\hat p+\hat q}{2}}}{g_{\rm str } } v_2\bigl(a^{\frac{\hat q}{2}}\tilde \zeta\bigr ) \bigr],\end{aligned}\ ] ] and @xmath74 and @xmath115 are defined in eq .
( [ smoothorthpoly ] ) . in this section , macroscopic
loop amplitudes are studied in the fractional - superstring critical points .
we adopt the daul - kazakov - kostov prescription @xcite and its multi - cut generalization @xcite to obtain macroscopic loop amplitudes .
first we note the relationship between macroscopic loop amplitudes and the lax operators @xcite@xcite@xcite@xcite .
the macroscopic loop amplitudes we consider are following : @xmath116 the basic idea comes from the exact expression of orthonormal polynomials @xcite , @xmath117 here @xmath118 is the matrix integral of the @xmath119 truncated matrices @xmath120 , @xmath121 from these relations , eqs .
( [ macroresol ] ) and ( [ orthalphabeta ] ) , we expect that the eigenvalues of the lax operators ( [ smoothorthpoly ] ) are related to the macroscopic loop amplitudes in the following way : @xmath122 in the large @xmath123 limit .
this claim has been argued and proved in various contexts @xcite @xcite @xcite@xcite . in terms of @xmath9 of eqs .
( [ pqacrel1 ] ) and ( [ pqacrel2 ] ) , this means that these amplitudes can be derived as the simultanious eigenvalues of the lax pair : @xmath124 in the weak string coupling limit , @xmath125 , ( or the dispersionless kp hierarchy limit ) .
this problem was solved in the one - cut two - matrix models @xcite and also in @xmath0 symmetric critical points of the multi - cut two - matrix models @xcite . in the following ,
we study this prescription in the fractional - superstring critical points ( [ fractionalsuperstringcriticalpoints ] ) .
although the correspondence seems to be between the single macroscopic loop amplitude @xmath126 and the @xmath1 eigenvalues of the operator @xmath127 , this is argued to be consistent in section [ multicutgeometrystokessection ] .
the system we consider is the douglas equation @xmath109= g_{\rm str}i_k$ ] with the pair of @xmath84 matrix valued differential operators @xmath128 and @xmath20 defined in eqs .
( [ fractionalsuperstringcriticalpoints ] ) and ( [ pqacrel1 ] ) in the @xmath29-rotated - potential critical points .
the real - potential critical points are studied separately in section [ realpotentialmodels ] . in the leading order of the week coupling expansion
, one can factorize the operators into a dimensionful part and a dimensionless part without any ambiguity of ordering : and @xmath127 to simplify the discussion . ]
@xmath129 given these ansatz , one can show that the matrices satisfy the leading douglas equation , @xmath130 = g_{str } i_k \quad \rightarrow \quad [ { \boldsymbol{\pi}}(z),{\boldsymbol{\xi}}(z ) ] = 0 \qquad \text { : the leading equation } , \label{leadingdouglasaa}\end{aligned}\ ] ] which means that the two matrix functions @xmath131 and @xmath132 commute with each other .
in the next leading order of the douglas equation , the ordering among @xmath133 variables in the expression ( [ dimensionseparation ] ) could cause some problem @xcite .
this ordering problem can be solved by first solving simultanious diagonalization of operators @xmath128 and @xmath20 in the week coupling limit , or equivalently @xmath131 and @xmath132 : @xmath134 here @xmath135 denotes equality upto some similarity transformation .
the next leading order of the douglas equation then turns out to be the ezj - dkk equation @xcite@xcite , @xmath136 note that constant appearing on the right - hand side of eq .
( [ ezj - dkkeq ] ) is common for all @xmath23 . up to now the basic procedure to solve the equations
is the same as the @xmath0 symmetric cases @xcite .
the major difference comes from the lax operators ( [ dimensionseparation ] ) which break the @xmath0 symmetry ; if there is the @xmath0 symmetry , the diagonalization is trivially solved @xcite . since there is no @xmath0 symmetry now , the characteristic equation of the lax operators : @xmath137 = 0 , \qquad \det\bigl[q i_k- \lambda^{\hat q } { \boldsymbol{\xi}}(z)\bigr ] = 0 \label{charapq}\end{aligned}\ ] ] can have various kinds of solutions which are not algebraic in general .
this makes it difficult to obtain general formulae of the eigenvalues in the fractional - superstring critical points . , is smaller than 5 , @xmath138 .
] in view of this , our strategy here is that we first propose a porper ansatz which can solve the ezj - dkk equation ( [ ezj - dkkeq ] ) , and then show that they are actually the solutions to the characteristic equations ( [ charapq ] ) . in this procedure , the uniqueness of the solution is not granted but our purpose is to obtain non - trivial solutions of the system .
, there are several exceptional solutions which have no direct correspondence in the liouville theory calculation @xcite .
therefore , it is also interesting to consider such a solution in our multi - cut system but it is out of the scope of this paper . ] our ansatz for the @xmath81 fractional - superstring critical points is as follows : _ the eigenvalues of @xmath139 and @xmath127 are given by hyperbolic cosines with some proper phase shifts _
, i.e. , the eigenvalues @xmath140 and @xmath141 are written as @xmath142 here we define _ deformed chebyshev functions _
, if @xmath143 .
] @xmath144 , which can also be expressed as @xmath145 where @xmath146 .
the polynomials @xmath147 and @xmath148 are the chebyshev polynomials of the first and second kinds : @xmath149 , @xmath150 .
the phase @xmath151 in front of the hyperboic cosine of the solutions ( [ ouransatz ] ) is necessary to have correct asymptotic behavior @xmath152 ( see eq .
( [ asympandq ] ) ) .
importantly , one can check that the deformed chebyshev functions are always solutions to the ezj - dkk equation , the ezj - dkk equation for the general series @xmath2 is obtained by changing the canonical pair @xmath153 to @xmath154 in @xcite . here
@xmath155 is the most relevant coupling constant ; @xmath22 is the cosmological constant and @xmath156 .
note that this extension is also applicable to the multi - cut cases : @xmath157 therefore , the defomed chebyshev polynomials are general solutions to the ezj - dkk equation . ] with an arbitrary phase shift , @xmath158 : @xmath159 therefore , what we need to check next is which phase shift @xmath160 is allowed by the characteristic equations ( [ charapq ] ) .
there are two constraints imposed by the characteristic equations ( [ charapq ] ) : * since the matrices @xmath131 and @xmath132 are matrix - valued polynomials in @xmath161 , the characteristic equations of @xmath139 and @xmath127 ( eq . ( [ charapq ] ) ) must be _ polynomial equations _ in @xmath162 of degrees @xmath163 for @xmath139 , and in @xmath164 of degrees @xmath165 for @xmath127 .
* since the coefficients of lax operators @xmath139 and @xmath127 are all real functions @xcite , _ all the coefficients of the characteristic equations ( [ charapq ] ) must be real functions .
_ noticing that the characteristic equations of eq .
( [ ouransatz ] ) are generally written as @xmath166 one can see that the first requirement results in @xmath167 and then the algebraic equation ( [ charapq ] ) is given by @xmath137 = \frac{\lambda^{k\hat p}}{2^{k-1}}\bigl ( ( -1)^n t_k(e^{\frac{\pi i n}{k}}\zeta/\lambda^{\hat p } ) - t_{k\hat p}(z ) \bigr)=0 . \label{charafirstsol } \end{aligned}\ ] ] furthermore , the second constraint on eq .
( [ charafirstsol ] ) ( i.e. @xmath168 ) requires that the phase shift must be @xmath169 therefore , the fractional - superstring critical points of the multi - cut matrix models admit two different kinds of solutions : @xmath170 = 2 \lambda^{k\hat p}\bigl(t_k(\zeta/\lambda^{\hat p } )
- t_{k\hat p}(z)\bigr ) = 0 , { \nonumber}\\ \underset{{\displaystyle}\vspace{1cm}(\delta = \pm 1/4)}{\underline{\text{sinh solution}}:}&\quad \left\ { \begin{array}{r } { \displaystyle}\pi^{(j)}_{\hat p}{}(z ) = \sinh\bigl(\hat p \tau + 2\pi i \frac{j-1}{k}\bigr ) = i \,t_{\hat
p}^{(\frac{j-1}{k}-\frac{1}{4})}(z ) , \cr { \displaystyle}\xi^{(j)}_{\hat q}{}(z ) = \sinh\bigl(\hat q \tau + 2\pi i \frac{j-1}{k}\bigr ) = i \ ,
t_{\hat q}^{(\frac{j-1}{k}-\frac{1}{4})}(z ) .
\end{array } \right .
\label{sinhsolution } \\
\text{with}\quad & \det \bigl [ ( 2 \zeta ) \,i_k - { \boldsymbol{p}}(z)\bigr ] = 2 \lambda^{k\hat p } \bigl ( ( -1)^{\frac{k}{2}}t_k(-i \zeta/\lambda^{\hat p } ) - t_{k\hat p}(z)\bigr ) = 0 . { \nonumber}\end{aligned}\ ] ] these solutions are referred to as _ cosh solutions _ and _ sinh solutions _ , respectively .
existence of these solutions depends on the parity of @xmath1 : @xmath171 note that the sinh solutions are essentially equivalent to the cosh solutions when @xmath172 : @xmath173 consequently , the cases of @xmath30 can be understood as a mixture of bosonic and type 0 superstring amplitudes .
in particular , the cases of @xmath174 include the one - cut phase and two - cut phase solutions of type 0 superstrings ( see @xcite ) at the same time . in this sense
, it tempts us to interpret this system as type 0 superstring theory with the @xmath175 fzzt branes , but the annulus amplitudes among them @xcite seem to be different . and @xmath176 are non - local @xcite _
if @xmath175 are treated equally_. however from the the free - fermion viewpoints @xcite , the @xmath177 sector and the @xmath178 sector are treated equally @xcite and the annulus amplitudes between the @xmath177 sector and the @xmath178 sector can not have non - local annulus amplitudes @xcite . ]
also note that , on the cft side , there is no zz brane which connects the @xmath179 fzzt branes @xcite ; in the case of @xmath30 , there are zz branes which connect the @xmath177 sector to the @xmath178 sector ( see the next section ) .
the real - potential critical points are given by the following lax pair : @xmath180 with the real - potential gamma matrix @xmath181 @xmath182 note that all the coefficient matrices @xmath183 and @xmath184 are real functions @xcite . by taking into account these facts and by using the same procedure as in the previous section
, one can find the following two types of solutions at the real - potential critical points : @xmath185 = 2 \lambda^{k\hat p}\bigl(t_k(\zeta/\lambda^{\hat p } ) + t_{k\hat p}(z)\bigr ) = 0 , \\
\underline{\text{sinh solution}}:&\quad \left\ { \begin{array}{r } { \displaystyle}\pi^{(j)}_{\hat p}{}(z ) = \sinh\bigl(\hat p \tau + 2\pi i \frac{2j-1}{2k}\bigr ) =
i \,t_{\hat p}^{(\frac{2j-1}{2k}-\frac{1}{4})}(z ) , \cr { \displaystyle}\xi^{(j)}_{\hat q}{}(z ) = \sinh\bigl(\hat q \tau + 2\pi i \frac{2j-1}{2k}\bigr ) = i \ ,
t_{\hat q}^{(\frac{2j-1}{2k}-\frac{1}{4})}(z ) .
\end{array } \right . \\ \text{with}\quad & \det \bigl [ ( 2 \zeta ) \,i_k - { \boldsymbol{p}}(z)\bigr ] = 2 \lambda^{k\hat p}\bigl((-1)^{\frac{k}{2}}t_k(-i\zeta/\lambda^{\hat p } ) +
t_{k\hat p}(z)\bigr ) = 0 . \end{aligned}\ ] ] appearance of these solutions also depends on the parity of the number of cuts @xmath1 : @xmath171 in the following , we compare the critical points from two types of the matrix - model potentials with different hermiticity : * when @xmath1 is odd , the cosh solutions of real potentials are equivalent to the cosh solution of @xmath29-rotated potentials : @xmath186 therefore , these systems are equivalent . *
when @xmath1 is even and @xmath187 is odd , the cosh and sinh solutions of real potentials are equivalent to the sinh and cosh solutions of @xmath29-rotated potentials , respectively : @xmath188 therefore , these systems are also equivalent . * when @xmath172 , the cosh solutions and sinh solutions of real potentials are equivalent to each other for the same reason as eq .
( [ 4nmidkseqc ] ) .
however the solutions of real potentials are _ distinct _ from the solutions of @xmath29-rotated potentials .
note that although these solutions in the real - potential models look similar , they are quantitatively _ distinct _ when @xmath172 . in particular , the @xmath30 cases of @xmath29-rotated - potential critical points can be viewed as a mixture of bosonic and superstring system ; but the @xmath30 cases of real - rotated - potential critical points are not . interestingly on the other hand ,
the cosh and sinh solutions of @xmath30 are degenerate in both @xmath29-rotated - potential and real - potential cases . therefore in this sense ,
the total number of different solutions are still _ two _ in the @xmath30 cases .
we have obtained the solutions by imposing two necessary conditions on the characteristic equations of lax operators . in the discussion above , however , no specific form of the matrix - valued lax pair @xmath189 is specified .
one may wonder whether some actual expression of the lax operators exists , or how does the expression look like .
the actual realization of the matrix - valued lax operators is also non - trivial and important .
this problem is referred to as _ matrix realization _ of the solutions .
purpose of this section is to give explicit matrix realizations of the cosh and sinh solutions . in order to identify any matrix realization of the cosh and sinh solutions of eqs .
( [ coshsolution ] ) and ( [ sinhsolution ] ) , we need to find out a pair of the matrix - valued polynomials @xmath131 and @xmath132 , which commute with each other , @xmath190=0 $ ] , and whose eigenvalues are given as eqs .
( [ coshsolution ] ) or ( [ sinhsolution ] ) of the cosh / sinh solutions . while it is possible to find several realizations , say for the @xmath191 cases , it turns out that a matrix realization of our solution is simply given by the following matrices @xmath192 : @xmath193 , \label{matrealcomplexcases}\end{aligned}\ ] ] with @xmath194 here @xmath147 is the chebyshev polynomial of the first kind , and the coefficients @xmath195 and the matrix @xmath196 are defined as @xmath197 then the claim is that the lax pair is given by these matrices : @xmath198 for every pair of @xmath2 .
the integer @xmath199 is basically a free parameter and is chosen to be an even number when @xmath1 is even .
it is because if @xmath1 is even , there is the residual @xmath101 symmetry ( [ residualz2sym ] ) and this requires the following condition , @xmath200 which results that @xmath199 should be even .
for later convenience , we collect relevant formulae among the matrices appearing here : @xmath201 in order to check that the matrix - realization ( [ matrealcomplexcases ] ) solves the douglas equation ( [ leadingdouglasaa ] ) , @xmath202 = 0 , \label{comofmatrixrealization}\end{aligned}\ ] ] for every pair of @xmath2 , we found that the following condition on @xmath203 is sufficient : @xmath204 \neq 0,\qquad [ ( \gamma\pm \gamma^{-1}),m_\pm ] = 0 .
\label{gammam}\end{aligned}\ ] ] by repeatedly using these relations ( [ gammam ] ) and the addition formulae of the chebyshev polynomials ( or the formulea of hyperbolic cosines ) , one can prove eq .
( [ comofmatrixrealization ] ) .
one can verify that the matrices @xmath205 in eq .
( [ matrixrealizationcomplex ] ) satisfy this sufficient condition .
secondly , one needs to show that these matrices in eq .
( [ matrealcomplexcases ] ) give rise to the algebraic equation for every number of @xmath1 : @xmath206 & = 2 \lambda^{k\hat p}\bigl(t_k(\zeta/\lambda^{\hat p } ) - t_{k\hat
p}(z)\bigr ) = 0 , { \nonumber}\\ \det \bigl [ ( 2 \zeta ) \,i_k - { \boldsymbol{q}}^{(\hat p)}_- ( z)\bigr ] & = 2 \lambda^{k\hat p } \bigl ( ( -1)^{\frac{k}{2}}t_k(-i \zeta/\lambda^{\hat p } ) - t_{k\hat
p}(z)\bigr ) = 0 .
\end{aligned}\ ] ] unfortunately , we have not found any general proof of this relation .
instead , we have checked this condition by using _
mathematica_@xmath207 up to very higher order of @xmath208 and @xmath1._{k\times k},\qquad t_k(z ) -t_k(\zeta ) = \frac{1}{2 } \det\bigl[\gamma z-\zeta i_k + \frac{1}{2}m_k\bigr]_{k\times k},\end{aligned}\ ] ] which are different from nash s determinant formula @xcite . ] as another non - trivial check , let us consider @xmath107 , the two - cut critical points .
since the gamma matrix satisfies @xmath209 , the lax operators can be simplified as and @xmath210 . ]
@xmath211 , { \nonumber}\\ \text{sinh solutions:}\qquad { \boldsymbol{q}}^{(\hat p)}_- & = \lambda^{\hat p}\ ,
\bigl[\gamma t_{\hat p}(z ) + 2e\delta \bigl(t_{\hat p-1}(z ) + t_{\hat p-3}(z)+ \cdots
\bigr ) + { \nonumber}\\ & \qquad + 2 \gamma \bigl(t_{\hat p-2}(z)+ t_{\hat p-4}(z)+\cdots\bigr ) \bigr]{\nonumber}\\ & = \lambda^{\hat p } \ ,
\bigl[u_{\hat p-1}(z ) \bigl(z\gamma + e\delta\bigr)\bigr].\end{aligned}\ ] ] this is identical to the matrix realization given in @xcite . the matrix realizations of the real - potential solutions are also obtained as follows : despite of the difference in the sinh solutions , actually the matrix realization is very similar .
what one needs to do is just replace @xmath11 by @xmath181 in the realization of @xmath29-rotated - potential critical points .
that is , if one introduces the following lax operators : @xmath212,\end{aligned}\ ] ] with @xmath213 the matrix realization of each solution is given as @xmath214 for further illustrations , we list some of the examples for @xmath205 in appendix [ examplemrealization ] .
in this section , we study the algebraic curves of the cosh and sinh solutions .
the central object is the algebraic equations @xmath215 of macroscopic loop amplitudes @xmath126 . as a definition of algebraic equations ,
we adopt the definition given in the free - fermion formulation @xcite , especially its multi - component extension @xcite , where the form of the equation is required from the @xmath216-constraints , i.e. the loop equations of the matrix models .
the definition is given as follows : macroscopic loop amplitudes @xmath126 and the eigenvalues of the lax operators @xmath9 are now related as @xmath217 the index @xmath48 is the label of the eigenvalues @xmath23 .
since there are many @xmath218 corresponding to the single @xmath126 , the algebraic equation @xmath215 is defined so that it includes _ all the eigenvalues @xmath218 as its solutions _ @xcite . the macroscopic loop amplitude @xmath218 is now written as @xmath219 ) } ( -i \zeta/\lambda^{\hat p } ) , \label{solqofj}\end{aligned}\ ] ] where the phase shifts @xmath26 are given as @xmath220 with @xmath23 . by introducing the @xmath208 branches of the loop amplitudes @xmath221 @xcite , and @xmath222 . ]
@xmath223+\frac{a}{\hat p } ) } ( -i \zeta/\lambda^{\hat p } ) & \text{(sinh solutions ) } , \label{qofzetaplusa } \end{array } \right.\end{aligned}\ ] ] the algebraic equations are defined as @xmath224 = 0.\end{aligned}\ ] ] in the case of @xmath225 , this goes back to the expression @xcite ( the one - cut two - matrix models ) and @xcite ( the @xmath216 approach ) .
generally this is the definition given in @xcite ( the multi - component cases )
. organization of this section is following : algebraic equations are in section [ algebraicequationssection ] , branch points are in section [ branchcurvessection ] , the multi - cut geometry is in section [ multicutgeometrystokessection ] and singular points are in section [ algsingularpointssection ] .
let us obtain the algebraic equations of our solutions .
first consider the multiplication among different branches with a fixed eigenvalue @xmath48 , which results in @xmath227+\frac{a}{\hat p } ) } ( -i \zeta/\lambda^{\hat p } ) \bigr ) = { \nonumber}\\ & \qquad = \frac{\lambda^{\hat p/\hat q}\ , i^{\hat p } } { 2^{\hat p-1 } } \times \bigl(t_{\hat p}(-i q/\lambda^{\hat q } ) -t_{\hat q}^{(-(\hat q-\hat p)[\nu_j-\frac{1}{4}])}(-i\zeta/\lambda^{\hat p})\bigr ) .
\end{aligned}\ ] ] introducing the following integers @xmath228 and @xmath229 @xcite , @xmath230 one can rewrite @xmath231 as @xmath232 and similarly , @xmath233 here @xmath234 is the part which appears only in the real - potential models . since @xmath235 and @xmath229 are coprime integers , the integer @xmath236 takes each value @xmath237 exactly @xmath238 times .
considering @xmath239 , one concludes that the algebraic equations of the cosh and sinh solutions at the @xmath81 critical points are expressed as @xmath240 here again @xmath234 is the part which appears only in the real - potential models .
we have introduced the @xmath241 labeling of minimal fractional superconformal field theory @xcite , @xmath242 note that this factorization property , @xmath243 , can be understood as a consequence of the accidental _
@xmath226 symmetry _ of this background .
that is , the @xmath1-cut two - matrix models include @xmath238 copies of @xmath81 minimal fractional superstring theory and the single minimal fractional superstring theory can be singled out after imposing @xmath244 gauging of the system @xcite .
odd models of minimal superstrings from the two - cut two - matrix models . ] in this sense , the cosh and sinh solutions are consistent with the @xmath226 gauging requirement which has been anticipated from the liouville side .
since these copies become _ identical _ after the @xmath226 gauging , the algebraic equations of the liouville side , @xmath245 , are given as @xmath246 let first consider branch points of the curves .
branch points here are defined as @xmath247 .
our current purpose is to see where the branch points exist on the branch ( [ solqofj ] ) .
these branches ( or solutions ) ( [ solqofj ] ) of the algebraic equation are particularly important because they correspond to macroscopic loop amplitudes ( [ solutionsinsecalgeq ] ) .
therefore , we here refer to these branches as _ physical branches_. for sake of simplicity , we put @xmath248 and consider the following loop amplitudes : @xmath249 the branch point of this expression is obtained from @xmath250 since the function @xmath251 is finite on @xmath252 , the branch points exist only if @xmath253\quad \leftrightarrow\quad
\zeta = ( -1)^m , { \nonumber}\\ \sinh(\hat q \tau + 2\pi i \nu)\neq 0 & \quad \leftrightarrow \quad \tau \neq \pi i \bigl[\frac{n-2\nu}{\hat q}\bigr]\quad \leftrightarrow\quad q^{(j)}\neq ( -1)^n,\end{aligned}\ ] ] with @xmath254 .
that is , there are generally _ two _ branch points @xmath255 on this branch @xmath218 .
it is also convenient to consider when one of the two points @xmath256 becomes a normal point : it happens if they satisfy @xmath257 = \pi i \bigl[\frac{n-2\nu}{\hat q}\bigr ] \quad & \leftrightarrow\quad \hat q m+\hat p n = 2 \nu ( \hat p-\hat q).\end{aligned}\ ] ] therefore , the condition that one of the branch points disappears is given as @xmath258 = \eta \frac{j-1}{\hat k } \in \frac{1}{2 } \mathbb z , { \nonumber}\\ \text{sinh solution:}&\qquad ( \hat q-\hat p ) \bigl[\frac{j-1}{k } -\frac{1}{4}\bigr ] = \eta \frac{j-1}{\hat k } - \frac{\hat q-\hat p}{4 } \in \frac{1}{2 } \mathbb z. \end{aligned}\ ] ] this indicates that the branch points only appear when @xmath259 .
the positions of cuts on these branches for the unitary cases @xmath260 are shown in fig.[physbranch ] . only include one cut and exactly the same branch as bosonic string .
the cases of @xmath178 are the same as the two - cut phase of type 0 superstrings ; in sinh solutions , the branches of @xmath178 only include one cut and exactly the same branch as bosonic string .
the cases of @xmath177 are the same as the two - cut phase of type 0 superstrings . in all the branches , there are maximally two branch points on each sheet . ]
note that the curves essentially depend on the effective phase shift @xmath261 .
these curves typically appear in the @xmath235-cut matrix models or @xmath235-fractional superstring theory . here
we refer to these curves as _ @xmath262-string _ curves . for example
, @xmath263- and @xmath264- string curves are equivalent to the curves of bosonic minimal string theory , and @xmath45-string curves are equivalent to minimal type 0 superstring theory . also note that one can also consider dual branch points @xmath265 , which result in the opposite condition , @xmath266 . in the previous subsection ,
we have obtained branch points of the physical branches of macroscopic loop amplitudes , @xmath218 , and it turns out that they include at most _ two _ cuts on each branch .
since we have started our discussion from the @xmath1-cut matrix models , there should be @xmath1 cuts in the macroscopic loop amplitudes @xmath126 .
that is , _ where is the multi - cut geometry ? _
a simple answer to this question is the follwing : the @xmath1 cuts exist on _ physical sheet _ of the @xmath1-cut matrix models .
so far , we have obtained eigenvalues @xmath267 of the lax operators
@xmath9 in the weak coupling limit , @xmath268 and calculated the corresponding algebraic equations ( [ ourcurvemat ] ) .
as one may notice , these algebraic curves are highly reducible .
for example , the algebraic equation of cosh solutions , .
] can be factorized in the following way : @xmath269 where @xmath26 is the phase shift of the solution , and @xmath270 is the irreducible algebraic equation of an eigenvalue @xmath218 of the lax operators : @xmath271 \bigl [ t_{\hat p}(q/\lambda^{\hat q})-t_{\hat q}^{(-(\hat q-\hat p)\nu)}(\zeta/\lambda^{\hat p})\bigr ] = 0 .
\label{factorizationcurve2}\end{aligned}\ ] ] this is the origin of the fact that there are maximally _ two _ cuts on the physical branch ( [ solqofj ] ) and the @xmath1 necessary cuts are missing . here we have used two different words _ physical branch _ and _ physical sheet_. although the word , physical sheet , is enough to explain everything in the usual one - cut cases , the multi - cut matrix models require to distinguish them .
_ physical sheet _ ,
@xmath272 , is a complex plane @xmath273 on which the resolvent @xmath274 ( or macroscopic loop amplitude @xmath126 ) is defined and each weak coupling infinity ( say , @xmath275 ) gives correct asymptotic behavior : @xmath276 therefore , in the @xmath1-cut matrix models , @xmath126 on the physical sheet is expected to have @xmath1 semi infinite cuts which separate the @xmath1 weak coupling regimes . on the other hand , if one solves the algebraic equation ( [ ourcurvemat ] ) , generally @xmath1 solutions ( i.e. branches ) satisfy the condition ( [ physicalsheet ] ) with some particular value of @xmath48 . due to the definition ( [ bakakhstokes ] ) ,
they are naturally identified with @xmath218 @xmath277 .
therefore , these solutions or branches are called _ physical branch_. physical sheet and physical branch are equivalent concepts in the one - cut cases .
note that only the cuts on physical sheets have the physical meaning as condensation of the matrix - model eigenvalues @xmath278 and the position of the cuts can have the geometrical meaning .
mathematically , position of cuts is meaningless and one can freely move it continuously . because of this fact , when some of the cuts on physical sheet have no branch point , the physical sheet is factorized into several pieces of curves , and these cuts on physical sheet disappear from the viewpoint of branch points in the algebraic equation .
this usually happens when branch points are joined and several cuts are connected .
examples are found in the one - cut phase of the two - cut matrix models @xcite and also `` one - cut phase '' counterpart of the @xmath1-cut two - matrix models of the @xmath0 symmetric critical points @xcite . in these cases , although the cuts are missing from the algebraic equations , one has to take into account the position of the cuts because they have the definite physical meaning . in our fractional - superstring solutions , the curves are factorized and many cuts are obviously missing .
this means that we need to add the cuts and glue the physical branches to make up the physical sheet .
identification of these cuts generally requires non - perturbative analysis , but at least in the unitary cases @xmath279 , , which includes @xmath2 odd models of type 0 superstring theory . ] the following connection rule of the macroscopic loop amplitude @xmath126 on the physical sheet ( with the @xmath1 eigenvalues @xmath218 of eq .
( [ bakakhstokes ] ) ) is consistent with the multi - cut geometry : @xmath280 that is , @xmath218 is the solution of the weak - coupling regime @xmath281 .
therefore , we propose it as a conjecture which can be checked by some non - perturbative analysis .
we show several examples of the positions of the cuts in fig .
[ fig1 ] and fig .
[ fig2 ] .
( @xmath282 ) .
the physical branch of each irreducible curve is shown , which corresponds to some fractional superstring theory of @xmath283 ( denoted as @xmath284-strings ) .
note that this is the geometry of unitary cases @xmath285 . ]
( @xmath282 ) .
the physical branch of each irreducible curve is shown , which corresponds to some fractional superstring theory of @xmath283 ( denoted as @xmath284-strings ) .
note that this is the geometry of unitary cases @xmath285 .
] note that , in the @xmath29-rotated - potential cases , the coordinate @xmath16 is rotated by @xmath29 in the complex plane ( see eq .
( [ relxandzeta ] ) ) . therefore , for example the cuts on the physical sheet in the two - cut matrix models should only appear along the pure - imaginary axis , although the cosh solutions only have cuts along the real axis .
our solution to this issue is that they are hided by another physical branch and do not appear on the physical sheet . in this way
, the connection relation ( [ connectionrelation ] ) of the macroscopic loop operator ( or resolvent ) provides consistent relationship between the physical branches and the multi - cut geometries .
although we can not deny the possibility that there is another meaningful connection rules , but we leave it as futur works . also note that we put each irreducible curve some particular names , as @xmath286-strings .
as one sees in eq .
( [ connectionrelation ] ) , these macroscopic loop amplitudes have many non - trivial jumps on the physical sheet .
this is understood as _ stokes phenomenon _ can have different asymptotic forms with respect to different angular regimes of @xmath16 if the expansion is not a convergent series .
each angular sector is now called a stokes sector .
the claim is that stokes sectors of the baker - akhiezer functions are regimes between cuts of the matrix models .
] of the baker - akhiezer functions @xcite : asymptotic behavior of the baker - akhiezer functions ( @xmath287 ) can be changed by an analytic continuiation of @xmath16 which crossese the cuts on the physical sheet .
therefore , the relation ( [ connectionrelation ] ) indicates how to connect there stokes sectors in the non - perturbative baker - akhiezer function . from the definition ( [ bakakhstokes ] ) and the relation ( [ connectionrelation ] ) ,
the leading behaviors of the recursive equations can be solved as @xmath288\ , \chi^{(j)}(t;\zeta ) , \label{asympsijei}\end{aligned}\ ] ] in @xmath289 . here
@xmath290 is a rank @xmath1 vector - valued function of sub leading corrections , and the expansion is carried out in some weak coupling regimes of the cosmological constant @xmath16 : @xmath291 . since the orthonormal polynomials are now related to the macroscopic loop amplitudes , @xmath292 these baker - akhiezer functions are connected to each other by an analytic continuation : , this relation has been shown to be true in @xcite ( i.e. eq . ( 4.16 ) ) by using the non - perturbative analysis of the baker - akhiezer functions @xcite . ]
@xmath293 another implication on the free fermion system is also noted in appendix [ zkchargefreefermion ] . in this subsection , we study singular points of the cosh and sinh solutions ( [ ourcurves ] ) .
the definition of singular points is @xmath294 one of the easy ways to obtain these points is to consider coincident points among the different branch expressions , @xmath221 , that is , the points @xmath295 which satisfy @xmath296 and eq .
( [ singularbranches ] ) .
this identification of singular points has been used in @xcite@xcite .
note that the pair @xmath297 can not be the same as @xmath298 . or equivalently , we need to solve for the following problem of finding the pair of @xmath299 : @xmath300 where @xmath301 with @xmath23 . in the case of the sinh solutions ,
the parameters @xmath51 and @xmath302 are shifted by @xmath303 : shift : @xmath304 , @xmath305 . ]
@xmath306 and @xmath295 is replaced by @xmath307 .
actually the discussion below does not depend on whether the solutions are cosh or sinh solutions and whether they come from @xmath29-rotated potentials or real potentials .
therefore , we assume that @xmath51 is a general rational number with a periodicity identification @xmath308 , that is , @xmath309 if one takes @xmath46 limit , the parameter @xmath310 will take the value in @xmath49 . the solutions to eq .
( [ singularbranchescond2 ] ) are obtained as the following pair of @xmath299 , which are parameterized by a pair of integers , @xmath311 : @xmath312,{\nonumber}\\ \tau'_{m , n } & = \pi i \bigl[+ \frac{1}{\hat p}\bigl(m-(\nu'-\nu)\bigr ) + \frac{1}{\hat q}\bigl(n-(\nu'+\nu)\bigr)\bigr],\end{aligned}\ ] ] and the corresponding points @xmath313 are expressed as @xmath314 , { \nonumber}\\ q_{m , n } & = ( -1)^{n}\lambda^{\hat q } \cos \bigl [ \pi \frac{(\hat q-\hat p ) ( \nu'-\nu ) - m \hat
q}{\hat p } \bigr ] .
\label{solofsingularpoints}\end{aligned}\ ] ] among these points , the branch points for @xmath126 ( or @xmath315 ) appear when @xmath316 for later convenience , here we introduce the @xmath317 charge according to the original @xmath0 charge conjugation of the @xmath1-cut matrix models , even though our critical points do not preserve the @xmath0 symmetry .
each eigenvalue of the operators @xmath139 and @xmath127 corresponds to each fzzt brane with a different charge @xcite . ] therefore , the phase @xmath51 defines the @xmath0 charge of fzzt branes as @xmath318 . \end{aligned}\ ] ] according to @xcite , the zz branes also have @xmath319 charges .
a natural definition here is that branch points ( [ chargebranch ] ) do not have this charge , which means that @xmath320 . \label{zkchargezzbrane}\end{aligned}\ ] ] the plus - minus sign means that there is a pair of zz brane and its charge conjugate at each singular point .
although the critical points break the @xmath0 symmetry , this system still admits the following @xmath101 charge conjugation , @xmath321 this @xmath101 charge conjugation is understood as a result of the hermiticity ( [ hermiticitypotentials ] ) , and the d - branes of @xmath322 are self - charge conjugate d - branes . in this section ,
we count the number of singular points .
first of all , one should resolve the basic periodicity of the parameters @xmath323 and @xmath324 in eq .
( [ singularbranchescond2 ] ) . since the integers @xmath2
are coprime , the paramters are identified as @xmath325 which can be rephrased in terms of the parametrization @xmath311 as follows : @xmath326 therefore , the parameterization @xmath311 can be restricted to the following fundamental domain , for instance , @xmath327 secondly , one has to avoid further double counting due to some accidental coincidence of the points @xmath328 in the expression eq .
( [ solofsingularpoints ] ) .
this happens if @xmath329 and @xmath330 satisfy either of the following two conditions : * the first condition is related to @xmath331 : @xmath332 in these cases , the expression ( [ solofsingularpoints ] ) turns out to be symmetric under the following transformations , respectively : @xmath333 here @xmath334 and @xmath335 are integers which satisfy @xmath336 with some proper integers of @xmath337 and @xmath338 . in the fundamental domain ( [ fundmn ] ) ,
there is a unique @xmath334 and are two different @xmath335 . *
the second condition is related to @xmath339 : @xmath340 in this case , the expression ( [ solofsingularpoints ] ) turns out to be symmetric under the following transformation : @xmath341 here @xmath334 and @xmath335 are half integers which satisfy @xmath342 with proper half integers of @xmath338 and @xmath337 . taking into account these facts
, we show the number of singular points in the following : [ [ section ] ] + the singular points among the same curves happens when the charges @xmath51 and @xmath302 satisfy @xmath343 in these cases , the points ( [ solofsingularpoints ] ) also include non - singular points like the branch points which happen if the pair @xmath344 satisfies the following condition : @xmath345 therefore , we have to exclude these cases from the counting . if @xmath346 or @xmath347 , then this can satisfy both of the conditions ( [ branchpointconditions1 ] ) and ( [ branchpointconditions2 ] ) . in this case
, we can say that there is no charged zz branes because this fzzt brane is self dual . by taking into account the accidental symmetry of eqs .
( [ accidentalsym1 ] ) and ( [ accidentalsym2 ] ) , one can count the number of zz branes which is given as @xmath348 the other cases satisfy part of the condition ( [ branchpointconditions1 ] ) and ( [ branchpointconditions2 ] ) . the cases of neutral zz branes ( @xmath349 )
are given as @xmath350 the cases of charged zz branes ( @xmath351 ) are given as @xmath352 [ [ section-1 ] ] + in these cases , all the zz branes are charged , and it is convenient to write the charges as follows : @xmath353 here @xmath354 are positive non - zero integers , @xmath355 are non - zero integers and @xmath356 and @xmath357 are pairs of coprime integers . in these cases , there is no branch point and the only thing one needs to take into account is the accidental coincidence ( [ accidentalsym1 ] ) and ( [ accidentalsym2 ] ) among @xmath358 points @xmath311 .
the numbers are following : @xmath359 and @xmath360 note that the last case should be @xmath361 which is automatic because @xmath362 and @xmath363 are coprime integers .
in this section , we calculate the fzzt- and zz - brane amplitudes from our macroscopic loop amplitudes .
we first review the relations between these brane amplitudes and the multi - cut matrix models before giving detailed derivations .
originally the fzzt brane is related to the orthonormal polynomials @xcite as @xmath364 = \exp\bigl[\frac{1}{g_{\rm str } } d_{\rm fzzt}(\zeta ) + \cdots\bigr ] .
\label{fzztact}\end{aligned}\ ] ] a new feature of the multi - cut matrix models is the @xmath0 charge with respect to the charge conjugation ( [ zkchargeconjint ] ) . although the @xmath0 charges are clearly formulated in the free - fermion system @xcite ( reviewed in appendix [ zkchargefreefermion ] ) , description of the @xmath0 charge within matrix models is not trivial .
a possible answer was provided in section [ multicutgeometrystokessection ] .
that is , the asymptotic behavior of the orthonormal polynomial ( [ fzztact ] ) depend on the regime of @xmath16 ( see eq .
( [ stokesbakeranaly ] ) ) . here
we claim that _ the @xmath0 charge @xmath365 of fzzt brane corresponds to the asymptotic regime @xmath281 of the orthonormal polynomials ( [ fzztact])_. consequently , there is only one kind of fzzt brane in this theory , and the @xmath0 charge of the fzzt brane expresses the position of the fzzt brane in the @xmath16 plane .
our description is natural because originally there is only one kind of matrix - model eigenvalue @xmath38 in the multi - cut matrix models , and also because this consideration is consistent with the original charge conjugation ( [ zkchargeconjint ] ) .
this also gives a support for our conjecture ( [ connectionrelation ] ) given in section [ multicutgeometrystokessection ] .
interestingly , since each asymptotic regime corresponds to a weak - coupling regime of string theory , this argument indicates that the ramond - ramond charge of fzzt brane is a concept which appears as a perturbative string artifact . in this description ,
the fzzt - brane charge is the label of asymptotic regimes in spacetime , and therefore resembles a spacetime coordinate . as noted in introduction , this description bears strong resemblance to the kaluza - klein compactification of gravity theory @xcite : compactification of a dimension results in @xmath49 gauge fields .
this idea will be elaborated in section [ secconclusion ] . according to this proposal ,
the @xmath0 charged fzzt - brane partition function @xmath366 with a charge @xmath365 is the orthonormal polynomials ( [ fzztact ] ) or the baker - akhiezer function @xmath367 in a particular perturbative regime : different functions as components . however it is known that they are related to the same fzzt brane with different background ( flux ) @xcite .
therefore , the leading disk amplitudes ( or fzzt - brane action ) of these components are the same . ]
@xmath368\qquad ( \zeta\to \infty\times e^{2\pi i \nu_0 } ) .
\label{fzztact2}\end{aligned}\ ] ] therefore , the fzzt - brane disk amplitude is @xmath369 see eq .
( [ dimensionseparation ] ) for the definitions of the functions @xmath370 and @xmath371 . with noting the relation of cosmological constant @xmath372 , the fzzt - brane disk amplitudes in the fractional superstring theory are calculated as @xmath373 , \label{fzztamp}\end{aligned}\ ] ] with @xmath374 .
the modular s - matrices on the liouville side are directly related to one - point function of ( bulk ) cosmological constant operator @xcite which can be easily calculated from the above expression as with @xmath16 fixed .
therefore , one needs to take into account eq .
( [ zetasigmab ] ) . ] @xmath375,\end{aligned}\ ] ] where we introduce the cft parameters , the liouville coupling @xmath376 and open channel momentum @xmath377 : @xmath378 the expression here is the same as bosonic and superstring liouville calculation ( see @xcite ) .
this means that the modular s - matrices for non - degenerate representations might be the same as bosonic or superstring cases .
the difference comes from the relation to the boundary cosmological constant : @xmath379 this information would be helpful for the boundary conformal bootstrap analysis @xcite of fractional super liouville field theory .
the instanton disk amplitudes were first studied in @xcite , and then realized in the free - fermion analysis @xcite .
the appearance of the zz - brane disk amplitudes in the two - matrix models was systematically studied in @xcite , and then also in the free - fermion analysis @xcite including two - cut cases @xcite which reproduce the results from the super - liouville calculations @xcite . according to @xcite , the definition of the amplitudes in the multi - component cases is simply given as ( [ defzz ] ) .
realization of charged zz branes in the multi - cut matrix models can also be understood by taking the differences of fzzt branes .
here we make a parallel comparison with @xcite to see that a straightforward generalization , eq .
( [ defzz ] ) , of the amplitude calculations can be justified in the cases of multi - cut two - matrix models . according to the saddle point method of the matrix models @xcite ( one - matrix ) and @xcite ( two - matrix ) ,
the partition function of the two - matrix model is given as @xmath380 in this mean field approximation ( in @xmath381 ) @xcite , one considers that the rest @xmath382 eigenvalues are in semi - classical configuration , ) , we make an intuitive argument below . ] and the expectation values are factorized : @xmath383,\end{aligned}\ ] ] and the effective action @xmath384 is expressed as follows : @xmath385 with @xmath386 as we have discussed in section [ multicutgeometrystokessection ] , these effective potentials ( or macroscopic loop amplitudes ) @xmath387 and @xmath388 take different asymptotic values ( as ( [ solqofj ] ) ) according to the position of @xmath38 or @xmath389 as in eq .
( [ connectionrelation ] ) .
therefore , the effective potentials ( outside the cuts ) ) .
however , it is not so important here because instanton actions always appear outside of the cuts . ]
are given as @xmath390 on each asymptotic regime ( @xmath391 ) of @xmath392 .
we should note eq .
( [ macrolaxrelation ] ) , especially the relation between dual macroscopic loop amplitudes @xmath393 and lax operators .
the eigenvalue @xmath38 and @xmath16 ( @xmath389 and @xmath394 ) are related as eq .
( [ relxandzeta ] ) .
therefore , on each sector the effective action is given as @xmath395 this is the origin of the @xmath0 charged zz branes in the multi - cut matrix models .
as in @xcite , the saddle point equations , @xmath396 , give possible saddle points @xmath397 , @xmath398 which are the intersection conditions of the curves ( which have been studied in section [ algsingularpointssection ] ) : @xmath399 among them , usually the branch points on physical sheet are most stable saddle points and this is the definition of perturbative action @xcite .
therefore , instanton actions are evaluated at the saddle points of singular points . by using partial integration
, the action ( [ effectiveactionsectors ] ) becomes @xmath400 the first term is the perturbative action which is usually non - universal and the second term is the instanton action @xcite . as one can see that the contour of the instanton action forms a closed loop @xmath401 in the whole curve @xmath215 ( see fig .
[ instanton ] ) , @xmath402 = \frac{-\lambda^{\hat p+\hat q}}{g_{\rm
str}}\oint_{\mathcal l } d\pi^{(j)}(z ) \ , \xi^{(j)}(z),\qquad z_{\rm branch } \in \mathcal l,\end{aligned}\ ] ] which can be decomposed into a sum of elemental pieces , i.e. zz - brane disk amplitudes ( see ( [ defzz ] ) ) , and is related to the vacuum energy of each irreducible curve in the weak - coupling limit .
in the multi - cut two - matrix models .
non - trivial connections among the irreducible curves ( drawn by doted lines ) are also singular points and are related to charged zz branes ( also see @xcite ) . each instanton loop forms a single topological loop and should include the starting point @xmath403 which defines the perturbative action . ]
the zz - brane amplitudes are now defined as the following difference between the corresponding pair of fzzt brane amplitudes in eq .
( [ fzztamp ] ) on the singular points : @xmath404 \sin\bigl [ \frac{(\hat q -\hat p)\bigl(m-(\nu_{j'}-\nu_j)\bigr)}{\hat p } \pi \bigr ] .
\label{defzz}\end{aligned}\ ] ] the derivative in the bulk cosmological constant @xmath22 is just given as @xmath405 \sin\bigl [ ( \dfrac{1}{b^{2 } } -1)\bigl(m-(\nu_{j'}-\nu_j)\bigr ) \pi \bigr ] .
\end{aligned}\ ] ] of course , this expression coincides with the one - cut calculation @xcite@xcite and the two - cut calculations in terms of the free fermion formulation @xcite and in the super - liouville theory @xcite .
in this paper , we have studied macroscopic loop amplitudes of @xmath2 minimal @xmath1-fractional superstring theory within the @xmath1-cut two - matrix models in the fractional - superstring critical points .
even though the system breaks the @xmath0 symmetry of the @xmath1-cut two - matrix models , the matrix models are still solvable in all of the critical points labeled by @xmath81 and the amplitudes can be expressed as a hyperbolic cosine or sine function .
although there is a difference between @xmath29-rotated critical points and real - potential critical points , the main structure of the solutions is the same and is simply given as @xmath406 with @xmath283 ( @xmath29-rotated potentials ) or @xmath407 ( real potentials ) .
since our investigation is a search for possible asymptotic solutions of the fractional - superstring critical points in the multi - cut matrix models , non - perturbative analysis and liouville theory calculations are pending , and we hope to address these points in future investigations .
possible issues related to the current work are listed in order : * there is only one type of solutions in the odd - cut models . according to the analogy to bosonic strings and type 0 superstrings @xcite
, this indicates that there is a bound in the cosmological constant , say @xmath408 .
this implies the following : the cosmological constant @xmath22 would be related to the `` bosonic '' cosmological constant @xmath409 as @xmath410 , which plays a role of suppressing large - area random surfaces @xmath411 in the liouville theory partition function @xmath412 of topology @xmath413 : @xmath414 in order to make this integral well - defined , there is a constraint on the cosmological constant , @xmath415 . on the other hand
when @xmath1 is even , it is natural to say that the cosh and sinh solutions would be exchanged with each other depending on the sign changing of the cosmological constant @xmath22 : @xmath416 .
these points should be checked in some actual formulation of fractional super - liouville theory and/or non - perturbative analysis of string equations .
* we have obtained two types of critical points : @xmath29-rotated potentials and real potentials .
our results show that they are distinct when @xmath30 . from the worldsheet point of view , the specialty of the @xmath172 cases is the appearance of an extra bosonic ( local ) current .
is even , among the parafermion fields @xmath417 , there is a special energy operator @xmath418 , whose conformal dimension is @xmath419 .
therefore , this is a local field .
this becomes a bosonic field when @xmath30 .
] therefore , it is interesting if there is a connection between this fact and our results . * as we have argued in section [ multicutgeometrystokessection ] , the highly reducibility of the algebraic equation indicates perturbative decoupling of several operators .
this should be observed also in the liouville calculations , as an extension of the calculation given in @xcite .
a succussful varification of this point will provide a non - trivial check of the correspondence between the multi - cut matrix models and fractional super - liouville theory @xcite . in this section ,
we give a discussion about a possible connection to the non - critical m theory proposed by p. hoava and c. a. keeler @xcite .
basic proposal given by @xcite is _ `` to identify the extra dimension of noncritical m - theory with the angular variable on the eigenvalue plane '' _ , and in order to realize this scenario , they introduced _
`` the double - scaled nonrelativistic fermi system in upside - down harmonic potential . '' _ in their proposal , the angular coordinate is identified so that the ramond - ramond flux of d0-brane in 0a background can be understood as kaluza - klein momentum of the kaluza - klein reduction along this angular direction . in this three ( or 2 + 1 ) dimensional
theory , type 0a and type 0b superstring theory are also shown to be included in its hilbert space @xcite .
on the other hand , in the multi - cut matrix models , there is the @xmath0 charged d - branes and the @xmath0 charge conjugation is the rotation in the eigenvalue space , or a translation of the stoke labeling @xmath51 : @xmath420 which moves one stokes sector to another .
therefore , the fractional superstring theory is also naturally understood as a kaluza - klein reduction about this non - perturbative direction @xmath51 in a three - dimensional gravity theory .
we here propose that this @xmath49 direction can be a possible candidate of _ the angular direction _ in the non - critical m theory .
first of all , let us mention the @xmath52 limit of our fractional - superstring amplitudes in order to compare the fractional - superstring amplitudes to the spectral curve of the hk non - critical m theory .
the @xmath52 limit is simply defined by the following limit of indices @xmath2 @xcite : @xmath421 this limit is interesting because it can relate the system to the @xmath422 non - critical fractional superstring theory in @xmath1-fractional superstring theory is given as @xmath423 . in this notation , @xmath52 means that spacetime dimension is two ( i.e. liouville + matter ) . ] or _ multi - cut matrix quantum mechanics _ @xcite .
our cosh and sinh solution are now expressed with the liouville parameters @xmath376 and @xmath377 in eq .
( [ liouvilleparam ] ) as @xmath424 one way to obtain the spectral curve of the @xmath52 system is to make use of the following combination : @xmath425 after taking the limit @xmath426 , this results in @xmath427 the universal eigenvalue density function @xmath428 or the canonical momentum of coordinate @xmath38 is extracted from @xmath429 in the following way @xcite : @xmath430 therefore , the phase space is given as @xmath431 in the two - cut cases , these phase spaces reproduce the spectral curve of type 0b superstring theory @xcite ( they are related to the @xmath177 sectors ) .
note that , since we consider complex integral of eigenvalues in the matrix model ( see ( [ definitionoftwomat ] ) and ( [ mcontour ] ) ) , the coordinate @xmath38 are generally complex and @xmath377 can be understood as proper time of the motion .
the trajectories on the complex @xmath38 space are shown in fig .
[ elliptic ] .
space trajectories in @xmath432 cut @xmath52 fractional strings ( cosh solution ) with an interpretation of real @xmath377 as proper time : @xmath433 with @xmath283 .
( a ) on the real axis ( @xmath177 ) , the trajectories are the same as bosonic strings : an eigenvalue comes from infinity , bounces and goes back .
( b ) on the imaginary axis ( @xmath434 ) , the trajectories are the same as type 0 superstrings : an eigenvalue comes from infinity , goes though the tip and goes to another infinity .
( c ) generally trajectories are similar to the superstring cases but the curves are slightly bent . ]
interestingly , each motion has a preferred cosmological constant , @xmath435 .
because of the hermiticity ( [ hermiticitypotentials ] ) , at these turning points on the real axes , the potential of the model is a purely real function and this @xmath436 can be understood as the energy level of the fermi sea . therefore
, this indicates that , for example , bosonic string vacua are more stable than the superstring vacuum and difference of energy level for one eigenvalue is finite . a main difference between the multi - cut matrix models and the hk non - critical m theory is the symmetry : although the spectral curve of the multi - cut matrix models looks the same as that of the hk non - critical m theory which preserve @xmath49 symmetry of the angular direction , the motions in the multi - cut matrix models break the @xmath49 symmetry .
it is because minimal fractional superstring theory generally breaks the @xmath0 ( i.e. @xmath49 ) symmetry .
however note that our solutions appear in the weak coupling analysis .
it is not so strange to have non - trivial geometry even if the theory is originally defined on a flat space coordinate .
is simply related to the flat coordinate @xmath437 as @xmath438 .
however , @xmath16 turns out to be on the non - trivial riemann surfaces in the weak coupling limit . ] our discussion here relies on the argument about the @xmath52 limit @xcite , but it should be important to define the proper multi - cut matrix quantum mechanics first and then to derive the above consideration from the first principle .
this also would clarify a further connection to the double - scaled fermion system .
we next also discuss the possibility that the multi - cut matrix models themselves can be interpreted as non - critical m theory .
not only the identification of the stokes labeling @xmath51 as the angular direction of the non - critical m theory , this theory implies an interesting way to include perturbative string theories ( i.e. bosonic or type 0 superstring theory ) as its vacua .
a hint is in the @xmath52 type 0b superstring theory @xcite : in this theory , there are two fermi seas and excitations in one of the fermi seas have no communication with another excitation in the other fermi sea in all order string coupling perturbation theory .
they communicate only through non - perturbative effects @xcite .
therefore , within the perturbation theory , there is no way to distingush the excitations as those of type 0b superstring theory or as those of bosonic string theory . in this sense
, one can also say that type 0b superstring theory includes perturbative bosonic string theory as _ a perturbatively isolated sector_. the case of the fractional - superstring critical points is following : * the origin of this _ perturbative isolation _ of sectors in the fractional superstring cases comes from the fact that the macroscopic loop amplitudes obtained in the critical points are highly reducible ( [ factorizationcurve1 ] ) and factorized into @xmath32 irreducible algebraic equations . as it has been mentioned in introduction ( footnote [ perturbativeisolation ] ) , this indicates that these irreducible algebraic curves communicate with each other only through non - perturbative effects .
therefore , the fractional superstring theory includes @xmath32 different perturbatively isolated sectors in its vacuum or in the weak - coupling landscape of this string theory . * from our results about the macroscopic loop amplitudes ( [ solutionsconclusion ] )
, one can conclude that every perturbatively isolated sector ( or perturbative string theory ) included in the @xmath1-cut fractional - superstring @xmath2 critical points is also realized as the perturbatively isolated sector in @xmath439-cut fractional - superstring @xmath2 critical points .
in particular , the fractional @xmath2 critical points of the infinite - cut two - matrix models includes all the perturbatively isolated sectors of @xmath2 minimal fractional superstring theory .
therefore , the infinite - cut two - matrix models can be understood as a _ mother theory _ of @xmath2 minimal fractional superstring theory , and the distribution of the isolated sectors can be interpreted as the landscape of @xmath2 minimal fractional superstring theory .
interestingly , in this limit , the angular direction @xmath51 discussed in the previous section becomes continuum coordinate , and the three - dimensional theory appears . *
appearance of the extra - third dimension in non - critical string theory does not contradict with the @xmath52 barrier @xcite .
it can be understood as follows : fractional superstring theory has two worldsheet bosons and parafermions : @xmath440 , @xmath441 , @xmath442 , @xmath443 .
although the ( matter ) central charge seems to exceed the @xmath444 bound , the exceeding central charge ( corresponding to that of parafermions @xmath442 and @xmath445 ) is gauged away by the fractional superconformal symmetry
. then we again have two - dimensional spacetime .
however , as we can see in the type 0 superstring theory , superstring theory remains @xmath101 discrete spacetime ( stokes sectors ) . in the same way ,
as is indicated in the multi - cut matrix models , general @xmath1-fractional superstring theory would have remaining @xmath0 discrete spacetime .
this extra discrete spacetime becomes the continuum third dimension in the @xmath46 limit . in this sense ,
the appearance of the third dimension has no contradiction with the barrier of @xmath52 in non - critical string theory ( or liouville theory ) . )
can only be observed by non - perturbative effects . ] * in the weak coupling regime @xmath287 , perturbative string theory description is favored and it is hard to see the neighborhood stokes sectors , i.e. extra dimension . in the center of the spectral curve @xmath446 ,
however , the theory is strong coupling and it is easy to see the extra dimension . in this sense , it is natural to ask the following question : _ what is the strong coupling dual theory of the @xmath2 minimal fractional superstring theory ? _
this strong coupling dual theory lives in the three - dimensional specetime , and controls all the perturbative @xmath2 minimal fractional superstring theories in the weak - coupling landscape .
therefore , it is natural to refer to this strong - coupling dual theory as _
minimal non - critical m theory_. whatever fundamental degree of freedom they have ( i.e. membrane or not ) , this three - dimensional theory is mathematically well defined and shares several features with the critical m theory .
* it is known that the multi - cut matrix models do not have good large @xmath123 expansion without fixing the filling fraction @xcite .
therefore , the strong - coupling dual theory should not have good expansion in terms of string coupling constant @xmath447 .
an interesting candidate for the strong coupling dual description is the kontsevich matrix models @xcite of fractional superstring theory because this matrix model does not rely on the string coupling topological expansion . * in this multi - cut matrix - model context , the direct connection to type 0a theory is not clearly understood : it is because the multi - cut two - matrix models are a natural generalization of type 0b superstring theory ( i.e. two - cut ) ; however the usual critical m theory is directly related to type iia superstrings . from the viewpoints of the orthonormal polynomial system ,
the difference of them is almost to put one eigenvalue on the origin ( 0a ) or not ( 0b ) , and this introduction of eigenvalue corresponds to introduction of the d0-brane @xcite ( or in this context kk modes of the third dimension @xcite ) .
therefore , it is interesting to consider such a 0a setting , and then to see to what extent the usual scenario from m theory to type iia superstrings can be realized in the multi - cut matrix models . *
our result indicates that each perturbative - string vacuum in fractional superstring theory is only a segment of the total non - perturbative theory . from the strong - coupling point of view
, the non - perturbative vacuum of this string theory is a superposition of perturbative vacua .
that is , amplitudes of full theory should include information of different perturbative - string vacua .
this suggests that it is these non - perturbative vacua which becomes important in the string landscape .
therefore , it is interesting to consider the following questions : _ what is the dynamical aspects of non - perturbative vacua ? _ and _ what is the general framework to control these vacua ? _ the multi - cut matrix models should play important roles to answer these questions .
although the discussion in this section relies on analogies and guessworks , we hope that this gives a new direction of non - critical string theory , a solvable and non - perturbative toy model for string landscape and possibly for m theory .
* acknowledgment * the authors would like to thank sheng - yu darren shih for collaboration and various discussions in the early stage of this work .
we would also like to thank bertrand eynard , kazuyuki furuuchi , pei - ming ho , hiroshi isono , shoichi kawamoto , ivan kostov , so okada , tomohisa takimi , tomohide terasoma , dan tomino , wen - yu wen and chen - pin yeh for useful discussions and comments .
hi would like to thank pei - ming ho for encouragement toward the study of `` matrix landscape , '' which partially led him to this work , and also every people in ist , ipht cea - saclay and ih ' es for their hospitality on his visit where part of the work has been done , especially to gabriel lopes cardoso , maxim kontsevich , ivan kostov , didina serban , shigenori seki and futoshi yagi .
ct is supported by national science council of taiwan under the contract no .
96 - 2112-m-021 - 002-my3 .
the authors are also supported by national center for theoretical science under nsc no .
98 - 2119-m-002 - 001 .
here we show some examples of the matrix realizations given in ( [ matrealcomplexcases ] ) .
that is , several examples of @xmath448 ( @xmath449 for the @xmath1-cut cases ) in the realization .
they also include other kinds of realizations which can not be written as ( [ matrixrealizationcomplex ] ) : the three - cut cases : @xmath450 the four - cut cases have sinh and cosh solutions : @xmath451 the five - cut case is given as @xmath452 or also @xmath453 below the matrices @xmath448 in the four - cut case with real potentials are shown : @xmath454 these are the basic realizations of our solutions and the other realization should be related to the basic realization ( [ matrixrealizationsinh ] ) by some proper similarity transformation : @xmath455 which commutes with @xmath11 and can give a real matrix @xmath456 .
in this section , we consider possible implications of the relation ( [ connectionrelation ] ) on the free - fermion formulation @xcite . as it has been pointed out in @xcite ,
_ different eigenvalues _ of the lax operators @xmath9 correspond to macroscopic loop amplitudes of fzzt - branes with _
different charges_. the meaning of this statement is following : in the free - fermion formulation , the charged fzzt branes are identified with the @xmath1-component free fermions @xmath457 @xmath277 .
it is because the @xmath0 charge is measured by the following fermion - number operator : @xmath458 where @xmath459 is the zero - th boson oscillator of the fermion @xmath460 . in this sense
, the macroscopic loop amplitudes @xmath461 of this charged fzzt brane should be given by the expectation value of the bosonization operator : @xmath462 on the other hand , it was shown in @xcite that each baker - akhiezer function ( i.e. eigenfunction ) @xmath463 of eq .
( [ bakakhstokes ] ) is written by the free - fermion operator @xmath457 of multi - component kp hierarchy in the following way : @xmath464 since difference of index @xmath199 in @xmath465 is difference of background flux and negligible in the week coupling limit @xcite , the behavior is uniformly given as @xmath466\ , \chi^{(j)}(t;\zeta),\end{aligned}\ ] ] in the week coupling limit @xmath467 .
therefore , the macroscopic loop amplitudes , @xmath461 , of fzzt brane of a charge @xmath48 are identified with the eigenvalues of the lax operators as @xmath468 however , since we now know that the baker - akhiezer functions are connected to each other , one should say that each fermion operators should be restricted in each stokes sector : @xmath469 otherwise the fermion operator should be renamed : @xmath470 in the non - perturbative sense .
this means that there is only one ( non - perturbative ) fundamental fzzt brane ( free fermion ) , @xmath471 , in this three - dimensional spacetime and that the the charge @xmath51 of fzzt branes is just a coordinate which measures which stokes sector the fzzt brane exists : @xmath472 .
this is quite natural because originally there is only one kind of eigenvalues in the matrix models and the position of eigenvalues ( which fermi sea the eigenvalue exists ) is interpreted as the charge of d - brane @xcite .
finally , we note an interesting expression of the baker - akhiezer function @xmath463 . according to eq .
( [ bakakhfermion ] ) , @xmath473 component gives simple one - point function of the fzzt brane with charge @xmath48 : @xmath474 .
this means that one has the following expression : @xmath475 in the sense of eq .
( [ changeofname ] ) .
therefore , amplitudes of fzzt brane with a charge @xmath48 are obtained by properly analytic continuing the @xmath48-th component of the orthonormal polynomials @xmath367 .
this is the observation given in @xcite .
a. m. polyakov , `` quantum geometry of bosonic strings , '' phys .
b * 103 * ( 1981 ) 207 ; `` quantum geometry of fermionic strings , '' phys .
b * 103 * ( 1981 ) 211 .
v. g. knizhnik , a. m. polyakov and a. b. zamolodchikov , `` fractal structure of 2d - quantum gravity , '' mod .
lett . a * 3 * ( 1988 ) 819 .
f. david , `` conformal field theories coupled to 2-d gravity in the conformal gauge , '' mod . phys .
a * 3 * ( 1988 ) 1651 ; + j. distler and h. kawai , `` conformal field theory and 2-d quantum gravity , or who s afraid of joseph liouville ?
, '' nucl .
b * 321 * ( 1989 ) 509 .
h. dorn and h. j. otto , `` two and three point functions in liouville theory , '' nucl .
b * 429 * ( 1994 ) 375 [ arxiv : hep - th/9403141 ] ; + a. b. zamolodchikov and al .
b. zamolodchikov , `` structure constants and conformal bootstrap in liouville field theory , '' nucl .
b * 477 * ( 1996 ) 577 [ arxiv : hep - th/9506136 ] .
j. teschner , `` on the liouville three point function , '' phys .
b * 363 * ( 1995 ) 65 [ arxiv : hep - th/9507109 ] .
v. fateev , a. b. zamolodchikov and al .
b. zamolodchikov , `` boundary liouville field theory .
i : boundary state and boundary two - point function , '' arxiv : hep - th/0001012 ; + j. teschner , `` remarks on liouville theory with boundary , '' arxiv : hep - th/0009138 .
a. b. zamolodchikov and al .
b. zamolodchikov , `` liouville field theory on a pseudosphere , '' arxiv : hep - th/0101152 .
r. h. poghosian , `` structure constants in the @xmath476 super - liouville field theory , '' nucl .
b * 496 * ( 1997 ) 451 [ arxiv : hep - th/9607120 ] ; + r. c. rashkov and m. stanishkov , `` three - point correlation functions in @xmath477 super liouville theory , '' phys .
b * 380 * ( 1996 ) 49 [ arxiv : hep - th/9602148 ] .
t. fukuda and k. hosomichi , `` super liouville theory with boundary , '' nucl .
b * 635 * ( 2002 ) 215 [ arxiv : hep - th/0202032 ] ; + c. ahn , c. rim and m. stanishkov , `` exact one - point function of @xmath476 super - liouville theory with boundary , '' nucl .
b * 636 * ( 2002 ) 497 [ arxiv : hep - th/0202043 ] .
e. brezin and v. a. kazakov ,
`` exactly solvable field theories of closed strings , '' phys .
b * 236 * ( 1990 ) 144 ; + m. r. douglas and s. h. shenker , nucl .
b * 335 * ( 1990 ) 635 ; + d. j. gross and a. a. migdal , `` nonperturbative two - dimensional quantum gravity , '' phys .
* 64 * ( 1990 ) 127 .
e. brezin , m. r. douglas , v. kazakov and s. h. shenker , `` the ising model coupled to 2-d gravity : a nonperturbative analysis , '' phys .
b * 237 * ( 1990 ) 43 ; + d. j. gross and a. a. migdal , `` nonperturbative solution of the ising model on a random surface , '' phys .
* 64 * ( 1990 ) 717 .
d. j. gross and a. a. migdal , `` a nonperturbative treatment of two - dimensional quantum gravity , '' nucl .
b * 340 * ( 1990 ) 333 .
m. fukuma , h. kawai and r. nakayama , `` continuum schwinger - dyson equations and universal structures in two - dimensional quantum gravity , '' int .
j. mod .
a * 6 * ( 1991 ) 1385 ; `` infinite dimensional grassmannian structure of two - dimensional quantum gravity , '' commun .
phys . * 143 * ( 1992 ) 371 ; `` explicit solution for @xmath478@xmath479 duality in two - dimensional quantum gravity , '' commun .
* 148 * ( 1992 ) 101 .
m. fukuma and s. yahikozawa , `` nonperturbative effects in noncritical strings with soliton backgrounds , '' phys . lett .
b * 396 * ( 1997 ) 97 [ arxiv : hep - th/9609210 ] ; `` combinatorics of solitons in noncritical string theory , '' phys .
b * 393 * ( 1997 ) 316 [ arxiv : hep - th/9610199 ] ; m. fukuma and s. yahikozawa , `` comments on d - instantons in c @xmath480 1 strings , '' phys .
b * 460 * ( 1999 ) 71 [ arxiv : hep - th/9902169 ] .
s. h. shenker , `` the strength of nonperturbative effects in string theory , '' j. polchinski , `` combinatorics of boundaries in string theory , '' phys .
d * 50 * ( 1994 ) 6041 [ arxiv : hep - th/9407031 ] .
j. mcgreevy and h. l. verlinde , `` strings from tachyons : the c = 1 matrix reloaded , '' jhep * 0312 * ( 2003 ) 054 [ arxiv : hep - th/0304224 ] .
d. gaiotto and l. rastelli , `` a paradigm of open / closed duality : liouville d - branes and the kontsevich model , '' jhep * 0507 * ( 2005 ) 053 [ arxiv : hep - th/0312196 ] .
d. j. gross and e. witten , `` possible third order phase transition in the large n lattice gauge theory , '' phys .
d * 21 * ( 1980 ) 446 .
v. periwal and d. shevitz , `` unitary matrix models as exactly solvable string theories , '' phys
. rev .
* 64 * ( 1990 ) 1326 ; `` exactly solvable unitary matrix models : multicritical potentials and correlations , '' nucl .
b * 344 * ( 1990 ) 731 .
m. r. douglas , n. seiberg and s. h. shenker , `` flow and instability in quantum gravity , '' phys .
b * 244 * ( 1990 ) 381 . c. r. nappi ,
`` painleve - ii and odd polynomials , '' mod . phys .
a * 5 * ( 1990 ) 2773 .
c. crnkovic , m. r. douglas and g. w. moore , `` loop equations and the topological phase of multi - cut matrix models , '' int .
j. mod .
phys . a * 7 * ( 1992 ) 7693 [ arxiv : hep - th/9108014 ] .
t. j. hollowood , l. miramontes , a. pasquinucci and c. nappi , `` hermitian versus anti - hermitian one matrix models and their hierarchies , '' nucl .
b * 373 * ( 1992 ) 247 [ arxiv : hep - th/9109046 ] .
t. takayanagi and n. toumbas , `` a matrix model dual of type 0b string theory in two dimensions , '' jhep * 0307 * ( 2003 ) 064 [ arxiv : hep - th/0307083 ] .
m. r. douglas , i. r. klebanov , d. kutasov , j. m. maldacena , e. j. martinec and n. seiberg , `` a new hat for the @xmath481 matrix model , '' arxiv : hep - th/0307195
. i. r. klebanov , j. m. maldacena and n. seiberg , `` unitary and complex matrix models as 1-d type 0 strings , '' commun .
* 252 * ( 2004 ) 275 [ arxiv : hep - th/0309168 ] .
h. irie , `` fractional supersymmetric liouville theory and the multi - cut matrix models , '' nucl .
b * 819 * ( 2009 ) 351 [ arxiv:0902.1676 [ hep - th ] ] . c. crnkovic and g. w. moore , `` multicritical multicut matrix models , '' phys .
b * 257 * ( 1991 ) 322 .
p. c. argyres , a. leclair and s. h. h. tye , `` on the possibility of fractional superstrings , ''
b * 253 * ( 1991 ) 306 .
a. b. zamolodchikov and v. a. fateev , `` parafermionic currents in the two - dimensional conformal quantum field theory and selfdual critical points in z(n ) invariant statistical systems , '' sov .
jetp * 62 * ( 1985 ) 215 [ zh .
* 89 * ( 1985 ) 380 ] .
m. fukuma and h. irie , `` a string field theoretical description of @xmath241 minimal superstrings , '' jhep * 0701 * ( 2007 ) 037 [ arxiv : hep - th/0611045 ] . m. sato , rims kokyuroku * 439 * ( 1981 ) 30 ; + e. date , m. jimbo , m. kashiwara and t. miwa , `` transformation groups for soliton equations .
operator approach to the kadomtsev - petviashvili equation , '' rims-358 ; + m. jimbo and t. miwa , `` solitons and infinite dimensional lie algebras , '' publ .
kyoto * 19 * ( 1983 ) 943 ; + v. g. kac and j. w. van de leur , `` the @xmath482-component kp hierarchy and representation theory , '' j. math .
* 44 * ( 2003 ) 3245 [ arxiv : hep - th/9308137 ] .
c. t. chan , h. irie , s. y. shih and c. h. yeh , `` macroscopic loop amplitudes in the multi - cut two - matrix models , '' nucl .
b * 828 * ( 2010 ) 536 [ arxiv:0909.1197 [ hep - th ] ] .
e. brezin , c. itzykson , g. parisi and j. b. zuber , `` planar diagrams , '' commun .
* 59 * ( 1978 ) 35 .
v. a. kazakov , `` the appearance of matter fields from quantum fluctuations of 2d gravity , '' mod .
lett . a * 4 * ( 1989 ) 2125 .
i. k. kostov , `` strings embedded in dynkin diagrams , '' cargese 1990 , proceedings , random surfaces and quantum gravity , pp.135 - 149
. i. k. kostov , `` loop amplitudes for nonrational string theories , '' phys .
b * 266 * ( 1991 ) 317 .
i. k. kostov , `` strings with discrete target space , '' nucl .
b * 376 * ( 1992 ) 539 [ arxiv : hep - th/9112059 ] .
t. banks , m. r. douglas , n. seiberg and s. h. shenker , `` microscopic and macroscopic loops in nonperturbative two - dimensional gravity , '' phys .
b * 238 * ( 1990 ) 279 .
g. w. moore , n. seiberg and m. staudacher , `` from loops to states in 2-d quantum gravity , '' nucl .
b * 362 * ( 1991 ) 665 .
j. m. daul , v. a. kazakov and i. k. kostov , `` rational theories of 2-d gravity from the two matrix model , '' nucl .
b * 409 * ( 1993 ) 311 [ arxiv : hep - th/9303093 ] . c. kristjansen , `` multiloop correlators for rational theories of 2-d gravity from the generalized kontsevich models , '' nucl .
b * 436 * ( 1995 ) 342 [ arxiv : hep - th/9409066 ] .
m. anazawa , a. ishikawa and h. itoyama , `` universal annulus amplitude from the two matrix model , '' phys .
d * 52 * ( 1995 ) 6016 [ arxiv : hep - th/9410015 ] .
m. anazawa , a. ishikawa and h. itoyama , `` macroscopic three loop amplitudes from the two matrix model , '' phys .
b * 362 * ( 1995 ) 59 [ arxiv : hep - th/9508009 ] .
m. anazawa and h. itoyama , `` macroscopic @xmath482-loop amplitude for minimal models coupled to two - dimensional gravity : fusion rules and interactions , '' nucl .
b * 471 * ( 1996 ) 334 [ arxiv : hep - th/9511220 ] .
f. david , `` phases of the large n matrix model and nonperturbative effects in 2-d gravity , '' nucl . phys .
b * 348 * ( 1991 ) 507 ; nonperturbative effects in matrix models and vacua of two - dimensional phys .
b * 302 * ( 1993 ) 403 [ arxiv : hep - th/9212106 ] .
v. a. kazakov and i. k. kostov , `` instantons in non - critical strings from the two - matrix model , '' arxiv : hep - th/0403152 . n. seiberg and d. shih , `` branes , rings and matrix models in minimal ( super)string theory , '' jhep * 0402 * ( 2004 ) 021 [ arxiv : hep - th/0312170 ] . j. m. maldacena , g. w. moore , n. seiberg and d. shih , `` exact vs. semiclassical target space of the minimal string , '' jhep * 0410 * ( 2004 ) 020 [ arxiv : hep - th/0408039 ] . n. seiberg and d. shih , `` flux vacua and branes of the minimal superstring , '' jhep * 0501 * ( 2005 ) 055 [ arxiv : hep - th/0412315 ] .
d. kutasov , k. okuyama , j. w. park , n. seiberg and d. shih , `` annulus amplitudes and zz branes in minimal string theory , '' jhep * 0408 * ( 2004 ) 026 [ arxiv : hep - th/0406030 ] .
m. fukuma , h. irie and y. matsuo , `` notes on the algebraic curves in @xmath241 minimal string theory , '' jhep * 0609 * ( 2006 ) 075 [ arxiv : hep - th/0602274 ] .
p. horava and c. a. keeler , `` noncritical m - theory in 2 + 1 dimensions as a nonrelativistic fermi liquid , '' jhep * 0707 * ( 2007 ) 059 [ arxiv : hep - th/0508024 ] .
m. l. mehta , `` a method of integration over matrix variables , '' commun .
phys . * 79 * , 327 ( 1981 ) .
t. tada and m. yamaguchi , `` @xmath483 and @xmath484 operator analysis for two matrix model , '' phys .
b * 250 * ( 1990 ) 38 ; + m. r. douglas , `` the two matrix model , '' _ in * cargese 1990 , proceedings , random surfaces and quantum gravity * 77 - 83 .
( see high energy physics index 30 ( 1992 ) no .
17911 ) _ ; + t. tada , `` @xmath485 critical point from two matrix models , '' phys .
b * 259 * ( 1991 ) 442 .
m. r. douglas , `` strings in less than one - dimension and the generalized kdv hierarchies , '' phys .
b * 238 * ( 1990 ) 176 .
g. w. moore , `` geometry of the string equations , '' commun .
* 133 * ( 1990 ) 261 ; `` matrix models of 2-d gravity and isomonodromic deformation , '' prog .
* 102 * ( 1990 ) 255 .
b. eynard and j. zinn - justin , `` large order behavior of 2-d gravity coupled to @xmath486 matter , '' phys .
b * 302 * ( 1993 ) 396 [ arxiv : hep - th/9301004 ] .
h. irie , `` notes on d - branes and dualities in @xmath241 minimal superstring theory , '' nucl . phys .
b * 794 * [ pm ] ( 2008 ) 402 [ arxiv:0706.4471 hep - th ] .
p. l. nash , `` chebyshev polynomials and quadratic path integrals , '' j. math .
phys . 27 ( 1986 ) 2963 .
b. eynard , `` large n expansion of the 2-matrix model , '' jhep * 0301 * ( 2003 ) 051 [ arxiv : hep - th/0210047 ] .
a. r. its and v. y. novokshenov , `` the isomonodromic deformation method in the theory of painlev ' e equations , '' springer - verlag ( 1986 ) ; + p. bleher and a. its , `` double scaling limit in the random matrix model : the riemann - hilbert approach , '' arxiv : math - ph/0201003 .
m. fukuma , h. irie and s. seki , `` comments on the d - instanton calculus in @xmath487 minimal string theory , '' nucl .
b * 728 * ( 2005 ) 67 [ arxiv : hep - th/0505253 ] .
t. kaluza , `` on the problem of unity in physics , '' sitzungsber .
berlin ( math .
* 1921 * ( 1921 ) 966 ; + o. klein , `` quantum theory and five - dimensional theory of relativity , '' z. phys . *
37 * ( 1926 ) 895 [ surveys high energ .
phys . * 5 * ( 1986 ) 241 ] .
g. bonnet , f. david and b. eynard , `` breakdown of universality in multi - cut matrix models , '' j. phys .
a * 33 * ( 2000 ) 6739 [ arxiv : cond - mat/0003324 ] .
p. h. ginsparg and j. zinn - justin , `` action principle and large order behavior of nonperturbative gravity , '' p.
di francesco and d. kutasov , `` world sheet and space time physics in two dimensional ( super ) string theory , '' nucl .
b * 375 * ( 1992 ) 119 [ arxiv : hep - th/9109005 ] .
i. r. klebanov , `` string theory in two - dimensions , '' arxiv : hep - th/9108019 . b. eynard , `` topological expansion for the 1-hermitian matrix model correlation functions , '' jhep * 0411 * ( 2004 ) 031 [ arxiv : hep - th/0407261 ] ; + b. eynard and n. orantin , `` invariants of algebraic curves and topological expansion , '' arxiv : math - ph/0702045 ; `` geometrical interpretation of the topological recursion , and integrable string theories , '' arxiv:0911.5096 [ math - ph ] . m. kontsevich , `` intersection theory on the moduli space of curves and the matrix airy function , '' commun .
* 147 * ( 1992 ) 1 . | multi - cut two - matrix models are studied in the @xmath0 symmetry breaking @xmath1-cut @xmath2 critical points which should correspond to @xmath2 minimal @xmath1-fractional superstring theory .
fzzt - brane or macroscopic
loop amplitudes are obtained in all of these critical points and found to have two kinds of solutions in general .
each of these solutions is expressed by hyperbolic cosine or sine functions with proper phase shifts .
the algebraic geometries and zz - brane disk amplitudes ( instanton actions ) of these solutions are also studied . in particular , our results suggest that minimal @xmath3-fractional superstring theory can be viewed as a mother theory which includes all the minimal @xmath1-fractional superstring theories @xmath4 as its perturbative vacua in the weak - coupling string landscape .
our results also indicate that , in the strong coupling regime of this fractional superstring theory , there is a three - dimensional theory which would be understood as the non - critical version of m theory in the sense proposed by p. hoava and c. a. keeler .
arxiv:1003.1626 + march 2010 * fractional - superstring amplitudes , + multi - cut matrix models and non - critical m theory * chuan - tsung chan@xmath5 , hirotaka irie@xmath6 and chi - hsien yeh@xmath6 _ @xmath7department of physics , tunghai university , taiwan , 40704 + @xmath8department of physics and center for theoretical sciences , + national taiwan university , taipei 10617 , taiwan , r.o.c _ |
the potential realization of majorana zero modes ( mzms ) at the ends of a nanowire - superconductor hybrid system @xcite has attracted broad interest to different ways of stabilizing mzms .
while there are proposals to exploit exotic statistics of mzms within quasi - one - dimensional networks @xcite , a two dimensional setting would be desirable for observing statistical properties of mzms .
a mzm can appear as a vortex bound state of triplet superfluids @xcite or superconductors @xcite .
unfortunately , naturally occurring triplet superconductors are rare , and hence the proposal by fu and kane @xcite to use the superconducting proximity effect on the topological insulator ( ti ) surface states raised enthusiasm as an alternative route to realizing mzms hosted in a two dimensional space .
however , most known three - dimensional ( 3d ) ti candidate materials , such as bi@xmath0se@xmath1 and bi@xmath0te@xmath1 , have both the surface states and the bulk states at the fermi energy @xcite .
recent experimental successes in inducing superconductivity in bi@xmath0se@xmath1 thin films through proximity effect @xcite makes it all the more urgent to address the superconducting proximity effect in such topological _ metals _ , where surface states and bulk states coexist . in the proposal by fu and kane @xcite for realizing mzms ,
superconductivity is induced to the surface states of a 3d ti by proximity to a trivial @xmath2-wave superconductor ( sc ) .
the argument for the existence of a mzm as a vortex bound state is based on the formal equivalence between a @xmath3 superconducting gap of a spinless fermion and a trivial @xmath2-wave gap after projection to the space of surface states . however , with only the surface states available at the fermi energy , the superconducting proximity effect is limited to the interface between the ti and the adjacent superconductor . on the other hand when the bulk band crosses the fermi energy , as they do in many 3d ti materials ,
there is a chance that the proximity effect can reach the naked surface .
the key questions then would be ( 1 ) when can proximity effect reach the naked surface and ( 2 ) whether the naked surface can host mzms .
these questions are the focus of this paper .
to be concrete , we consider a bi@xmath0se@xmath1-sc heterostructure , where the bi@xmath0se@xmath1 takes the form of a finite thickness slab , so that we can study its naked surface [ fig .
[ fig : heterostructure ] ] .
we first study how the proximity effect propagates differently depending on the location of the chemical potential , by solving the bogoliubov - de gennes ( bdg ) equation in the heterostructure .
we then study the vortex bound state spectra with the gap structure inferred from the solution and investigate the stability of a mzm on the naked surface depending on chemical potential .
the heterostructure of interest consists of a slab of bi@xmath0se@xmath1 for @xmath4 and superconductor for @xmath5 .
the electronic structure of bi@xmath0se@xmath1 is described by an effective two - orbital hamiltonian on a simple cubic lattice with lattice constant @xmath6 . given the slab geometry with periodic boundary conditions in the @xmath7 and @xmath8 directions , we choose as basis @xmath9 , a state with momentum @xmath10 within an @xmath11 plane at @xmath12 for @xmath13 , with orbital @xmath14 and spin @xmath2 .
as the normal - state hamiltonian of the model we take a lattice version of the four - band continuum model for 3d ti as given in ref .
@xcite consisting of two parts : intra - layer terms @xmath15 and the inter - layer hopping ( from @xmath16 to @xmath17 ) terms @xmath18 written as @xmath19 \hat{\tau}_z \nonumber\\ & + \lambda \sin ( k_y a ) \hat{\tau}_x \hat{\sigma}_x - \lambda \sin ( k_x a ) \hat{\tau}_x \hat{\sigma}_y \nonumber \\
\hat{h}_\bfk^{(1 ) } = & - t_2 - m_2 \hat{\tau}_z - i \frac{\lambda'}{2 } \hat{\tau}_y \label{eq : hamiltonian}\end{aligned}\]]where @xmath20(@xmath21 ) for @xmath22 are pauli matrices in the orbital ( spin ) space . the parameters of the hamiltonian in eq
. are chosen such that the model matches the continuum model for bi@xmath0se@xmath1 from ref .
@xcite up to @xmath23 for @xmath24 : @xmath25 ev , @xmath26 ev , @xmath27 ev , @xmath28 ev , @xmath29 ev , @xmath30 ev , and @xmath31 ev . the reference chemical potential @xmath32ev has been chosen such that the degeneracy point of the surface state branch lies at @xmath33 when @xmath34 . to explicitly define what we mean by a topological metal ( tm )
it is important to recall the well - known band structure of the above model . as shown in fig .
[ fig : bandstructure ] , the spectrum of the hamiltonian contains a ( degenerate ) gapless branch in addition to the bulk states separated by a finite gap . depending on the chemical potential , we now define three regimes : topological insulator ( ti ) , tm , and metal ( m ) .
the ti is a bulk insulating state with the chemical potential within the bulk band gap [ fig .
[ fig : bandstructure ] , @xmath35mev ] . in the ti regime ,
gapless states at the fermi level are highly localized at the two surfaces of the slab . on the other hand ,
when the chemical potential is well within the bulk conduction band , all the states at the fermi level , including the ones from the branch that contains surface states in the ti regime , are extended over the entire slab [ fig .
[ fig : bandstructure ] , @xmath36mev ] . here
, we refer to this regime as metal ( m ) . in between these two regimes
, there is a range of chemical potential where the branch that is an extension of the dirac cone coexists with the bulk states at the fermi level , but nevertheless it remains surface - localized and spin - momentum locked [ fig . [
fig : bandstructure ] , @xmath37mev ] .
experimentally , this regime can be identified through the spin - momentum locking of dirac - cone states outside the bulk band - gap , which has been observed in bi@xmath0se@xmath1 by spin - angle - resolved photoemission spectroscopy ( arpes ) @xcite .
we refer to this regime as topological metal @xcite .
note that while the existence of the in - gap surface states is protected by topology , its dispersion depends on material specific details .
therefore , the exact ranges of chemical potential of the three regimes will also be material dependent . nevertheless , the surface states and the bulk states have qualitatively different contributions to the proximity effect as we will see below , and therefore we expect the three regimes in a real material to show qualitatively the same features as the corresponding regimes in our calculation .
for the superconductor part ( @xmath38 ) we again use a two - orbital model of the same form as eq . to describe its normal state , with @xmath39 for @xmath40 .
the same parameters as bi@xmath0se@xmath1 are used , except that we flip the sign of the `` mass term '' ( @xmath41 ) and make the resulting band structure trivial , by choosing @xmath42 .
also , since the inter - layer hopping in both parts of the heterostructure is described by the same term @xmath18 , we use it to describe the tunneling between the two parts .
in order to compare the proximity effect in the three regimes , we impose an orbital - independent @xmath2-wave superconducting gap of strength @xmath43 on the superconductor ( @xmath38 ) and diagonalize the bdg hamiltonian .
we then study how the resulting pair amplitude depends on the distance from the interface . because the pair amplitude is a matrix in both the spin and the orbital basis , it is convenient to look at its projection onto different spin channels . as pointed out in ref .
@xcite , spin - singlet @xmath44 pairing term induces spin - singlet @xmath44 and spin - triplet @xmath45 components of the pair amplitude matrix in the presence of spin - orbit coupling of the form eq . .
the spin singlet and triplet components @xmath46 and @xmath47 are themselves @xmath48 matrices in the orbital space , given by @xmath49_{s_1 s_2 } u_{\bfk z \alpha s_1}^{\phantom { * } } v_{\bfk z \beta s_2}^ * , \label{eq : f}\end{aligned}\ ] ] where @xmath50 is the number of @xmath51 points in the @xmath11-plane and the sum is over every positive - energy bdg eigenstate @xmath52 . in eq .
@xmath53 and @xmath54 are the respective form factors for spin - singlet and triplet defined by @xmath55 with @xmath56 the ( @xmath48 ) identity matrix . in the self - consistent approach with attractive interaction @xmath57 in the bcs channel ,
the superconducting gap @xmath58 is proportional to the pair amplitude ( @xmath59 ) . here
, however , no such self - consistency is imposed , and the pair amplitude inside the bi@xmath0se@xmath1 is completely due to the andreev reflection from the interface @xcite .
we study the @xmath60-dependence of the pair amplitudes in bi@xmath0se@xmath1 side ( @xmath61 ) in the three regimes : ti , m , and tm .
for this purpose , we pick for each @xmath60 in each spin channel the largest eigenvalue @xmath62 of the @xmath63 matrix @xmath64 , which indicates the leading instability in the given spin channel . in all three regimes ,
both spin - singlet and spin - triplet pair amplitudes are expected to be non - zero because of the spin - orbit coupling term in the hamiltonian . in fig .
[ fig : pairamplitude ] , we plot @xmath62 as a function of @xmath60 . in the ti regime [ fig .
[ fig : pairamplitude - a ] ] , we find that the pair amplitude is confined to the buried interface with exponential decay , since it is carried entirely by the surface states with such spatial profile .
in addition , singlet and triplet components of the pair amplitude have the same magnitude as a result of spin - momentum locking of the surface states . in the m regime [
[ fig : pairamplitude - b ] ] , on the other hand , the pair amplitudes show friedel oscillations with an envelop that decays algebraically as a function of @xmath60 .
( see the supplemental material for an analytic understanding of the @xmath60 dependence of the pair amplitudes in the m regime . ) in addition , the singlet channel dominates over the triplet channel in the m regime .
the results in the tm regime [ fig .
[ fig : pairamplitude - b ] ] can be understood by combining the two pictures of the ti and the m regimes . in the tm regime ,
the pair amplitude consists of two components : the surface - states contribution and the bulk - states contribution , each of which should be qualitatively the same as the pair amplitude in the ti and the m regimes , respectively . at large distances where the bulk - states contribution is dominant
, the pair amplitude should show a power - law - like decay .
friedel oscillation should also be present in principle , but in fig .
[ fig : pairamplitude - c ] , the large wavelength of the oscillation makes it difficult to identify the oscillation . with the power - law decay of the pair amplitude at large distances , superconductivity can be induced on the naked surface by proximity effect in the tm .
this induced pairing on the naked surface is a mixture of singlet and triplet components .
the two components , however , lead to the identical effective bdg hamiltonian for the surface states , as the surface states are fully spin - momentum locked .
+ next , we ask whether the naked surface of a tm with proximity - induced superconductivity can host mzms . formally related to the system of our interest is the 3d bulk superconducting cu - doped bi@xmath0se@xmath1 .
for this system hosur _
et al . _
@xcite predicted a vortex parallel to the @xmath65-axis to host a surface mzm even when the chemical potential is within the bulk conduction band , as long as it is below a critical value of @xmath66 from the bottom of the band .
the chemical potential of an undoped bi@xmath0se@xmath1 falls within this range @xcite , and so does our definition of tm in our model .
hence a vortex in a tm proximity - coupled to a superconductor is likely to host a protected mzm at the naked surface .
however , the effect of @xmath60-axis - dependent proximity - induced pairing strength on the naked surface and energetic stability of the mzm are not known _ a priori_. for concreteness , we solve the bdg equation on a cylindrical slab of bi@xmath0se@xmath1 with thickness @xmath67 and radius @xmath68 , with chemical potential in the ti and tm regimes . with the axis of the cylinder aligned along the @xmath60 axis , we take the @xmath11-coordinates to be continuous , while keeping the @xmath60 coordinate discrete .
the normal state hamiltonian is then described by eq . , with @xmath69 and @xmath70 for @xmath71 .
informed by our proximity effect calculation above , we impose an @xmath2-wave superconducting gap of the following respective profiles for ti and tm : @xmath72 where @xmath73 is the cylindrical coordinate of the system . @xmath74 and @xmath75 are superconducting correlation lengths in the radial and the axial directions , respectively .
we chose @xmath76 such that the bottom - most layer ( @xmath77 ) of the ti / tm has a gap of magnitude @xmath43 , and a positive exponent @xmath78 is used for the gap profile to decay as @xmath60 increases .
because of the rotation symmetry of the system , it is convenient to use as basis the circular harmonics @xmath79 where @xmath80 is the bessel function of the first kind of order @xmath81 and @xmath82 is its @xmath83th zero . expressed in terms of @xmath84 , the hamiltonian can be block diagonalized into different sectors of @xmath85 , where @xmath86 and @xmath87 are orbital angular momentum and spin of a quasiparticle in the @xmath60 direction , and @xmath88 is its charge in units of @xmath89 ( @xmath90 for electron ) .
one can then diagonalize each block of the hamiltonian , and find the low energy eigenstates .
each eigenstate @xmath91 can be identified using its spatial probability density defined as @xmath92 figures [ fig : vbs_ti_lowest ] and [ fig : vbs_tm_lowest ] show @xmath93 of the lowest excitation in the ti and tm regimes . in the ti regime
, the superconducting gap decays exponentially away from the bottom surface , becoming negligible on the top surface . as a result a zero - energy vortex bound state appears only on the bottom surface , and the top surface remains metallic [ fig .
[ fig : vbs_ti_lowest ] ] .
the resulting spectrum is shown in the inset of fig .
[ fig : vbs_ti_lowest ] . in the tm regime , on the other hand , the superconducting gap at the top surface is sizable , and a well - defined majorana vortex bound state exists on both the top and the bottom surfaces .
hence the tm regime brings the best of both worlds : a stable zero mode on the experimentally accessible top surface .
trivially yields no zero mode , as expected . ]
in summary , we studied the proximity effect in _ topological metals _ , i.e. , topological insulators with bulk states at the fermi level coexisting with well - defined surface states exhibiting spin - momentum locking . against the common belief that ideal topological insulators should be bulk insulating , we showed that the existence of bulk carriers can be a feature for the proximity effect as the induced gap will be observable at the naked surface .
most importantly , we showed that a vortex line in a tm - sc structure will host an energetically stable majorana bound state at the naked surface .
although we focused on the proximity effect due to an @xmath2-wave superconductor for concreteness , our results are applicable to the proximity effect due to a @xmath94-wave superconductor such as the high-@xmath95 cuprates as long as the induced gap is dominantly @xmath2-wave .
in fact wang _
_ @xcite observed an isotropic gap opening on the dirac branch on a thin film of bi@xmath0se@xmath1 on a @xmath96 substrate below the superconducting transition temperature . while the mechanism for the larger value of the inferred surface - state gap compared to the bulk gap in ref .
@xcite remains unknown @xcite and the results of ref .
@xcite have not been reproduced to date @xcite , our results should apply as long as the induced isotropic gap is dominantly @xmath2-wave .
the setup of bi@xmath0se@xmath1 proximity coupled to superconducting nbse@xmath97 recently studied using arpes and point - contact transport in ref .
@xcite actually satisfy the condition of tm - sc structure as defined in this paper , according to their spin - momentum locking observations .
our results imply that the same system can support majorana bound states at vortex cores with spatial separation between the top ( naked ) surface majorana and the bottom ( buried ) surface majorana .
so far little attention has been given to experimentally distinguishing the two surfaces of ti in such a heterostructure , although ref .
@xcite showed how the spectral gap at the dirac point depends on the film thickness presumably due to varying degrees of coupling between the two surfaces .
one way to experimentally identify the surface would be to use arpes and look for the normal - state fermi surface of the substrate .
the dirac state signal probed simultaneously with the substrate will be coming from both the top surface and the interface .
when the film is thick enough to not show the substrate fermi surface , the dirac state signal will be coming from the naked top surface . in order to test our predictions
we propose in - field stm measurements looking for majorana bound states in a tm - sc setup like that of ref .
@xcite in which spin - momentum locking is confirmed , with further attention given to distinguishing signals from each surface .
* acknowledgements*. we thank z. hasan for discussions that motivated the work and h. yao for useful discussions .
k. lee , a. vaezi , and e .- a.k . were supported in part by nsf career grant no .
m.h.f . and
e .- a.k . were supported in part by nsf grant no .
dmr-1120296 to the cornell center for materials research .
33ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1126/science.1222360 [ * * , ( ) ] link:\doibase 10.1038/nphys2479 [ * * , ( ) ] link:\doibase 10.1038/nphys2429 [ * * , ( ) ] link:\doibase 10.1021/nl303758w [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.217005 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.126406 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.214509 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.077001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.177002 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.134508 [ * * , ( ) ] link:\doibase 10.1103/physrevb.44.9667 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/7/47/002 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.096407 [ * * , ( ) ] link:\doibase 10.1038/nphys1762 [ * * , ( ) ] link:\doibase 10.1126/science.1216466 [ * * , ( ) ] link:\doibase 10.1038/nphys2744 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.045122 [ * * , ( ) ] link:\doibase 10.1038/nature08234 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.245432 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.186403 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.104202 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.235114 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.205438 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.220506 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.25.4515 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.097001 [ * * , ( ) ] link:\doibase 10.1038/nphys1274 [ * * , ( ) ] @noop link:\doibase 10.1103/physrevlett.113.067003 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys3139 [ * * , ( ) ] | much interest in the superconducting proximity effect in three - dimensional ( 3d ) topological insulators ( tis ) has been driven by the potential to induce majorana bound states at the interface .
most candidate materials for 3d ti , however , are bulk metals , with bulk states at the fermi level coexisting with well - defined surface states exhibiting spin - momentum locking . in such topological metals ,
the proximity effect can differ qualitatively from that in tis . by studying a model topological metal - superconductor ( tm - sc ) heterostructure within the bogoliubov - de gennes formalism ,
we show that the pair amplitude reaches the naked surface , unlike in a topological insulator - superconductor ( ti - sc ) heterostructure where it is confined to the interface .
furthermore , we predict vortex - bound - state spectra to contain a majorana zero - mode localized at the naked surface , separated from the bulk vortex - bound - state spectra by a finite energy gap in such a tm - sc heterostructure .
these naked - surface - bound modes are amenable to experimental observation and manipulation , presenting advantages of tm - sc over ti - sc . |
with the successful start of collisions at 7 tev a few months ago , the large hadron collider has put another milestone towards a thorough exploration of the terascale . the large energy available
will open many multi - particle channels that have being widely investigated @xcite . in view of a correct interpretation of the signals of new physics which might be extracted from data , it is of considerable interest to reduce our theoretical uncertainty for the physical processes under study , especially when large qcd backgrounds are involved . in this respect
, the need of next - to - leading - order ( nlo ) corrections for the lhc is unquestionable .
many efforts have been made in the last few years in developing efficient techniques for one - loop calculations , using both highly - refined traditional methods and the new approaches of opp and generalized unitarity .
a significant achievement of this work has been the calculation of complete nlo predictions for several interesting @xmath1 processes at lhc @xcite . in this contribution
we describe the main features of the ` helac - nlo ` approach to nlo calculations .
we provide an overview of the computational framework and briefly summarize some physics results obtained with this system .
the calculation of the virtual corrections is organized in the framework of the opp method .
any one - loop amplitude is characterized by a set of denominators given as products of loop propagators , @xmath2 , and a polynomial in the loop momentum function called numerator , @xmath3 , such that it can be represented in the form @xmath4 where both the loop momentum @xmath5 and the numerator function are considered to be evaluated in @xmath6 dimensions . on the other hand , it is well known that any one - loop amplitude admits a decomposition in terms of a basis of @xmath6-dimensional scalar integrals . in the limit @xmath7
, this decomposition can be cast into the form @xmath8 where box , triangle , bubble and tadpole denote the known scalar loop functions and @xmath9 is a rational function in the invariants also known as rational term . the basic idea of opp is to decompose the numerator of the integrand function , @xmath10 , into a basis of monomials of the propagators appearing in the denominator : the coefficients of the expansion are the @xmath11 appearing in eq.[opp ] plus additional , @xmath12-dependent , coefficients also known as spurious terms , whose analytical dependence on the loop momentum is known .
it is therefore possible to evaluate the coefficients of the expansion by computing @xmath10 a sufficient number of times ( equal the number of coefficients to be determined ) , by appropriate choices of @xmath12 .
one possibility is to select those values of the loop momentum that nullify certain sets of propagators the so - called unitarity cuts and solve the resulting linear system of equations in terms of the expansion coefficients .
if the numerator @xmath10 is considered in @xmath13 dimensions , it is necessary to decompose the rational terms , @xmath14 : we call @xmath15 the part of the rational terms that is derived through the reduction starting with a @xmath13 numerator function .
@xmath16 is the part of the rational terms that originates from the explicit @xmath17 dependence of the numerator function evaluated in @xmath6 dimensions .
as has been proven in @xcite @xmath16 can be fully determined by including new interaction - vertices with up to four legs , in full similarity with the usual uv counter - terms .
on the one hand the method can be fully implemented numerically , resulting to a very fast reduction of arbitrary multi - leg amplitudes . on the other hand , being inherently algebraic , provides exact results ( with arbitrary precision ) for the coefficients multiplying the one - loop scalar functions as well as for the rational terms .
the opp method thus provides a systematic framework to evaluate the cut - constructible part of the amplitude as well as the @xmath15 term , given a numerator function as input ( see @xcite for details ) . in the approach we follow ` helac-1loop ` @xcite , based on ` helac - phegas ` @xcite , automatically generates all the input numerators for the opp reduction as well as all tree - order - like contributions including the new vertices needed for the full determination of the rational part ( @xmath16 ) @xcite , as well as uv - renormalization counter - terms . for a detailed description of the algorithm
we refer to @xcite .
to summarize , our approach for the evaluation of one - loop amplitudes relies upon three basic elements : 1 ) an implementation of the opp reduction ; 2 ) a library of scalar loop integrals ; 3 ) an algorithm to automatically evaluate numerator functions and @xmath16 terms .
these tasks are solved respectively by the codes ` cuttools ` , ` oneloop ` and ` helac-1loop ` @xcite .
we organize the calculation of virtual corrections using a re - weighting technique and adopt helicity and colour sampling methods in order to optimize the performance of our system @xcite .
the treatment of the qcd colour is based on the colour connection representation .
given a generic qcd amplitude , @xmath18 , composed by @xmath19 gluons and @xmath20 quarks ( antiquarks ) with colour indices @xmath21 and @xmath22 ( @xmath23 ) respectively , it is possible to get a uniform representation for quarks and gluons by contracting each gluon index @xmath21 with a corresponding @xmath24 : @xmath25 the resulting amplitude can be now decomposed in terms of colour - stripped partial amplitudes @xmath26 : @xmath27 with @xmath28 , and @xmath29 denoting a permutation of the set @xmath30 .
each @xmath26 is interpreted as the contribution to the amplitude associated to one possible way of connecting the external particles through colour flow , the connection being established by the indices of the @xmath31 s .
the feynman rules that allow the calculation of all @xmath26 s have been described in refs .
@xcite . at the level of squared matrix element ,
the following equivalence holds : @xmath32 where the colour matrix , @xmath33 , can be evaluated automatically by counting the number of cycles that are obtained by closing the colour flows of @xmath34 .
full colour summation is performed using the right - hand side of eq .
[ fullcolorsum ] .
while appealing , the colour - connection representation has a drawback in that the computation becomes less efficient for processes with high parton multiplicities .
indeed , the number of partial amplitudes that have to be evaluated for a given phase - space point increases factorially with the number of the external coloured particles . a possible way out
is given by the colour sampling approach where , following the left hand side of eq .
[ fullcolorsum ] , one generates for each phase - space point a random colour assignment and computes the matrix element @xmath35 accordingly : due to eq.[colordec ] , only the subset of the @xmath26 s which have a colour flow compatible with the given assignment contribute .
in fact , as shown in table [ table:1 ] , the number of subamplitudes that have to be computed in this way is sensibly reduced , on the average , for increasing complexity of the colour structure . [ cols="<,<,<,<",options="header " , ] + at the lhc with @xmath36 , where @xmath37 .
the blue dotted curve corresponds to the lo , the red solid to the nlo result whereas the green dashed to the nlo results with a jet veto of 50 gev.,width=257,height=207 ] at the lhc .
the red solid line refers to the nlo results while the blue dotted one to the lo one.,width=257,height=207 ] at the lhc .
the red solid line refers to the nlo results while the blue dotted one to the lo one.,width=257,height=207 ] at the lhc .
the red solid line refers to the nlo results while the blue dotted one to the lo one.,width=257,height=207 ]
it is amazing how much progress has been realized over the last few years in the field of multi - leg nlo calculations . `
helac - nlo ` is a complete framework for nlo calculations , including virtual and real corrections . in its present form can be used for nlo qcd calculations for arbitrary processes including up to one - particle - irreducible 8-point one - loop functions .
it has been successfully used for a number of processes of current interest , such as @xmath38 including the higgs signal contributions and @xmath39 jets . in the immediate future
, we plan a complete analysis of @xmath40jets production including the full description of top decays and non - resonant contributions , i.e. @xmath41 . plans of further development include also the incorporation of all @xmath16 and uv counter - terms for full electroweak corrections .
_ acknowledgement : this work was funded in part by the rtn european programme mrtn - ct-2006 - 035505 heptools - tools and precision calculations for physics discoveries at colliders .
m.c . was supported by the heisenberg programme of the deutsche forschungsgemeinschaft .
m.w . was supported by the initiative and networking fund of the helmholtz association , contract ha-101 ( `` physics at the terascale '' ) .
m.v.g . and r.p .
thank the financial support of the mec project fpa2008 - 02984 .
was additionally supported by the infn . _
j. r. andersen _ et al . _ [ sm and nlo multileg working group ] , arxiv:1003.1241 [ unknown ] .
r. keith ellis , k. melnikov and g. zanderighi , phys .
d * 80 * ( 2009 ) 094002 [ arxiv:0906.1445 [ hep - ph ] ] . c. f. berger _
et al . _ ,
d * 80 * ( 2009 ) 074036 [ arxiv:0907.1984 [ hep - ph ] ] .
a. bredenstein , a. denner , s. dittmaier and s. pozzorini , phys .
* 103 * ( 2009 ) 012002 [ arxiv:0905.0110 [ hep - ph ] ] . g. bevilacqua , m. czakon , c. g. papadopoulos , r. pittau and m. worek , jhep * 0909 * ( 2009 ) 109 [ arxiv:0907.4723 [ hep - ph ] ] .
k. melnikov and g. zanderighi , phys .
d * 81 * ( 2010 ) 074025 [ arxiv:0910.3671 ] .
t. binoth , n. greiner , a. guffanti , j. reuter , j. p. guillet and t. reiter , phys .
b * 685 * ( 2010 ) 293 [ arxiv:0910.4379 ] .
a. bredenstein , a. denner , s. dittmaier and s. pozzorini , jhep * 1003 * ( 2010 ) 021 [ arxiv:1001.4006 [ unknown ] ] .
g. bevilacqua , m. czakon , c. g. papadopoulos and m. worek , phys .
* 104 * ( 2010 ) 162002 [ arxiv:1002.4009 ] . c. f. berger _
_ , arxiv:1004.1659 .
g. ossola , c. g. papadopoulos and r. pittau , nucl .
b * 763 * ( 2007 ) 147 [ arxiv : hep - ph/0609007 ] .
g. ossola , c. g. papadopoulos and r. pittau , jhep * 0805 * ( 2008 ) 004 [ arxiv:0802.1876 [ hep - ph ] ] .
a. cafarella , c. g. papadopoulos and m. worek , comput .
commun .
* 180 * ( 2009 ) 1941 [ arxiv:0710.2427 [ hep - ph ] ] .
a. van hameren , c. g. papadopoulos and r. pittau , jhep * 0909 * ( 2009 ) 106 [ arxiv:0903.4665 [ hep - ph ] ] .
p. draggiotis , m. v. garzelli , c. g. papadopoulos and r. pittau , jhep * 0904 * ( 2009 ) 072 [ arxiv:0903.0356 [ hep - ph ] ] .
m. v. garzelli , i. malamos and r. pittau , jhep * 1001 * ( 2010 ) 040 [ arxiv:0910.3130 ] . w. t. giele , z. kunszt and k. melnikov , jhep * 0804 * ( 2008 ) 049 [ arxiv:0801.2237 [ hep - ph ] ] .
r. k. ellis , w. t. giele , z. kunszt and k. melnikov , nucl .
b * 822 * ( 2009 ) 270 [ arxiv:0806.3467 [ hep - ph ] ] .
a. lazopoulos , arxiv:0812.2998 [ hep - ph ] .
r. britto , f. cachazo and b. feng , nucl .
b * 725 * ( 2005 ) 275 [ arxiv : hep - th/0412103 ] . c. anastasiou , r. britto , b. feng , z. kunszt and p. mastrolia , phys .
b * 645 * ( 2007 ) 213 [ arxiv : hep - ph/0609191 ] .
g. ossola , c. g. papadopoulos and r. pittau , jhep * 0803 * ( 2008 ) 042 [ arxiv:0711.3596 [ hep - ph ] ] .
a. kanaki and c. g. papadopoulos , comput .
commun .
* 132 * ( 2000 ) 306 [ arxiv : hep - ph/0002082 ] .
a. kanaki and c. g. papadopoulos , arxiv : hep - ph/0012004 .
s. catani and m. h. seymour , nucl .
b * 485 * ( 1997 ) 291 [ erratum - ibid .
b * 510 * ( 1998 ) 503 ] [ arxiv : hep - ph/9605323 ] .
s. catani , s. dittmaier , m. h. seymour and z. trocsanyi , nucl .
b * 627 * ( 2002 ) 189 [ arxiv : hep - ph/0201036 ] .
m. czakon , c. g. papadopoulos and m. worek , jhep * 0908 * ( 2009 ) 085 [ arxiv:0905.0883 [ hep - ph ] ] . c. g. papadopoulos , comput .
commun .
* 137 * ( 2001 ) 247 [ arxiv : hep - ph/0007335 ] .
a. van hameren , arxiv:1003.4953 [ unknown ] . z. nagy and z. trocsanyi , phys .
d * 59 * ( 1999 ) 014020 [ erratum - ibid .
d * 62 * ( 2000 ) 099902 ] [ arxiv : hep - ph/9806317 ] .
z. nagy , phys .
d * 68 * ( 2003 ) 094002 [ arxiv : hep - ph/0307268 ] .
the atlas collaboration ( g. aad _ et al .
_ ) , arxiv:0901.0512v4 [ hep - ex ] | achieving a precise description of multi - parton final states is crucial for many analyses at lhc . in this contribution
we review the main features of the ` helac - nlo ` system for nlo qcd calculations . as a case study , nlo qcd corrections for @xmath0 production at lhc are illustrated and discussed . |
the third - order quantum electrodynamical process of photon splitting in a strong magnetic field , currently popular in several astrophysical models of different neutron star sources , was first studied over three decades ago . due to analytic complexities encountered when investigating this interaction , it was not until the beginning of the 1970s that a body of correct and uncontroversial results emerged .
these early splitting calculations used either effective lagrangian @xcite or variations of schwinger s proper - time techniques @xcite , the expediency of which yielded compact analytic forms for the rates when specializing to low energy ( ) or low field ( ) cases . after a hiatus of nearly two decades , photon splitting became of interest again in the literature @xcite following the publication of an s - matrix calculation in the landau representation of its rates by mentzel , berg and wunner@xcite , specifically because of their contention that the earlier works cited above had seriously underestimated the strength of this process .
the rates computed in @xcite were later retracted in @xcite , with a sign error in their numerical coding having been discovered and corrected .
mentzel et al.s analytic derivation was the first comprehensive presentation of the application of a landau representation technique specifically to magnetic photon splitting , though the qed formalism presented by melrose and parle @xcite virtually provided an equivalent enunciation of such s - matrix forms for splitting amplitudes .
more recently , weise , baring & melrose @xcite confirmed the analytic derivation of @xcite .
the landau representation calculations and most of the earlier effective lagrangian and proper - time presentations were generally applicable to non - dispersive regimes below the pair creation threshold ( ) , where the momentum vectors of the initial and final photons are collinear , and arbitrary field strengths . below pair threshold , the effective lagrangian approach of @xcite and the proper - time calculations in @xcite
appear much more amenable for the purposes of numerical evaluation than the s - matrix formulation in the landau representation .
this arises because effective lagrangian and proper - time ( collectively referred to by the label elp here ) methods produce results that involve triple integrals over relatively simple ( hyperbolic and exponential ) functions , while the s - matrix amplitudes integrate over the parallel momentum and include a triple summation over the landau level quantum numbers of the intermediate pair states .
both techniques start from different but equivalent @xcite forms of the electron propagator , and hence s - matrix computations @xcite should yield identical results to proper - time numerics @xcite . for the specific case of magnetic pair creation ,
such an equivalence of the s - matrix and proper - time methods has been demonstrated@xcite , but only via continuous asymptotic approximations that smoothly average out the exact `` sawtooth '' resonance structure .
yet the s - matrix landau representation approach explicitly retains the resonances in the scattering amplitudes above pair threshold , whereas the elp methods eliminate such information early during developments .
photon splitting becomes effectively first - order in at any one of a multitude of pair resonances , generated when the intermediate states become `` on - shell . ''
hence it is quite possible that splitting can compete effectively with pair creation as a photon absorption mechanism above pair threshold .
ascertaining whether this is true is an interesting physics question .
moreover , if splitting is approximately as probable as pair creation above threshold , then it manifestly changes the character of vacuum dispersion , so that quadratic ( and by inference perhaps higher order ) contributions to the vacuum polarization tensor become significant relative to the standard linear ones used in the derivation @xcite of kinematic selection rules for splitting .
hence , the generation of exact and compact expressions for the rates for valid both below and above pair creation threshold is clearly a worthwhile enterprise from a physics perspective .
developed expressions for the rates for photon splitting are also important for astrophysical applications of this process , particularly to effect efficient and accurate computations of such rates .
these applications have so far focused on neutron star magnetospheres , primarily on models of soft gamma repeaters ( sgrs ) and strongly - magnetized pulsars , both being extremely topical in the astrophysics community at present .
the potential importance of splitting in neutron star environments was suggested by @xcite .
possible formation of splitting cascades has been explored in models of sgr transient outbursts as a means of softening the spectrum efficiently with no production of pairs @xcite .
if both polarizations can split , or if polarization switching is active during sgr outbursts , then the properties of the splitting cross - section guarantee emergent spectra in the observed range ( 20150 kev ) and of the observed shape for all fields in excess of around gauss @xcite , provided that the emission region is not concentrated near the polar cap .
the spectral properties of sgrs in quiescent emission appear to be distinct from those during outburst .
pulsations and temporal increases of their periods ( i.e. spin - down ) have now been observed @xcite for two of the four confirmed sgrs ( sgr 1806 - 20 and sgr 1900 + 14 ) , leading to inferences of fields in the vicinity of gauss .
the connection between these pulsars of extremely high magnetization , so - called _ magnetars _ , and conventional radio / x - ray / gamma - ray pulsars is not well - understood .
baring & harding @xcite postulated that radio quiescence , a property of the sgrs , may be common in magnetars due to the efficient action of photon splitting and other effects in suppressing the creation of pairs .
photon splitting also has spectral implications for such pulsars with more modest fields : @xcite demonstrated that the unusual absence of mev emission in the gamma - ray pulsar psr 1509 - 58 ( whose spin - down field is gauss ) can naturally be explained by the operation of in the intense magnetic and gravitational fields near its surface .
several desirable goals are immediately identifiable on the basis of this historical path for the study of the physics of photon splitting , and the needs of the astrophysics community . it would be satisfying ( i ) to obtain analytic expressions for rates that are valid above pair creation threshold using the landau representation methodology , ( ii ) to know whether the analytic formalism of mentzel et al .
@xcite can be developed and simplified , and ( iii ) to demonstrate a formal equivalence between this s - matrix landau representation approach and extant results from proper - time / effective lagrangian techniques .
this paper addresses these issues , using the verified analytic formalism of mentzel et al . as the starting point for mathematical developments .
the analysis here considers all the polarization modes that are permitted by the cp invariance symmetry ( , and ) , and applies for collinear momenta of the incoming and outgoing photons , i.e. when the effects of vacuum dispersion are neglected . a significant development provided in this paper is the dramatic simplification incurred by algebraically performing the summation over the spin states that are incorporated in the electron propagators .
the resulting expressions in section [ sec : reduce ] ( first stated in @xcite ) are relatively compact , and of an appearance familiar to landau representation / s - matrix theory applications to magnetized environments ( i.e. including associated laguerre functions ) .
furthermore , here the integrations over the momentum parallel to the field are performed analytically for the first time in section [ sec : pzint ] , rendering the splitting rates in most amenable forms ( see eqs . [
[ eq : totratefin ] ] and [ [ eq : mform ] ] ) that are optimal for numerical applications : the analytic forms presented consist of just triple summations over landau level quantum numbers of the intermediate states . these general results are valid both below and above pair threshold at non - resonant photon energies , and provide substantial advances over the work of @xcite ; they are much more suitable for numerical evaluation since many cancellations have been eliminated algebraically .
two specializations are discussed in section [ sec : limits ] , primarily to ( partially ) demonstrate equivalence of the landau representation formalism presented here with extant proper - time / effective lagrangian limiting forms for splitting rates , and simultaneously to serve as a check on the mathematical manipulations of this paper .
results are presented for all three polarization modes permitted by cp invariance in the limit of zero dispersion .
the first asymptotic regime is ( see section [ sec : bgg1 ] ) for highly supercritical fields , , where in the case of , the limit was found to concur with a recent analytic result that was obtained by baier et al .
@xcite , while new results were obtained for the other two modes . in the second specialization , in section [ sec : omegall1 ] ,
asymptotic results for energies well below pair creation threshold were obtained , reproducing the cubic energy dependence of the amplitudes obtained by other qed techniques . moreover ,
new and compact expressions for the scattering amplitudes in this low energy limit are derived in terms of the logarithm of the gamma function , its integral and their derivatives .
these simplified forms in eqs .
( [ eq : mperptoparparwll1 ] ) and ( [ eq : mperptoperpperpwll1 ] ) are also produced from extant integral forms for splitting matrix elements derived first in @xcite , thereby facilitating the first analytic demonstration of the equivalence of splitting rates obtained by the s - matrix formulation in the landau representation and those derived using schwinger - type techniques .
the rates for photon splitting within an s - matrix formulation can be developed using a variety of conventions ; here the landau representation used by mentzel , berg and wunner @xcite is adopted , and formal developments lead to an independent confirmation of their analytic derivation .
specifically , for a field * b * , this approach uses a representation of the electron / positron wavefunctions as eigenstates of the magnetic moment ( or spin ) operator ( with ) in cartesian coordinates within the confines of the landau gauge .
such states turn out to be very convenient because they generate useful symmetry properties ; they were identified by sokolov and ternov @xcite , who dubbed them states of `` transverse polarization . ''
let , and denote the polarizations of the initial and final ( primed ) photons ( ) , and , and denote the absorbed and produced photon four momenta .
then the total rate for splitting via the polarization mode can be written , using eqs .
( 27)(29 ) of @xcite , in terms of the s - matrix element , which is the sum of six terms corresponding to the six viable time - ordering possibilities : @xmath1 where and denote the volume and time associated with the interaction calculation and the factor of out the front avoids double counting of the final states .
the priming convention adopted throughout the paper is one and two primes for the produced photons and no prime for the initial photon .
since the s - matrix element contains a delta function prescribing four - momentum conservation for splitting , it is squared in the usual way using .
note that the s - matrix element should possess a cubic dependence on photon energies when well below pair creation threshold , due to parity symmetry , photon gauge invariance , and the antisymmetric nature of the electromagnetic field tensor ; details are discussed in @xcite . before writing down expressions for the s - matrix element terms ,
it is appropriate to identify the dimensionless convention that shall be adopted throughout this paper . since the electron rest mass is the only mass that enters into this qed problem , we opt to scale all energies by and momenta by unless otherwise specified .
this includes a scaling of for photon frequencies . in the spirit of this convention
, we choose to use the symbol to represent dimensionless electron energies and reserve ( ) to denote `` dimensional '' energies as in @xcite .
in addition , the magnetic field will be expressed in terms of the quantum critical field hereafter , so that denotes a field of gauss .
the convention for polarizations is identical to that assumed in @xcite , who opted for real polarization vectors with zero time components .
the polarization states and are defined according to whether the photon s electric vector lies either perpendicular or parallel ( respectively ) to the plane containing the photon s momentum and the ( uniform ) magnetic field * b * vectors , the convention of @xcite . in the limit of zero dispersion ,
three polarization modes are permitted by charge / parity ( cp ) invariance in qed , namely , and .
however , adler @xcite showed ( see also @xcite ) that for weak vacuum dispersion ( roughly delineated by ) , where the refractive indices for the polarization states are very close to unity , energy and momentum could simultaneously be conserved only for the splitting mode .
this kinematic selection rule applies to gamma - ray pulsar magnetospheres where plasma dispersion is negligible . in magnetar models of soft gamma repeaters , where supercritical fields are employed
, strong vacuum dispersion arises .
in such a regime , it is not clear whether adler s selection rules still endure , since his linear dispersion analysis omits higher order ( quadratic ) contributions @xcite to the vacuum polarization tensor ( e.g. those that couple to photon absorption via splitting ) that may become significant in supercritical fields .
furthermore , plasma dispersion effects , which can nullify the vacuum selection rules , may be quite pertinent @xcite to soft gamma repeater magnetospheres , rendering them distinctly different from those of conventional pulsars .
therefore , in the interests of generality , consideration of all three cp - permitted splitting modes is adopted throughout this paper .
the derivation of the s - matrix element proceeds along lines identical to those in mentzel , berg & wunner @xcite , with the result being an exact reproduction of their analytic formalism , as reported in weise , baring & melrose @xcite ; for details , one is referred to @xcite .
of the six contributions to eq .
( [ eq : totrate ] ) , it is sufficient to explicitly present just one : @xmath2 where is the fine structure constant , and the energies and are defined in eq .
( [ eq : erg ] ) below , with similar definitions for the primed energies involving primed quantum numbers and momenta of the virtual electrons . using the notation in eq .
( [ eq : jeval ] ) below , @xmath3\ ; e''_-\nonumber\\ & & + \ ; j(-k''_x \,\vert\ , n',\ , -k''_y\,\vert\ , n-1,\ , 0 ) \ , \vsp [ \kappa'^{\ast}_4\kappa_1 + \kappa'^{\ast}_2\kappa_3]\ ; e''_+\nonumber\\ & & -\ ; \bigl\ { j(-k''_x \,\vert\ , n',\ , -k''_y \,\vert\ , n,\ , 0)\ , [ \kappa'^{\ast}_2\kappa_4 + \kappa'^{\ast}_4\kappa_2 ] - j(-k''_x \,\vert\ , n'-1,\ , -k''_y \,\vert\ ,
n-1,\ , 0)\ , [ \kappa'^{\ast}_1\kappa_3 + \kappa'^{\ast}_3\kappa_1]\bigr\ } e''_z \ ; , \nonumber\\ { \cal d}^{+-}_{n\ , n''}(\hbox{\bf k } ' ) & = & \vsp j(-k'_x \,\vert\ , n-1,\ , 0 \,\vert\ , n'',\ , k'_y)\ , [ \kappa^{\ast}_1 \kappa''^{\ast}_2 + \kappa^{\ast}_3\kappa''^{\ast}_4]\ ; e'_-\nonumber\\ & & + \ ; j(-k'_x \,\vert\ , n,\ , 0 \,\vert\ , n''-1,\ , k'_y)\ , \vsp [ \kappa^{\ast}_2\kappa''^{\ast}_1 + \kappa^{\ast}_4\kappa''^{\ast}_3 ] \ ; e'_+ \label{eq : cald}\\ & & -\ ; \bigl\ { j(-k'_x \,\vert\ , n,\ , 0 \,\vert\ , n'',\ , k'_y)\ , [ \kappa^{\ast}_2\kappa''^{\ast}_2 + \kappa^{\ast}_4\kappa''^{\ast}_4 ] - j(-k'_x \,\vert\ , n-1,\ , 0 \,\vert\ , n''-1,\ , k'_y)\ , [ \kappa^{\ast}_1 \kappa''^{\ast}_1 + \kappa^{\ast}_3\kappa''^{\ast}_3 ] \bigr\ } e'_z \ ; , \nonumber\\ { \cal d}^{-+}_{n''n'}(\hbox{\bf k } ) & = & \vsp j(k_x \,\vert\ , n''-1,\ , k'_y \,\vert\ , n',\ , -k''_y)\ , [ \kappa'_4\kappa''_3 + \kappa'_2\kappa''_1]\ ; e_-\nonumber\\ & & + \ ; j(k_x \,\vert\ , n'',\ ,
k'_y \,\vert\ , n'-1,\ , -k''_y ) \ , \vsp [ \kappa'_3\kappa''_4 + \kappa'_1\kappa''_2]\ ; e_+\nonumber\\ & & -\ ; \bigl\ { j(k_x \,\vert\ , n'',\ , k'_y \,\vert\ , n',\ , -k''_y ) \ , [ \kappa'_4\kappa''_4 + \kappa'_2\kappa''_2 ] - j(k_x \,\vert\ , n''-1,\ , k'_y \,\vert\ , n'-1,\ , -k''_y ) \ , [ \kappa'_3\kappa''_3 + \kappa'_1\kappa''_1 ] \bigr\ } e_z \ ; , \nonumber\end{aligned}\ ] ] where the polarization vector is specified by and , and similarly for the final photon polarizations ( primed ) . the other three s in eq . ( [ eq : mbweq25 ] ) are not displayed here for brevity ; they can be obtained from those in eq .
( [ eq : cald ] ) simply by the interchange of polarization components ( and similarly for primed components ) and a relabelling of the s that produces a correspondence .
several notations need to be identified .
first , the particles have energies , and momentum components along the field . the energies and
that appear here are , respectively , with and without the parallel momentum : @xmath4 with denoting the landau level quantum numbers , as usual .
the other quantum number pertaining to the eigenstates of is , which signifies the spin state of the fermions ( @xcite used the label ; here the notation of melrose and parle @xcite is preferred ) , and satisfies .
it does not appear explicitly in , but is embedded in the spinor coefficients : @xmath5 for and where the spin quantum number takes on two values except for the ground state ( zeroth landau level ) , where only is permissible .
here is the familiar kronecker delta .
the primed coefficients and are similarly defined in terms of primed momenta and landau level quantum numbers , subject to the momentum conservation implicit in eq .
( [ eq : mbweq25 ] ) .
the functions that appear in eq .
( [ eq : cald ] ) are integrals over the oscillator functions ( hermite polynomial products ) , a form undeveloped in @xcite . here , eq .
( 7.377 ) of @xcite is employed to express these integrals analytically in terms of generalized laguerre polynomials , ( see also eq . ( 47 ) of @xcite ) : @xmath6\bigr)\ , \bigl\ { ie^{-i\psi}\bigr\}^{n'-n}\ , i_{n',n}\bigl(\dover{\rho^2}{2b}\bigr ) \label{eq : jeval}\ ] ] where ( ) and the phase are introduced for convenience of notation : and . note that the s are always s and is always a . here
the functions follow the sokolov and ternov convention @xcite up to a factor of , being related to the functions of melrose and parle @xcite , and both are defined in terms of the generalized laguerre polynomials ( see @xcite ) : @xmath7 values for are obtained by interchanging indices , as indicated . hereafter ,
the ( modified ) sokolov and ternov convention for writing the laguerre polynomials will be adopted .
complex conjugation of eq .
( [ eq : jeval ] ) can be used to establish the identity @xmath8 noting that the products are either purely imaginary or real , for all choices of spin quantum numbers .
the three factors like appearing in the second product of three s in eq .
( [ eq : mbweq25 ] ) cancel , leading to this product being just the complex conjugate of the first three s. this useful symmetry property clearly underlines the convenience of the sokolov and ternov choice of wavefunctions when adopting real components for the photon polarization . the form of the contribution to the s - matrix element in eq .
( [ eq : mbweq25 ] ) is _ identical _ to that for given in eq . (
25 ) of mentzel , berg and wunner @xcite . in the same fashion
, it can be found that the expression derived here for is absolutely identical to eq .
( 26 ) of @xcite , thereby providing confirmation of their analytic developments ; it can be obtained by using the substitutions , and in eq .
( [ eq : cald ] ) .
all other contributions result from application of the cyclic permutations @xmath9
\label{eq : perm}\\[-5.5pt ] p_{-1 } : & & -k_{\mu},\ ; e_{\mu}^{\ast}\;\to\ ; k_{\mu}',\ ; e_{\mu}'\quad , \quad k_{\mu}',\ ; e_{\mu}'\;\to\ ; k_{\mu}'',\ ; e_{\mu}''\quad , \quad k_{\mu}'',\ ; e_{\mu}''\;\to\ ; -k_{\mu},\ ; e_{\mu}^{\ast}\quad , \nonumber\end{aligned}\ ] ] where , , etc . observe also that a minus sign and the complex conjugation of the polarizations are always associated with the initial photon since it is absorbed in the process .
given these permutations , the crossing symmetry for splitting is manifested in the following relationship between the various terms like those in eq .
( [ eq : mbweq25 ] ) that contribute to eq .
( [ eq : totrate ] ) : @xmath10 where the permutations act as operators .
this symmetry can be expressed in a multitude of ways using the identities and .
it is important to remark that the derivation of analytic forms by mentzel , berg and wunner is not the first in the literature relating to s - matrix applications to photon splitting .
the papers by melrose and parle @xcite dealing with various aspects of qed in strong magnetic fields , specifically from a wave dispersion / response tensor approach , constructed the s - matrix element for splitting in eqs .
( 46 ) and ( 47 ) of @xcite , which incorporated the quadratic vacuum response tensor given in eq .
( 36 ) of @xcite .
this tensor is obviously of a standard s - matrix landau representation appearance .
( [ eq : mbweq25 ] ) and ( [ eq : cald ] ) can be generated directly ( and also ) from the melrose and parle evaluation after a modicum of algebra .
hence , eqs .
( [ eq : mbweq25 ] ) and ( [ eq : cald ] ) here , and eqs . ( 25 ) and ( 26 ) can be used as reliable starting points for further s - matrix developments . the form in eqs .
( [ eq : mbweq25 ] ) and ( [ eq : cald ] ) is quite cumbersome .
it can be simplified considerably by ( i ) specializing to specific but representative directions of photon propagation and ( ii ) analytically performing the summations over spin states , and . restricting the photon motion to the -direction yields photon motion perpendicular to the field : since splitting is collinear in the non - dispersive limit discussed earlier in this paper , it follows that .
this choice dramatically simplifies coefficients of the laguerre polynomials in eq .
( [ eq : cald ] ) . without significant loss of generality ,
setting removes nearly all of the phase factors in the definition of the s in eq .
( [ eq : jeval ] ) , leaving just .
three such factors emerge in the triple product of s , leading to a factor of .
the cp symmetry possessed by the splitting process becomes most evident at this point , since it is now simple to derive the cp selection rules .
the specification of and yields only one possible component of polarization perpendicular to the field , and one conceivable component of polarization parallel to the field , ( and similarly for primed quantities ) .
the polarization ( electric field ) vector of the photons is , of course , normal to the photon momentum vector , which automatically spawns the notation for the two possible polarization states : and . from the presence of subtractions in the numerators of the integrands of eq .
( [ eq : mbweq25 ] ) together with the complex conjugation property in eq .
( [ eq : caldsymm ] ) and the proportionality of the s to factors like , it follows that only terms with an odd number of factors contribute to , i.e. terms proportional to , , and .
all other terms cancel identically to zero . by virtue of the permutation symmetries in eq .
( [ eq : perm ] ) , this is also true for all other .
it is then trivial to deduce the cp selection rules for photon splitting , namely that the only permitted transitions are @xmath11 the three other splitting transitions all have s - matrix elements that are exactly zero for collinear photon momenta , and hence are forbidden .
this technique for cp selection rule derivation was implemented in @xcite .
these restrictions are simply consequences of the charge conjugation ( c ) and parity ( p ) symmetries of the splitting process , i.e. relating to the transformations and .
the summation over the spin states , and ( ) produces a dramatic simplification in the appearance of the s - matrix elements .
such spin summations act only on the products of the s that appear in eq .
( [ eq : cald ] ) ; the algebra is lengthy but straightforward , being facilitated by pairing terms with denominators that differ only in the sign of their photon energies .
the total splitting rate in eq .
( [ eq : totrate ] ) can be written in the form @xmath12 where is implicitly understood from the conservation of four - momentum . while these rates will be expressed for photon propagation normal to the uniform magnetic field , the results for general photon obliquities to * b *
can be obtained via a simple lorentz transformation : , , , together with an extra multiplicative factor of applied to the rate in eq .
( [ eq : totratefin ] ) .
the comparative simplicity of the reduced form of the s - matrix element relative to eq .
( [ eq : mbweq25 ] ) is both notable and comforting . unlike eqs .
( 25 ) and ( 26 ) of @xcite , this developed form of the splitting s - matrix element has an appearance familiar to s - matrix applications of qed in the landau representation to strongly - magnetized systems , with products of generalized laguerre polynomials multiplied by simple combinations of energies and momentum components .
examples of previous work bearing such familiar forms focus largely on lower - order qed processes and include studies of synchrotron radiation @xcite , single photon pair creation @xcite , and vacuum @xcite and plasma @xcite polarization . for the purposes of the analysis in the next section , it is pertinent to define the cyclic permutations @xmath19
\label{eq : permsymm}\\[-6.5pt ] & & n \;\to\ ; n''\;\ ; , \quad n ' \;\to\ ; n\;\ ; , \quad n '' \;\to\ ; n'\;\ ; , \nonumber\end{aligned}\ ] ] in the spirit of the permutation in eq .
( [ eq : perm ] ) .
these permutations will appear repeatedly in the developments below , and lead to the following transformation properties of eq .
( [ eq : sigmadef ] ) : @xmath20 with being invariant , symmetries that are consequences of the arrangements of electron and positron propagators in the feynman diagram for splitting .
these translate into obvious mappings between , and and an invariance of the .
it is also easily seen that under this cyclic permutation , the factor in braces in the summation for is invariant , while the equivalent factor in the summation for maps over ( up to a minus sign ) to the factor in braces in the summation . as will become evident in section [ sec : limits ] , the remaining powers of in the summations do not provide any unsatisfactory interference in the limits of low photon energy ( ) and high fields ( ) , so that permutation symmetry can be extended to the total amplitudes in these specific parameter regimes .
further analytic development is not only possible , but also desirable , given that the integrations over the momentum parallel to the field can be expressed compactly in terms of elementary functions .
such tractability facilitates both numerical evaluations and the derivation of asymptotic limits . in proceeding ,
since results are sought at energies sufficiently remote from pair creation resonances , the imaginary parts of the denominators in the are dropped in all further considerations , i.e. , we consider only the functions .
it turns out that carefully - constructed contour integrations in the complex plane do not facilitate the integrations .
hence the first step in integrating over is effected by the more cumbersome and less elegant approach of completing the squares and rationalizing the denominators using products of factors like .
these factors define poles of the integration for and being some combination of , and .
such poles fall into two types : pair creation ones ( e.g. see @xcite ) that contribute only above pair threshold , due to the structure of the splitting rate , and cyclotronic ones that must be considered below pair threshold .
the appearance of such cyclotronic poles is an artifact of the rationalization of denominators , so that they are really pseudo - poles of the subsequent analysis ; a consistency check on the algebra is that the s - matrix element be effectively continuous across them .
it is convenient to define energies that correspond to the poles : @xmath21 and three others paired with these , which are obtained via the relations , and . here
the notation @xmath22 is used for the purposes of abbreviation .
observe that , taking advantage of the subjectivity of such definitions , a minus sign appears in front of the expression for , a choice that preserves symmetries induced by the mapping in eq .
( [ eq : permsymm ] ) in the results that follow .
these definitions spawn the following useful identities for the momentum poles : @xmath23 which immediately imply the possibility of poles along the imaginary axis .
in fact , , with equality for , and likewise for the other poles . note that for the one - vertex calculations of cyclotron emission and single photon pair creation and annihilation , the requirement that such poles be real , corresponding to real components of particle momenta on external lines , is precisely what generates thresholds ( e.g. @xcite ) and kinematic cutoffs ( e.g. @xcite ) for transitions involving various states .
the rationalization of the denominators yields relatively compact decompositions for these sums , after much cancellation and simplification .
they take the form @xmath24 the simplicity of which is contingent upon the energy - conservation restriction . here
@xmath25 identities such as prove useful in the ensuing analysis .
the and coefficients assume simple forms when expressed as partial fractions . consider first the result for , which has the coefficients @xmath26 \label{eq : sigma11coeff}\\[-5.5pt ] t_{1,1}^{\erg'\erg '' } & = & \dover{\erg_{nn''}}{p_{n''n}^2 - p_z^2 } + \dover{\erg_{nn'}}{p_{nn'}^2 - p_z^2 } \nonumber\\[2pt ] t_{1,1}^{\erg''\erg } & = & \dover{\erg_{n'n}}{p_{nn'}^2 - p_z^2 } + \dover{\erg_{n'n''}}{p_{n'n''}^2 - p_z^2 } \;\ ; . \nonumber\end{aligned}\
] ] observe that a cyclic symmetry is immediately apparent : is invariant under the permutation in eq .
( [ eq : permsymm ] ) , as is evident from its original definition in eq .
( [ eq : sigmadef ] ) .
similarly , the algebraic developments yield coefficients for the sum , which appears in the terms , as @xmath27 + \dover{\erg_{n''n}\erg_{nn '' } } { p_{n''n}^2 - p_z^2 } \;[\erg_{n''n}-\omega ] \,\biggr\ } \nonumber\\ t_{-1,-1}^{\erg\erg ' } & = & 0 \nonumber\\[-5.5pt ] & & \label{eq : sigmam1m1coeff}\\[-5.5pt ] t_{-1,-1}^{\erg'\erg '' } & = & - \dover{\erg_{n''n'}+\omega'}{p_{n'n''}^2 - p_z^2 } - \dover{\erg_{nn'}}{p_{nn'}^2 - p_z^2 } \nonumber\\[2pt ] t_{-1,-1}^{\erg''\erg } & = & - \dover{\erg_{n''n}-\omega}{p_{n''n}^2 - p_z^2 } - \dover{\erg_{n'n}}{p_{nn'}^2 - p_z^2 } \;\ ; . \nonumber\end{aligned}\ ] ] the coefficients for the sum that appears in the terms and the coefficients for the sum that appears in the terms are similar : there is little need to state them explicitly , since the coefficients possess a relationship to each other due to the permutation symmetry enunciated in eq .
( [ eq : sigmasymm ] ) . given these decompositions , it is now fairly straightforward to evaluate the integrations over , expressing them in terms of the an elementary function with real arguments : @xmath28 for real .
the identity with has been used to map across the singularities at ( cyclotronic below pair threshold ) and guarantee bounded and continuous behaviour of at . the integral identity in eq .
( [ eq : ffuncdef ] ) can be established quickly with the aid of result 3.513.2 in @xcite , using the substitution and partial fractions .
note that real values ( either positive or negative ) of are guaranteed by the formalism here , with being improbable due to the discreteness of the quantum numbers , and .
the integration of the coefficients of the terms for each of the polarization modes are then straightforward , and the identities in eq .
( [ eq : poleident ] ) can be used to advantage .
similar terms appear in the integrations of parts of the coefficients of the other terms , which also possess integrands with terms proportional to , and that formally lead to divergences that cancel each other ( an artifice introduced by the rationalization of the denominators ) . using partial fractions , the divergent contributions can be written as integrals over the finite range , rearranging to subtract off exactly - cancelling terms , and then taking the limit as .
similar manipulations are used for the integration over , where again the leading order terms are individually divergent yet collectively convergent .
partial fractions can again be used to enable rearrangements and separate the divergent terms , which are then integrated over finite ranges as with the evaluation .
the results are encapsulated in the identities @xmath29 where @xmath30 and @xmath31 no further integration is necessary : the cyclic permutations in eq .
( [ eq : permsymm ] ) can be used to quickly derive expressions for and from eq .
( [ eq : sigmaint ] ) . at this point , it is salient to remark that the divergences at in the functions pose no problem for the integral evaluations in eqs .
( [ eq : sigmaint ] ) , because these functions always appear two at a time . below the pair threshold ,
these divergences are cyclotronic in nature , being encountered when or for similar circumstances for the other photon energies . as tends to such a limit ,
for example , we observe that and when ( without loss of generality ) .
this opposition of signs guarantees cancellation of divergences when the form of is used ( as ) , so that continuity across cyclotron `` pseudo - resonances '' emerges naturally from eq .
( [ eq : sigmaint ] ) , consistent with the continuity of the functions .
continuity across pair resonances does not arise above pair threshold , so that true divergences emerge .
the incorporation of eq .
( [ eq : sigmaint ] ) into the scaled matrix elements in eq .
( [ eq : mform ] ) constitutes the final product of the general analytic developments in this paper , providing rates valid for all energies below pair threshold ( and applicable for non - resonant energies above threshold ) , and for photon propagation normal to the uniform magnetic field .
they are eminently suitable for numerical computations , having improved upon the analytic formalism of mentzel , berg and wunner @xcite ( i.e. eq . [
[ eq : mbweq25 ] ] ) by performing the summations of the spin states and integration over the momenta parallel to the field that are associated with the electron propagators .
such developments are prudent prior to numerical evaluations due to the large degree of cancellation in these sums and integrations .
a fruitful extension of this analysis is the exploration of the simplification of the scattering amplitudes and rates in two particular asymptotic regimes , namely the limit of highly supercritical fields , , and the specialization to photon energies well below threshold , i.e. .
the benefits of such an investigation are twofold .
first , it provides the first unequivocal analytic demonstration of the equivalence of splitting results from the s - matrix formulation in the landau representation and effective lagrangian / proper - time results from schwinger - type formalisms in well - defined parameter regimes . in doing so
, it serves as a powerful check on the developments here .
second , in the case , it identifies a new , satisfyingly compact representation of the scattering amplitudes in terms of special functions that leads to an efficient means of computation .
these two parameter regimes are encompassed under the single limit , which thereby identifies the appropriate series expansion of the generalized laguerre polynomials that appear in the amplitudes . for small arguments ,
the leading order terms in the series for can be found in the appendix of @xcite . given that , and cluster in a manner such that , this series converges rapidly provided .
hence actually represents the true expansion parameter here , with and being similarly bounded .
the leading order terms of such expansions for the are linear in the photon energies , while the next higher order terms are cubic ; a more detailed exposition can be found in weise , baring and melrose @xcite .
the series for the integrations of , namely , , , and ( which do not depend on the polarization mode ) are expansions in rather than .
they are independent of photon energy to leading order , with a quadratic scaling with energy to next order .
the series for possess logarithmic character in the quantum numbers in situations when no two of them are equal ( i.e. ): @xmath32 \label{eq : iapprox}\\[-5.5pt ] { \cal i}_1 \;\approx\ ; - { \cal j } '' & \approx & -\dover{2 { \cal n } \log_e { \cal n } } { ( { \cal n}-{\cal n } ' ) ( { \cal n}''-{\cal n } ) } - \dover{2 { \cal n } ' \log_e { \cal n } ' } { ( { \cal n}'-{\cal n } '' ) ( { \cal n}-{\cal n } ' ) } - \dover{2 { \cal n } '' \log_e { \cal n } '' } { ( { \cal n}''-{\cal n } ) ( { \cal n}'-{\cal n}'')}\;\ ; , \nonumber\end{aligned}\ ] ] and additionally involve inverse trigonometric functions when two s ( e.g. for ) are in fact equal : @xmath33 \label{eq : iapprox2}\\[-5.5pt ] { \cal i}_1 & \approx & - \dover{2 { \cal n}''}{({\cal n}-{\cal n}'')^2}\ , \log_e \dover{{\cal n}}{{\cal n } '' } - \dover{2}{{\cal n}-{\cal n } '' } + \dover{4{\cal
n}-(\omega'')^2}{{\cal n } ( { \cal n}-{\cal n } '' ) } \ ; q\bigl ( \dover{\omega''}{2\sqrt{{\cal n } } } \bigr ) \;\approx\ ; \dover{4}{({\cal n}-{\cal n } '' ) } \ ; q\bigl ( \dover{\omega''}{2\sqrt{{\cal n } } } \bigr ) - { \cal j}''\;\ ; , \nonumber\end{aligned}\ ] ] where @xmath34 and the identity has been invoked .
this retention of the inverse trigonometric functions is particularly relevant for determining the high limiting forms of the scattering amplitudes .
relations similar to eq .
( [ eq : iapprox2 ] ) exist for and , obtained by the cyclic permutations through s and photon energies . the lengthier higher order ( quadratic ) terms
are not explicitly stated for the sake of brevity .
this concludes the preamble that guides the reader in the subsequent specializations .
this regime is of particular relevance to the study of magnetars such as soft gamma repeaters .
for the two modes and , only the leading order terms for the and the momentum integrals presented in eqs .
( [ eq : iapprox ] ) and ( [ eq : iapprox2 ] ) are required .
consider first the reduction of . here
the and terms contribute leading order terms only through , and , cases , respectively , where it is necessary to use the full forms in equation ( [ eq : iapprox2 ] ) , and inverse trigonometric functions appear through the function , which assumes the arguments and .
a similar , term is identically equal to zero by virtue of the factor .
the contributions from the and terms possess an entirely different character , being infinite summations over , with the values of and being constrained by , producing five groupings of the indices .
the series is evaluated by truncating the sum at , relabelling one of the logarithmic terms , and then taking the limit .
the net result is ( for ) @xmath35 which , when combined with equation ( [ eq : totratefin ] ) , yields the asymptotic high - b result derived by baier et al .
@xcite , and reproduced independently by baring & harding @xcite ; the overall rate for approaches a value independent of .
observe that the manifestations of the pair creation threshold for each of the final photons of polarization ( i.e. at and ) are the individually - divergent coefficients of the inverse trigonometric functions .
yet , collectively , due to the energy conservation relation , such divergences cancel each other to yield a finite overall result as . for the incident photon of polarization , the pair threshold of is remote from so that it would only become explicitly apparent when the amplitude was evaluated to higher order in .
note also that the limit of eq .
( [ eq : mperptoparparbgg1 ] ) is and reproduces results obtained in @xcite and @xcite .
the functional form of eq .
( [ eq : mperptoparparbgg1 ] ) is plotted in figure [ fig : scattampbgg1 ] . 0.5truecm the equivalent result for the splitting mode requires little additional algebra given that it can be obtained from the analysis just above using the cyclic symmetry transformations of eq .
( [ eq : permsymm ] ) .
carefully keeping track of signs and all photon frequencies by relabelling at the beginning of the manipulations , the roles of the and terms are interchanged , and the obvious result emerges : @xmath36 while not established before in the literature , the low energy limit of this , namely , yields the differential rate from previous expositions @xcite of low energy approximations .
the form of eq .
( [ eq : mpartoperpparbgg1 ] ) is displayed in figure [ fig : scattampbgg1 ] , exhibiting the asymmetry expected under interchanges . in this case ,
pair threshold structure in the amplitude appears again for the two photons of parallel polarization ( i.e. at and ) , and is also absent for the produced photon , being of higher order in .
consequently , the amplitude possesses a real divergence at , a noteworthy occurrence that is illustrated by comparing the two panels of figure [ fig : scattampbgg1 ] .
such divergences , which are not integrable over ( and therefore patently different in nature from the resonances encountered in rates for ) , are characteristic of the photon splitting rate near resonances at and above the pair threshold of , corresponding to the creation of virtual pairs in various excited states .
in fact , near such resonances , photon splitting necessarily becomes first order in like pair creation as the intermediate states `` go on - shell . ''
the rapid increase of the rate of relative to that of is exhibited in figure [ fig : ratesbgg1 ] , where the rates have been scaled by the low energy ( ) limiting forms ( ) discussed in the next subsection .
this particular scaling is chosen to illustrate deviations from the asymptotic forms , and therefore to demonstrate the need for relinquishing use of them when sampling photon energies near pair threshold , a parameter regime very relevant to certain astrophysical calculations ( e.g. see @xcite ) .
the dominance of the over near apparent in these results becomes substantive in parameter regimes where the weakly - dispersive vacuum ( i.e. for ) polarization selection rules for splitting derived by adler @xcite ( which prohibit and splittings ) may not apply if non - linear contributions to vacuum polarization or plasma effects are significant .
this underlines the saliency of a detailed determination of the dispersive properties of the magnetized vacuum or plasma medium appropriate to a particular astrophysical scenario . the derivation of the form for the amplitude for differs significantly from the results just expounded .
first , contributions from , and , and , combinations are identically equal to zero by virtue of each of the associated factors .
this automatically implies that no inverse trigonometric functions that have arguments independent of appear in the amplitude , a property not possessed by the other splitting modes .
the consequences of this are twofold .
first , this cancellation implies that the scattering for is of a higher order in than for the other two splitting modes .
second , since any potential appearance of inverse trigonometric functions spawned by the forms in eq .
( [ eq : iapprox2 ] ) involves arguments that depend on through the s , these arguments are always small when , precipitating a redundancy with the low energy limit .
hence , it follows immediately that the scattering amplitude for in the regime of highly super - critical fields is identical to that of the specialization of the low energy ( ) limit .
as the latter has been derived in various papers in the literature ( e.g. see @xcite and the subsequent section ) , here it is sufficient to merely state the result : @xmath37 this extremely simple form differs profoundly from those of the other two modes because of the absence of photons of polarization in the interaction .
hence any signatures of the pair threshold of of photons are absent in the domain of , and a scaling - type form with obvious cyclic symmetry emerges .
the low energy limit is of interest not only because it was the regime where compact analytic expressions for the splitting rates were first obtained @xcite , but also because the analysis that follows derives simple and elegant representations of the scattering amplitudes in terms of well - known special functions that provide a convenient alternative option for numerical evaluations .
the amplitudes for each of the splitting modes should exhibit a cubic energy dependence @xcite when .
hence , a necessary product of the landau representation formalism is that terms linear in photon energies should contribute exactly zero . for the polarization modes and ,
whose amplitudes are identical in the low energy limit @xcite , the demonstration of this is not dissimilar to the analysis .
the restriction generates a single infinite series in due to the clustering of and around .
the ensuing algebra in the simplification of this series is moderately lengthy , and requires re - indexing of the logarithmic terms to assume forms involving , and also some relabelling of the rational functions .
care must be taken in these rearrangements due to the infinite nature of the series , and the technique adopted is outlined just below . the terms linear in photon energy result in zero , as expected : for more details , the reader is referred to @xcite .
the next order contribution is cubic in energy , as desired , with terms coming from a mixture of ( i ) the linear terms of the combined with the quadratic higher order terms of the integrals , and ( ii ) the cubic terms in conjunction with the leading order ( constant ) terms from the integrals .
the algebra is straightforward , but lengthy and tedious , generating an exact cancellation of all but terms proportional to .
this approach leads to a reproduction of the listed in entirety in appendix b of weise , baring & melrose @xcite for both the and modes of splitting .
hence there is little point in replicating these expressions here ; the reader is referred to @xcite for details .
these results are expressed as single infinite series in the label , which sometimes starts at , and sometimes begins at higher integer values ( up to ) .
hence , an aesthetic goal is to rearrange some of these series so that the summations in each contribution begin at , and then add the terms in the series together .
this is a non - trivial exercise , given the divergent nature of the series in many of the individual contributions .
hence , considerable care must be taken when performing the rearrangements , for which there is no unique prescription .
one choice for relabelling the sums is adopted by weise , baring & melrose @xcite , though their end results expressed in their appendix c do not facilitate analytic development in the most expedient manner , and were in fact erroneous ( discussed briefly below ) . an alternative and preferable choice for rearrangement of the multitude of series over the label is adopted here , outlined as follows .
inspection of the various contributions in appendix b of @xcite reveals that they always consist of three types of terms : ( i ) logarithmic ones proportional to , for , ( ii ) rational functions of for , and ( iii ) polynomials in .
a unique method for rearrangement is to truncate all series to finite ones with , and then perform relabellings so that the first two types of terms consist only of terms and rational functions of .
this approach provides no particular focus on series that originate with labels , but requires careful accounting of the remainder terms at the upper and lower ends of the sums , for which significant cancellation arises .
the coefficients of the logarithmic functions , originally cubic in , reduce to linear functions of in this development .
the consequent simplification of the series terms is counterbalanced by the transferral of complexity to the constant remainder terms , which are purely functions of and .
taking the limit of achieves the desired ( and convergent ) result .
after considerable algebra collecting together all the constituent series in appendix b of @xcite , and performing the rearrangement as just prescribed , one arrives at the following series representation of the scaled scattering amplitudes : @xmath38 for , where @xmath39 + \biggl ( \dover{2}{3}+\dover{1}{2b^2 } \biggr)\ ; \dover{1}{1 + 2nb } - \dover{1}{3(1 + 2nb)^2}\nonumber\\[-5.5pt ] \label{eq : mseries1tdef}\\[-5.5pt ] t_{\ppp}(n,\ , b ) & = & \dover{3}{2b^2 } \log_e\bigl[n+\dover{1}{2b}\bigr ] - \dover{3}{2b^2}\ ; \dover{1}{1 + 2nb } + \dover{1}{2b^2 } \ ; \dover{1}{(1 + 2nb)^2}\nonumber\end{aligned}\ ] ] defines the series terms .
the remainders are quite lengthy , and are listed in appendix a. consider first the polarization mode . while possibly only marginally simpler than eq .
( c1 ) of @xcite , the series and remainder in eqs .
( [ eq : mseries1 ] ) , ( [ eq : mseries1tdef ] ) and ( [ eq : mseries1rdef ] ) naturally enable the development of a special function representation of the scattering amplitude .
the finite summation over terms like in eq .
( [ eq : mseries1tdef ] ) can be expressed using result 44.1.2 of @xcite in terms of an integral of the logarithm of the gamma function . at this juncture ,
the analysis begins to image parts of that generated in expressing the polarization properties of a magnetized vacuum via effective lagrangian or proper - time techniques @xcite , as should be expected .
hence , it is appropriate to adopt definitions from such literature as much as possible . following @xcite ,
here a definition for the generalized gamma function of @xmath40 is adopted .
properties of this function , which include , are discussed at length in @xcite and outlined in appendix b. using eq .
( [ eq : sumnlogn ] ) , one soon arrives at an expression for the scattering amplitude in terms of a handful of special ( polygamma ) functions , namely , and and its derivatives .
this representation consists of two parts , one independent of , and one that involves a limit as of the remainder in eq .
( [ eq : mseries1rdef ] ) , combined with several terms incorporating the special functions with arguments that depend on . in evaluating this limit , most terms can be handled in a straightforward manner , and standard asymptotic series ( e.g. see @xcite ) for and as prove useful .
however , the treatment of the term involving the function that appears in the limit contribution is non - trivial . a series representation for this function for large arguments is required , and is presented in eq .
( [ eq : loggamma1asymp ] ) . assembling the various pieces , the limiting result as is @xmath41 \label{eq : mperptoparparwll1}\\[-6pt ] & & \quad - \dover{1}{12b^2 } \,\psi ' \bigl(\dover{1}{2b}\bigr ) - \dover{1}{6 } + \dover{1}{6b } - \dover{4 l_1}{b } + \dover{1}{4b^2 } \bigl ( \log_e2\pi -1 -3 \log_e 2b \bigr ) \biggr\ } \nonumber\end{aligned}\ ] ] for .
this is the sought - after compact analytic form that is comparable in simplicity to the one - loop effective lagrangians calculated in @xcite . using series and asymptotic expansions for all the special functions present , it is routine to establish that for , while for
, one finds , a result obtainable from eq . ( [ eq : mperptoparparbgg1 ] ) .
the developments are similar for the mode : this representation again consists of two parts , one independent of , and one that involves a limit as of the remainder in eq .
( [ eq : mseries2rdef ] ) , combined with several terms incorporating polygamma functions with arguments that depend on .
this limit can easily be evaluated using asymptotic series to yield ( for ) @xmath42 \label{eq : mperptoperpperpwll1}\\[-6pt ] & & \quad + \dover{1}{8b^4 } \,\psi ' \bigl(\dover{1}{2b}\bigr ) + \dover{1}{3b } + \dover{1}{2b^2 } - \dover{1}{b^3 } + \dover{3}{4b^2 } \bigl ( \log_e2\pi + \log_e 2b \bigr ) \biggr\}\;\ ; .
\nonumber\end{aligned}\ ] ] using series and asymptotic expansions for all the special functions present , it is routine to establish that for , while for , one finds , the result stated in eq .
( [ eq : mperptoperpperpbgg1 ] ) .
it must be remarked in passing that the expressions for in eqs .
( [ eq : mseries1tdef ] ) , ( [ eq : mseries2rdef ] ) and ( [ eq : mperptoperpperpwll1 ] ) can not be derived from the series in eq .
( c2 ) of @xcite , principally because that series expression is divergent , and therefore erroneous .
such an error was introduced by an inappropriate rearrangement of individually - divergent contributing series ( leading to the addition of infinite contributions ) , a mistake that is avoided by the careful technique employed here in manipulating the results of appendix b in @xcite .
notwithstanding , the numerical results for the mode presented in @xcite were effectively evaluated before any series rearrangement , and therefore remain valid .
the compact analytic forms presented in eqs .
( [ eq : mperptoparparwll1 ] ) and ( [ eq : mperptoperpperpwll1 ] ) represent the culmination of the focus here .
it is the first time such simple forms for the scattering amplitudes involving just special functions have been calculated in this limit , though somewhat more convoluted , yet essentially equivalent , expressions have been put forward in @xcite . a distinct advantage of the expressions in eqs .
( [ eq : mperptoparparwll1 ] ) and ( [ eq : mperptoperpperpwll1 ] ) is the ease with which they can be accurately computed numerically .
their dependence on is illustrated in figure [ fig : scattampwll1 ] , replicating the numerics of @xcite and earlier effective lagrangian determinations @xcite , which are just as expedient since ( see eqs . [ [ eq : mperptoparparwll1_elp ] ] and [ [ eq : mperptoperpperpwll1_elp ] ] below ) they involve just integrals of elementary functions .
the low frequency result for is not presented explicitly since it reproduces that for ( e.g. see @xcite ) ; this is due to the crossing symmetries involved .
note that while the cubic dependences of all the modes at low energies reflect the lack of an energy scale in this domain ( i.e. such as pair threshold ) , the normalizations are dependent on the polarization mode , particularly at highly supercritical fields where the amplitude is highly suppressed .
this effectively represents how the rate normalization is sensitive to the ( virtual ) pair creation thresholds for the polarization states involved in a particular splitting mode . to conclude this presentation focusing on the specialization ,
an obvious objective is the re - derivation of eqs .
( [ eq : mperptoparparwll1 ] ) and ( [ eq : mperptoperpperpwll1 ] ) starting with extant and well - known effective lagrangian / proper - time ( elp ) results , and thereby demonstrating analytically the equivalence of the s - matrix formulation in the landau representation and schwinger - type formalisms in the low energy limit .
consider first the mode , for which such a determination is somewhat involved .
the starting point is the integral expression @xcite that corresponds to the scaled scattering amplitude that generates the same form for the rate as in eq .
( [ eq : totratefin ] ) : @xmath43 which has and limits matching those of eq .
( [ eq : mperptoparparwll1 ] ) . in the subsequent analysis
, it is useful to manipulate integrations using the variable .
the first step is to recognize that times the factor in curly braces in eq .
( [ eq : mperptoparparwll1_elp ] ) is a perfect derivative , namely , where .
integration by parts is obviously the operative method , with the goal of retaining functions explicitly , combined with powers of .
after some algebra , one finds that @xmath44\ ; ( s \coth s-1 ) \nonumber\\[-6pt ] \label{eq : mperptoparparwll1_elp2}\\[-6pt ] & & \qquad - \dover{1}{b } \int^{\infty}_{0 } \dover{ds}{s^3}\ , e^{-s / b}\ , \biggl[s \coth s-1 - \dover{s^2}{3 } \biggr ] \nonumber\end{aligned}\ ] ] results .
the integral on the first line can be performed using identities 3.551.3 and 3.554.4 of @xcite , yielding the gamma function and polygamma functions ( or equivalently generalized riemann zeta functions ) in addition to elementary functions .
the only subtle part pertains to the second term of this integral , namely that contributed by the factor .
this can be differentiated with respect to , evaluated to yield a function , and then the result integrated , noting the behaviour as .
the evaluation of the integral on the second line of eq .
( [ eq : mperptoparparwll1_elp2 ] ) is much more involved .
however , it has been performed before in the literature , and appears explicitly in calculations @xcite of the one - loop effective lagrangian describing refractive indices of the magnetized vacuum in qed .
hence the motivation for the particular partitioning of integrations chosen in eq .
( [ eq : mperptoparparwll1_elp2 ] ) .
details of the determination of this integral are found in dittrich et al .
@xcite , and the second line of eq .
( [ eq : mperptoparparwll1_elp2 ] ) can be equated to times the lagrangian ( see eqs . ( 2.4 ) and ( 3.16 ) of @xcite ) , thereby introducing the function .
collecting together the terms neatly generates an analytic form for that is identical to eq .
( [ eq : mperptoparparwll1 ] ) , so that the desired demonstration of equivalence of the landau representation and effective lagrangian forms is achieved . the procedure for the mode is similar , though somewhat less involved .
the equivalent scaled scattering amplitude obtained @xcite from effective lagrangian / proper - time techniques is @xmath45 recognizing that the factor in curly braces can be written as , integration by parts is again indicated , with identities 3.551.3 and 3.554.4 of @xcite again proving useful . with manipulations similar to ( but simpler than :
the function is not involved here ) those for the , a modicum of algebra leads to the derivation of eq .
( [ eq : mperptoperpperpwll1 ] ) from eq .
( [ eq : mperptoperpperpwll1_elp ] ) , as desired .
this equivalence is a satisfying indication of the verity of the landau representation analysis in this paper .
this paper has provided a detailed development of the s - matrix formulation of the qed process of magnetic photon splitting in the landau representation , focusing on the case of zero dispersion where photon propagation is collinear . the formalism in section [ sec :
formalism ] rederives and extends the exposition of mentzel , berg & wunner @xcite .
the two principal general developments offered here are an analytic reduction via the summation over the spins of the intermediate pair states , discussed briefly in @xcite , and the analytic integration over the momenta parallel to the field incorporated in the electron propagators .
this latter accomplishment is presented here for the first time .
the cumulative product of these developments is a satisfyingly simple and elegant form in eq .
( [ eq : mform ] ) for the scattering amplitude for each of the polarization modes permitted by cp invariance .
these amplitudes possess products of generalized laguerre polynomials that are common to qed processes in external magnetic fields , and elementary functions involving the photon energies and the various pair thresholds associated with the propagators .
moreover , the analytic forms presented consist of just triple summations over landau level quantum numbers of the intermediate states , and are eminently suitable for accurate numerical computations both below and above pair creation threshold .
the applicability of these results to regimes above pair threshold is a benefit of the s - matrix expansion in the landau representation that is not afforded by effective lagrangian and proper - time calculations : while these ( latter ) schwinger - type techniques elegantly formulate splitting rates below pair threshold , they eliminate the resonance structure early on in their mathematical developments , a severe limitation above . as an embellishment to these general results , specializations in two significant domains
have been obtained .
the first is for highly supercritical fields , , reproducing in particular the result of @xcite for the mode , and deriving new results for the other two modes permitted by cp invariance in the limit of zero dispersion .
the second group of asymptotic results are for energies well below pair creation threshold , where new and compact expressions for the scattering amplitudes have been derived in eqs .
( [ eq : mperptoparparwll1 ] ) and ( [ eq : mperptoperpperpwll1 ] ) in terms of the logarithm of the gamma function , its integral and their derivatives
. these two domains of specialization herein have facilitated the first analytic demonstration of the equivalence of splitting rates obtained by the s - matrix formulation in the landau representation and those derived using schwinger - type effective lagrangian / proper - time techniques .
i thank my collaborators jeanette weise , don melrose and alice harding , and also carlo graziani and george pavlov for many productive discussions , and also the referee stephen adler for suggestions that improved the presentation .
i also thank the rcfta at the university of sydney for hospitality during part of the period when this work was performed . here the remainders that appear in the series representation in eq .
( [ eq : mseries1 ] ) for the specializations to the splitting amplitudes for polarization modes and are presented : @xmath46 \log_e\bigl[k+2+\dover{1}{2b}\bigr ] \nonumber\\ & + & \dover{1}{4b } \biggl [ 11k^3 + k^2 \bigl ( 21 + \dover{21}{2b}\bigr ) + k \bigl ( 10 + \dover{14}{b } + \dover{5}{2b^2}\bigr ) + \dover{5}{2b } + \dover{5}{4b^2 } \biggr ] \log_e\bigl[k+1+\dover{1}{2b}\bigr ] \nonumber\\ & - & \dover{1}{4b } \biggl [ 11k^3 + k^2 \bigl ( 16 + \dover{21}{2b}\bigr ) + k \bigl ( 5 + \dover{9}{b } + \dover{5}{2b^2}\bigr ) + \dover{5}{4b^2 } \biggr ] \log_e\bigl[k+\dover{1}{2b}\bigr]\nonumber\\[-5.5pt ] \label{eq : mseries1rdef}\\[-5.5pt ] & - & \dover{k}{4b } \biggl [ k^2 + \dover{k}{2b } - 1 \biggr ] \log_e\bigl[k-1+\dover{1}{2b}\bigr ] \nonumber\\ & - & \biggl ( \dover{b}{3 } ( k+1 ) + \dover{1}{6 } + \dover{k+1}{2b } \biggl)\ ; \dover{1}{1 + 2(k+1)b } + \dover{k}{2b[1 + 2 kb ] } \nonumber\\ & - & \dover{9k^2}{2b } - \dover{5k}{b } - \dover{7k}{4b^2 } - \dover{5}{8b } - \dover{9}{8b^2 } - \dover{3}{4b^2}\,\log_e 2b\;\ ; , \nonumber\end{aligned}\ ] ] and @xmath47 \nonumber\\ & & -\dover{k+1}{16b } \biggl [ \dover{13}{3 } k^2 + k \bigl ( \dover{32}{3 } + \dover{7}{2b}\bigr ) + 4 + \dover{7}{2b}\biggr ] \log_e\bigl[k+2+\dover{1}{2b}\bigr ] \nonumber\\ & & + \dover{1}{16b } \biggl [ \dover{71}{3 } k^3 + k^2 \bigl ( 38 + \dover{9}{2b}\bigr ) + k \bigl ( \dover{79}{3 } - \dover{4}{b}\bigr ) + 12 - \dover{2}{b}\biggr ] \log_e\bigl[k+1+\dover{1}{2b}\bigr ] \nonumber\\ & & -\dover{1}{16b } \biggl [ \dover{71}{3 } k^3 + k^2 \bigl ( 13 + \dover{9}{2b}\bigr ) - k \bigl ( \dover{20}{3 } - \dover{13}{b}\bigr ) + \dover{13}{2b}\biggr ] \log_e\bigl[k+\dover{1}{2b}\bigr ] \nonumber\\[-6pt ] \label{eq : mseries2rdef}\\[-6pt ] & & + \dover{k}{16b } \biggl [ \dover{13}{3 } k^2 + k \bigl ( 2 + \dover{7}{2b}\bigr ) - \dover{19}{3}\biggr ] \log_e\bigl[k-1+\dover{1}{2b}\bigr ] \nonumber\\ & & \;\ ; + \dover{k(k-1)(k-2)}{12b}\ ; \log_e\bigl[k-2+\dover{1}{2b}\bigr ] + \dover{(k+1)(k+2)^2}{2[1 + 2(k+2)b ] } \nonumber\\ & & \;\ ; - \dover{(k+1)^2(5k+2)}{2[1 + 2(k+1)b ] } - \dover{k^2(5k-1)}{2[1 + 2k b ] } + \dover{k(k-1)^2}{2[1 + 2(k-1)b ] } \nonumber\\ & & \quad + \dover{7k^2}{4b } + \dover{7k}{4b } + \dover{9}{8b } + \dover{k+1}{b^2 } + \dover{3}{4b^2}\,\log_e2b\;\ ; .
\nonumber\end{aligned}\ ] ] in this appendix , various useful properties of the function , the integral of the logarithm of the gamma function , that are needed in the specializations are stated .
given the definition of in eq .
( [ eq : gamma1def ] ) , it is elementary to establish , using 44.1.2 of @xcite , that @xmath48 taking successive derivatives with respect to , one quickly arrives at well - known finite series representations of and its logarithmic derivative ; see @xcite for discussions of these functions and their series representations .
an asymptotic series representation for the function for large arguments is useful , and can be derived with the aid of the following series representation ( see result 8.343.2 of @xcite ) for the logarithm of the gamma function : @xmath49 \label{eq : loggammaseries}\ ] ] from which stirling s asymptotic expansion can be derived .
here , is the generalized riemann zeta function , defined in 9.511 and 9.521.1 of @xcite .
the integration of this series is effected using the identity , and is mostly uneventful .
however , the treatment of the term in the summation is somewhat more subtle , due to the singular nature of , and requires taking the limit , assuming to be a continuous variable .
then result 8.362.1 of @xcite comes in handy , and the series identity @xmath50 \nonumber\\[-6pt ] \label{eq : loggamma1series}\\[-6pt ] & & \quad + \dover{1}{2 } \sum_{m=2}^{\infty } \dover{1}{(m+1)(m+2)}\ ; \biggl [ \zeta ( m,1 ) - \zeta ( m , x ) - 1 + \dover{1}{x^m } \biggr ] \nonumber\end{aligned}\ ] ] follows , where is euler s constant .
this series , which adequately substitutes for an asymptotic representation , can be used very effectively for numerical evaluations for all . for the range , this series also effects accurate evaluation of via use of the recurrence relation , an identity derivable from eq .
( [ eq : gamma1def ] ) with the aid of 6.441.3 in @xcite . for large
, it then follows that @xmath51 where @xmath52 \label{eq : l1def}\\[-6pt ] & \equiv & \lim_{k\to\infty } \biggl\ { \sum_{n=1}^k n\,\log_e n - \biggl ( \dover{k(k+1)}{2 } + \dover{1}{12 } \biggr ) \ , \log_e k + \dover{k^2}{4 } \biggr\ } \nonumber\end{aligned}\ ] ] with numerical value .
this is just the constant appearing in the magnetized vacuum polarization analyses of @xcite , where the raabe integral form for it can be found .
the second definition of in eq . ( [ eq : l1def ] ) can be obtained by setting in eq .
( [ eq : sumnlogn ] ) , and is a result noted by @xcite .
m. mentzel , d. berg & g. wunner , phys . rev . *
d50 * , 1125 ( 1994 ) .
s. l. adler , j. n. bahcall , c. g. callan & m. n. rosenbluth , phys .
lett . * 25 * , 1061 ( 1970 ) . z. bialynicka - birula & i. bialynicki - birula , phys
d2 * , 2341 ( 1970 ) .
s. l. adler , ann .
ny * 67 * , 599 ( 1971 ) .
v. o. papanyan & v. i. ritus , sov .
jetp * 34 * , 1195 ( 1972 ) . r. j. stoneham , j. phys .
* a12 * , 2187 ( 1979 ) . v. n. baier , a. i. milstein & r. zh .
shaisultanov , sov .
jetp * 63 * , 665 ( 1986 ) . v. n. baier , a. i. milstein & r. zh .
shaisultanov , phys .
* 77 * , 1691 ( 1996 ) . s. l. adler & c. schubert , phys .
lett . * 77 * , 1695 ( 1996 ) .
m. g. baring & a. k. harding , astrophys .
j. * 482 * , 372 ( 1997 ) .
j. i. weise , m. g. baring & d. b. melrose , phys rev .
d. * 57 * , 5526 ( 1998 ) . c. wilke & g. wunner , phys .
d. * 55 * , 997 ( 1997 ) .
d. b. melrose & a. j. parle , aust .
* 36 * , 775 ( 1983 ) .
d. b. melrose & a. j. parle , aust .
* 36 * , 799 ( 1983 ) .
d. b. melrose & a. j. parle , aust .
* 36 * , 755 ( 1983 ) .
tsai & t. erber , phys .
d. * 10 * , 492 ( 1974 ) .
j. k. daugherty & a. k. harding , astrophys .
j. * 273 * , 761 ( 1983 ) .
mitrofanov , i. g. , et al .
, sov . astron . *
30 * , 659 ( 1986 ) .
baring , m. g. , mnras * 235 * , 79 ( 1988 ) .
m. g. baring , astrophys .
* 440 * , l69 ( 1995 ) .
m. g. baring & a. k. harding , astr .
space sci . * 231 * , 77 ( 1995 ) .
a. k. harding & m. g. baring , proc .
huntsville grb workshop ( aip 384 , new york ) , p. 941
chang , k. chen , e. e. fenimore , & c. ho , proc .
huntsville workshop on gamma - ray bursts , ( aip 384 , new york ) , p. 921
kouveliotou , c. , et al . , nature * 393 * , 235 ( 1999 ) .
hurley , k. , et al .
, nature * 397 * , 41 ( 1999 ) .
kouveliotou , c. , et al . , astrophys .
* 510 * , l116 ( 1999 ) .
m. g. baring & a. k. harding , astrophys .
* 507 * , l55 ( 1998 ) .
a. k. harding , m. g. baring & p. l. gonthier , astrophys .
j. * 476 * , 246 ( 1997 ) .
a. a. sokolov & i. m. ternov , _ synchrotron radiation _
( pergamon , oxford , 1968 ) .
h. herold , h. ruder & g. wunner , astr .
astrophys . * 115 * , 90 ( 1982 ) .
shabad , a. e. , ann . phys . *
90 * , 166 ( 1975 ) .
bulik , t. & miller , m. c. , mnras * 288 * , 596 ( 1997 )
. i. s. gradshteyn & i. m. ryzhik , _ table of integrals , series , and products _ ( academic , new york , 1980 ) .
d. b. melrose & r. j. stoneham , j. phys .
a. * 10 * , 1211 ( 1977 ) .
g. g. pavlov , yu . a. shibanov & d. g. yakovlev , astr .
space sci . *
73 * , 33 ( 1980 ) .
e. r. hansen , _ a table of series and products _
( prentice - hall , englewood cliffs , nj ) .
tsai & t. erber , phys .
d. * 12 * , 1132 ( 1975 ) . w. dittrich , w .- y .
tsai , & k .- h .
zimmermann , phys .
d. * 19 * , 2929 ( 1979 ) .
b. v. ivanov , physics lett .
b * 282 * , 228 ( 1992 ) .
l. bendersky , acta math . * 61 * , 263 ( 1933 ) .
j. s. heyl & j. hernquist , phys rev .
d. * 55 * , 2449 ( 1997 ) .
a. erdlyi , higher transcendental functions , volumes i iii , ( krieger publishing , malabar , florida ) | calculations of reaction rates for the third - order qed process of photon splitting in strong magnetic fields traditionally have employed either the effective lagrangian method or variants of schwinger s proper - time technique .
recently , mentzel , berg and wunner @xcite presented an alternative derivation via an s - matrix formulation in the landau representation .
advantages of such a formulation include the ability to compute rates near pair resonances above pair threshold .
this paper presents new developments of the landau representation formalism as applied to photon splitting , providing significant advances beyond the work of @xcite by summing over the spin quantum numbers of the electron propagators , and analytically integrating over the component of momentum of the intermediate states that is parallel to field .
the ensuing tractable expressions for the scattering amplitudes are satisfyingly compact , and of an appearance familiar to s - matrix theory applications .
such developments can facilitate numerical computations of splitting considerably both below and above pair threshold .
specializations to two regimes of interest are obtained , namely the limit of highly supercritical fields and the domain where photon energies are far inferior to that for the threshold of single - photon pair creation . in particular , for the first time the low - frequency amplitudes are simply expressed in terms of the gamma function , its integral and its derivatives .
in addition , the equivalence of the asymptotic forms in these two domains to extant results from effective lagrangian / proper - time formulations is demonstrated .
# 1@xmath0 # 1#2 # 10= -0.0125em0 - 0 0.025em0 - 0 -0.0125em0.0433em0 = cmr5 = cmbx5 = cmr7 = cmbx7 = cmr8 = cmmi8 = cmsy8 = cmbx8 = cmtt8 = cmti8 = cmsl8 = cmr6 = cmmi6 = cmsy6 = cmbx6 |
this work was partially supported by the ministry of science and technology of taiwan r.o.c .
, under grant most-103 - 2112-m-006 - 004-my3 ( chc ) .
s. n. gninenko and n. v. krasnikov , phys .
b * 513 * , 119 ( 2001 ) [ hep - ph/0102222 ] .
p. fayet , phys .
d * 75 * , 115017 ( 2007 ) [ hep - ph/0702176 [ hep - ph ] ] .
n. arkani - hamed , d. p. finkbeiner , t. r. slatyer and n. weiner , phys .
d * 79 * , 015014 ( 2009 ) [ arxiv:0810.0713 [ hep - ph ] ] .
m. pospelov , phys .
d * 80 * , 095002 ( 2009 ) [ arxiv:0811.1030 [ hep - ph ] ] .
h. davoudiasl , h. s. lee and w. j. marciano , phys .
d * 89 * , no . 9 , 095006 ( 2014 ) [ arxiv:1402.3620 [ hep - ph ] ] .
k. petraki , l. pearce and a. kusenko , jcap * 1407 * , 039 ( 2014 ) [ arxiv:1403.1077 [ hep - ph ] ] .
t. araki , f. kaneko , t. ota , j. sato and t. shimomura , phys .
d * 93 * ( 2016 ) no.1 , 013014 [ arxiv:1508.07471 [ hep - ph ] ] .
s. baek , phys .
b * 756 * , 1 ( 2016 ) [ arxiv:1510.02168 [ hep - ph ] ] .
h. s. lee and s. yun , phys .
d * 93 * , no . 11 , 115028 ( 2016 ) [ arxiv:1604.01213 [ hep - ph ] ] .
ko and y. tang , arxiv:1608.01083 [ hep - ph ] .
a. j. krasznahorkay _ et al .
_ , phys .
lett . * 116 * , no . 4 , 042501 ( 2016 ) [ arxiv:1504.01527 [ nucl - ex ] ] . j. l. feng , b. fornal , i. galon , s. gardner , j. smolinsky , t. m. p.
tait and p. tanedo , arxiv:1604.07411 [ hep - ph ] .
p. h. gu and x. g. he ,
arxiv:1606.05171 [ hep - ph ] . l. b. chen , y. liang and c. f. qiao , arxiv:1607.03970 [ hep - ph ] .
y. liang , l. b. chen and c. f. qiao , arxiv:1607.08309 [ hep - ph ] .
p. fayet , nucl .
b * 187 * , 184 ( 1981 ) .
m. suzuki , phys .
lett . * 56 * , 1339 ( 1986 ) .
v. m. braun and i. e. filyanov , z. phys .
c * 48 * , 239 ( 1990 ) [ sov . j. nucl .
phys . * 52 * , 126 ( 1990 ) ] [ yad .
fiz . * 52 * , 199 ( 1990 ) ] .
p. ball , jhep * 9901 * , 010 ( 1999 ) [ hep - ph/9812375 ] .
p. ball , jhep * 9809 * , 005 ( 1998 ) doi:10.1088/1126 - 6708/1998/09/005 [ hep - ph/9802394 ] . c. h. chen and h. hatanaka , phys .
d * 73 * , 075003 ( 2006 ) [ hep - ph/0602140 ] .
n. carrasco , p. lami , v. lubicz , l. riggio , s. simula and c. tarantino , phys .
d * 93 * , no .
11 , 114512 ( 2016 ) [ arxiv:1602.04113 [ hep - lat ] ] . k.a .
( particle data group ) , chin .
c , 38 , 090001 ( 2014 ) . | the inconsistent conclusions for a light gauge boson @xmath0 production in the @xmath1 exist in the literature .
it is found that the process can be generated by the tree - level @xmath2-boson annihilation and loop - induced @xmath3 .
we find that it strongly depends on the @xmath4 limit or the unique gauge coupling to the quarks , whether the @xmath5 decay , which is from the @xmath2-boson annihilation , is suppressed by @xmath6 ; however , no such suppression is found via the loop - induced @xmath7 .
the constraints on the relevant couplings are studied .
a light gauge boson @xmath0 has been studied widely due to various motivations @xcite . especially , a mass of around 17 mev boson with more than @xmath8 significance is indicated by the measurement of @xmath9 angular correlations in @xmath10 transition @xcite , where the implications of this light gauge boson are investigated @xcite .
however , contradictory conclusions associated with the @xmath0 emission in rare @xmath11 decay occur in the literature , where the authors @xcite concluded that the longitudinal component in the @xmath5 decay is enhanced by @xmath12 , but the authors @xcite claimed that this decay amplitude should be suppressed by @xmath13 . in this study
, we revisit this issue by exploring the situations with and without the @xmath4 limit and unique gauge coupling , when the @xmath14 transition arises from the @xmath2-boson annihilation .
in addition , we also study the contributions from the loop - induced flavor - changing neutral current ( fcnc ) process @xmath3 .
since our purpose is to investigate the property of a light gauge boson emission from @xmath14 transition , we do not focus on a specific gauge model .
instead , we study the case that the @xmath0-boson vectorially couples to the standard model ( sm ) quarks and the interactions are dictated by : @xmath15 in general , the couplings @xmath16 are flavor dependence , and the fcncs at the tree level are then induced .
since we have less knowledge on the flavor mixings , the associated fcnc parameters are completely free ; therefore , we skip the discussions on the tree - level fcnc effects in this work by assuming that they are small , or they can be always constrained by the low energy physics .
eq .
( [ eq : lx ] ) can be applied to dark photon models if @xmath16 for the same type of quarks are the same @xcite . in order to demonstrate the characteristics of the @xmath5 decay
, we analyze the hadronic effects with leading - twist parton distribution amplitudes ( das ) for the @xmath17 and @xmath11 mesons . as usual , the twist-2 da of a pseudoscalar meson is defined by @xcite : @xmath18 where @xmath19 , @xmath20 , and @xmath21 is the decay constant of a meson @xmath22 .
the da can be expanded by gegenbauer polynomials as : @xmath23 where the gegenbauer moments @xmath24 for the @xmath17 and @xmath11 mesons at @xmath25 gev are @xmath26 , @xmath27 , @xmath28 , @xmath29 , and @xmath30 @xcite .
it can be seen that due to the su(3 ) breaking , the odd moments in the @xmath11 meson do not vanish . to calculate the @xmath0 emission from the @xmath11 and @xmath17 mesons , we adopt the spin structure for incoming meson as @xcite : @xmath31 where @xmath32 is the number of colors ; and the spin structure for outgoing meson can be obtained by using @xmath33 instead of @xmath34 . according to eq .
( [ eq : da ] ) , the @xmath21 can be obtained by taking the trace in spinor space as : @xmath35 where the @xmath36 in the numerator is from the sum of color charges of quark line @xmath37 q_{1\alpha}$ ] .
decay , where the square boxes and numbers denote the possible places to emit the @xmath0-boson.,width=283 ] with the couplings in eq .
( [ eq : lx ] ) , the @xmath5 decay can arise from the tree and one - loop diagrams .
since the hadronic effects from the tree level are more copious than those from one - loop , we first discuss the tree contributions in detail .
the flavor diagrams for the @xmath1 decay are shown in fig .
[ fig : kpix ] , where the square boxes and numbers denote the possible places to emit the @xmath0-boson . according to eq .
( [ eq : da ] ) , the decay amplitude for the @xmath0-boson emitting from place-1 can be derived as : @xmath38 it is found that the decay amplitude for the @xmath0-boson emitting from place-@xmath39 can be formulated as : @xmath40 where @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46 , and @xmath47 are defined as : @xmath48 in order to understand the @xmath4 limit and @xmath16 dependence of @xmath49 , we study the cases by requiring exact / partial @xmath4 limit and different / same @xmath16 . :
+ since @xmath50 and @xmath51 are the multiplier factors , they are irrelevant to the discussions of @xmath4 limit ; therefore , we do not need to set @xmath52 . due to the kinematics
, @xmath53 has to be satisfied ; thus , we need to leave the factors @xmath54 and @xmath55 , which appear in the denominators of eq .
( [ eq : r_j ] ) , alone . from eqs .
( [ eq : a_j ] ) and ( [ eq : r_j ] ) , we then get : @xmath56 and @xmath57 .
we consider @xmath58 because the @xmath59 and @xmath60 involve the same gauge coupling @xmath61 .
one can alternatively use @xmath62 depending on the convenience .
it is clearly seen that the decay amplitude for the @xmath5 is proportional to @xmath63 .
this result can match the conclusions given in @xcite .
we note that when calculating @xmath64 , we have used the property @xmath65 , where this condition is suitable for the @xmath17 meson , and it is violated in the @xmath11 meson due to the @xmath4 breaking . as a result , the leading - twist contributions can not lead to an interesting result in the @xmath4 limit . furthermore
, if we further set @xmath66 , it can be found that @xmath67 .
+ when we release the condition @xmath68 , the first term in the numerator of eq .
( [ eq : a1pa3 ] ) inside the integral becomes @xmath69 , and the denominator is @xmath70 . with @xmath71 ,
we find : @xmath72 where the @xmath73 dependence of numerator and denominator in the integral is cancelled , and @xmath74 is applied . by this analysis ,
it is clear that when we put back the @xmath4 breaking effect with @xmath75 , the decay amplitude is not proportional to @xmath63 anymore .
this result confirms the conclusions given in @xcite . furthermore , if we take all gauge couplings to be the same and @xmath76 , we find that @xmath77 is still satisfied
. we can understand the cancellations from another viewpoint : by using @xmath78 , from eqs .
( [ eq : a_j ] ) and ( [ eq : r_j ] ) it can be easily found that @xmath79 for @xmath80 and @xmath81 for @xmath82 .
+ with partial @xmath4 limit , which is conditioned by @xmath83 , we have seen that the decay amplitude for the @xmath5 is not suppressed by @xmath63 ; however , it diminishes when @xmath84 .
it is intriguing to see whether the cancellations work or not when the @xmath4 breaking effects are taken into account in the da of the @xmath11 meson . from eqs .
( [ eq : a_j ] ) and ( [ eq : r_j ] ) , we can easily get the result @xmath85 when the @xmath61 and @xmath73 in @xmath60 are replaced by the @xmath86 and @xmath87 .
the connection between @xmath60 and @xmath88 is based on the property @xmath89 , where the odd gegenbauer moments vanish .
that is , if @xmath90 , the cancellation between @xmath60 and @xmath88 still works . unlike the da of pion , @xmath91 due to nonvanishing odd gegenbauer moments ,
e.g. , @xmath92 . hence , we have : @xmath93 \label{eq : a1pa2}\ ] ] in order to examine eq .
( [ eq : a1pa2 ] ) , we simplify the analysis by taking the limit @xmath94 .
thus , with @xmath95 , we find : @xmath96 according to the result , it can be seen that the decay amplitude for the @xmath5 with @xmath97 is @xmath98 , where the function @xmath99 is from the integration in @xmath73 and only depends on the @xmath63 .
we then conclude that if the @xmath0 gauge couplings to the quarks are the same , the decay amplitude of the @xmath5 process from the leading - twist da is suppressed by @xmath100 .
it is of interest to examine the @xmath0 emission from @xmath2-boson propagator shown in fig .
[ fig : kpix ] . to estimate the contribution , we parametrize the lorentz covariant gauge coupling @xmath101 to be : @xmath102 w^{-\alpha } w^{+\beta}x^\mu\,,\ ] ] where @xmath103 is the trilinear gauge coupling ; @xmath104 , @xmath105 , and @xmath106 are the momenta of @xmath107 , @xmath108 , and @xmath0 gauge bosons , respectively , and the momenta are chosen to flow into the vertex . accordingly , the decay amplitude for the @xmath5 via the trilinear coupling can be obtained as : @xmath109 it is clear that the contribution from the coupling @xmath101 is suppressed by @xmath100 .
this result is nothing to do with the @xmath4 limit . without the @xmath4 limit
, it can be seen from eq .
( [ eq : r_j ] ) that the numerators of @xmath110 are related to @xmath111 and those of @xmath112 are associated with @xmath113 .
since @xmath114 is around 3.5 times larger than @xmath115 , numerically , the values of @xmath116 are one order of magnitude smaller than those of @xmath117 . as a result ,
the @xmath118 is sensitive to the difference between @xmath86 and @xmath61 . to present the numerical analysis
, we adopt @xmath61 and @xmath119 as the free parameters and set @xmath120 .
we show the contours for @xmath118 ( in units of @xmath121 ) as a function of @xmath61 and @xmath122 ( in units of @xmath123 ) in fig .
[ fig : wa ] , where we have used @xmath124 gev and @xmath125 gev , and the numbers on the lines denote the values of @xmath118 . with the assumption of @xmath126 ,
we obtain @xmath127 , where the current measurement is @xmath128 @xcite .
therefore , the dashed lines in the plot can be regarded as the central value of the experimental measurement . from the figure
, it can be seen that @xmath129 can not be larger than @xmath130 . in units of @xmath121 as a function of @xmath61 and @xmath131 in units of @xmath123 , where the numbers on the lines denote the values of @xmath118 , and dashed lines are the central value of @xmath132 . , width=302 ] besides the tree - level @xmath2-boson annihilation , the fcnc coupling @xmath133 , which is induced from one - loop and is depicted in fig . [
fig : loop ] , can also contribute to the @xmath5 process .
according to the interactions in eq .
( [ eq : lx ] ) , the effective coupling of @xmath133 from each up - type quark loop can be derived as : @xmath134 where we have dropped the small effects from @xmath135 , the factor @xmath136 is from the mass inserted twice in @xmath137-quark propagator , and @xmath138 is the loop integral . from eq .
( [ eq : sdx ] ) , the associated cabibbo - kobayashi - maskawa ( ckm ) matrix elements for top - quark loop are @xmath139 , due to the enhancement of @xmath140 , its contribution is comparable to that from charm - quark , in which the essential factor is @xmath141 .
the contribution from @xmath142-quark loop can be ignored because of the @xmath143 suppression .
additionally , it can be clearly seen that although the @xmath0 couplings to quarks are vector - like , the induced coupling @xmath144 indeed is chiral . .
, width=283 ] unlike the @xmath2-boson annihilation shown in fig .
[ fig : kpix ] , the dominant hadronic effect for the @xmath5 decay from @xmath144 interaction is formulated by @xcite : @xmath145 where @xmath146 at @xmath147 . as a result ,
the corresponding transition amplitude and br are respectively given by : @xmath148 it can be seen that eq .
( [ eq : kpix_loop ] ) is not suppressed by @xmath100 but enhanced by @xmath149 . to show the bounds on the gauge couplings @xmath150 and @xmath151 independently , we present the numerical values of eq .
( [ eq : kpix_loop ] ) as a function of @xmath150 and @xmath151 in fig .
[ fig : br_loop ] , where @xmath152 mev is used , the solid ( dashed ) line stands for the result of @xmath153 in units of @xmath154 , and the horizontal dotted line is the central value of @xmath155 . with @xmath126 , the bounds from penguin diagrams are stronger than those from the @xmath2-boson annihilations . therefore , we confirm the conclusion given in @xcite , where the effective coupling arisen from the penguin diagrams gets a stricter bound .
we note that the @xmath0-boson emitting from the @xmath2 propagator shown in fig .
[ fig : loop ] can also contribute to @xmath5 ; however , due to the @xmath156 suppression , we neglect its contribution in the numerical analysis . in summary ,
a light gauge boson predominantly decaying to @xmath9 in rare @xmath11 decay is studied .
the process @xmath157 can be generated from the tree and penguin diagrams .
it is found that the decay amplitude for @xmath5 from @xmath2-boson annihilation can be directly proportional to @xmath158 when the @xmath4 limit is applied , or the gauge couplings satisfy @xmath159 .
when these conditions are relaxed , it is found that if @xmath61 is of the @xmath160 , @xmath129 has to be less than @xmath130 .
by contrast , the @xmath63 suppression is not found in the loop - induced @xmath5 process .
we show that the loop effects indeed give more severe bounds on the gauge couplings @xmath150 and @xmath151 .
our results are consistent with the conclusions given in @xcite . as a function of @xmath16 ( in units of @xmath154 ) , where the solid and dashed lines denote the contributions from top- and charm - quark loops , respectively .
the horizontal dotted line is the central value of @xmath155 .
, width=302 ] |
low - dimensional spin-1/2 systems have been extensively studied during the last decade due to strong quantum fluctuations that may lead to unusual ground states and low - temperature properties . magnetic frustration in low - dimensional spin systems
additionally suppresses long - range ordering and results in fascinating phenomena such as the formation of a resonating valence bond ( rvb ) spin - liquid ground state @xcite or the magnetoelectrical effect in spin spirals.@xcite recent studies @xcite indicated the importance of combining experimental ( measurement of thermodynamic properties , neutron and magnetic resonance studies ) and computational ( microscopic modeling via electronic structure calculations ) approaches to gain an understanding of the unusual physical properties of low - dimensional spin-1/2 systems .
vanadium phosphates present a promising and mainly unexplored field for the search of novel low - dimensional ( and , possibly , frustrated ) spin systems .
structural reports on vanadium phosphates are often accompanied by magnetic susceptibility data @xcite but few of the systems gained thorough physical investigation , the notable examples being ( vo)@xmath0p@xmath0o@xmath1 ( refs . ) and vo(hpo@xmath2h@xmath0o.@xcite the careful study of these compounds revealed complex superexchange pathways and showed the failure of the straightforward interpretation of experimental results on the structural basis only.@xcite recently , we studied magnetic properties of a novel vanadium phosphate pb@xmath0vo(po@xmath3)@xmath0 and found an interesting realization of the frustrated square lattice system with both ferro- and antiferromagnetic interactions resulting in columnar antiferromagnetic ordering.@xcite thus , vanadium phosphates can reveal very unusual spin systems , and a detailed study of the respective compounds is of high interest .
below , we present the investigation of the magnetic properties of a silver vanadium phosphate ag@xmath0vop@xmath0o@xmath1.@xcite we use both experimental and computational techniques in order to achieve a reliable description of the exchange couplings in the system under investigation . the results are discussed with respect to the structural correlations for magnetic interactions in vanadium phosphates .
the outline of the paper is as follows . in sec .
[ structure ] , we briefly describe the crystal structure and analyze the possible pathways of superexchange interactions .
methodological aspects are given in sec .
[ method ] .
[ experimental ] deals with magnetic susceptibility data , while sec .
[ band ] presents the results of band structure calculations and estimates for the exchange integrals .
experimental and computational results are compared and discussed in sec .
[ discussion ] followed by our conclusions .
the crystal structure of ag@xmath0vop@xmath0o@xmath1 is formed by [ vop@xmath0o@xmath1 ] layers separated by silver cations .
the structure has monoclinic symmetry ( @xmath5 , @xmath6 a , @xmath7 a , @xmath8 a , @xmath9 , @xmath10 ) .
one unit cell contains two layers and , in particular , four vanadium atoms two from each of the layers ( fig .
[ fig - structure ] ) .
every layer includes vo@xmath11 octahedra joined by po@xmath3 tetrahedra .
vanadium has the oxidation state of + 4 corresponding to the electronic configuration @xmath12 , i.e. , @xmath13.@xcite magnetic measurements down to 77 k indicate paramagnetic behavior of ag@xmath0vop@xmath0o@xmath1,@xcite but , as we will show in the following , considerable deviation from the curie law starts immediately below this temperature .
to understand the possible magnetic interactions in ag@xmath0vop@xmath0o@xmath1 , one has to consider the local environment of vanadium atoms as well as the connections between vanadium polyhedra .
the vo@xmath11 octahedra are strongly distorted due to the formation of a short v
o bond typical for v@xmath14 ( see ref . ) .
the distortion of the octahedron results in a non - degenerate orbital ground state with the half - filled @xmath15 orbital lying in the equatorial plane of the octahedron,@xcite i.e. , almost parallel to the [ vop@xmath0o@xmath1 ] layers . the orientation of the half - filled orbital implies that interlayer interactions should be negligibly small while the magnitude of in - layer interactions depends on the connections between the vanadium octahedra . according to fig .
[ fig - structure ] , v o v superexchange is impossible in ag@xmath0vop@xmath0o@xmath1 ; therefore , all the superexchange pathways must include po@xmath3 tetrahedra .
assuming that the interactions via two consecutive tetrahedra ( i.e. , the pyrophosphate p@xmath0o@xmath1 group ) are weak , we derive three possible couplings : @xmath16 , and @xmath17 indicating a 1d spin system .
one can consider such a system as an alternating zigzag chain ( see bottom panel of fig .
[ fig - structure ] ) , since two nearest - neighbor interactions @xmath16 are not equivalent by symmetry , while @xmath17 is the next - nearest neighbor interaction .
interchain couplings are weak as they include p@xmath0o@xmath1 groups or single tetrahedra that do not match the half - filled orbital of vanadium . below ( sec . [ band ] ) , we justify the weakness of the interchain couplings by analyzing the local density approximation ( lda ) band structure with a tight - binding model that accounts for _ all _ the interactions between nearest and next - nearest neighbors .
( color online ) crystal structure of ag@xmath0vop@xmath0o@xmath1 ( upper panels ) and a schematic image of an alternating zigzag chain ( bottom panel ) .
the upper left panel shows a single [ vop@xmath0o@xmath1 ] layer , while the upper right panel presents the stacking of the layers . in the upper left panel ,
faces of the two vo@xmath11 octahedra are removed in order to illustrate the orientation of short v
o bonds ( shown as thicker lines ) .
solid , dashed , and dotted lines in the bottom panel denote @xmath16 , and @xmath17 , respectively . ]
polycrystalline samples of ag@xmath0vop@xmath0o@xmath1 were prepared according to the procedure reported in ref . .
stoichiometric amounts of agno@xmath18 , ( nh@xmath19hpo@xmath3 , and vo@xmath0 were carefully mixed , placed in corundum crucibles and annealed in air .
the regime of the annealing was as follows : 12 hours at 200 @xmath20c , 1 day at 400 @xmath20c , and 1 day at 500 @xmath20c with regrindings after each step .
the final product was a single phase as checked by x - ray diffraction ( xrd ) ( huber g670f camera , cuk@xmath21 radiation , imageplate detector ) .
however , the xrd pattern revealed a rather high background signal that probably indicated low crystallinity of the prepared samples or the presence of amorphous impurities .
unfortunately , we failed to improve the quality of the samples , since further annealings ( in air or in vacuum ) did not lead to any notable changes in the xrd patterns or resulted in the appearance of non - identified impurity peaks .
we also tried to prepare an isostructural compound na@xmath0vop@xmath0o@xmath1 ( ref . ) but the samples always contained a mixture of two polymorphic modifications [ one is isostructural to ag@xmath0vop@xmath0o@xmath1 while the other one has fresnoite - type structure ( see ref . ) ] .
magnetic susceptibility was measured using quantum design squid between 2 and 300 k in the fields @xmath22 of 0.5 , 1 , and 5 t. scalar - relativistic density - functional ( dft ) band structure calculations were performed using a full - potential local - orbital scheme@xcite ( fplo5.00 - 19 ) and the exchange - correlation potential of perdew and wang.@xcite ag@xmath23 , v@xmath24 , p@xmath25 , and o@xmath26 orbitals were employed as the basis set , while all lower - lying orbitals were treated as core states .
the @xmath27 mesh included 864 points within the first brillouin zone ( 296 in its irreducible part ) .
the convergence with respect to the number of @xmath27 points was carefully checked .
first , a lda calculation was performed employing the crystallographic unit cell and the full symmetry of the crystal structure ( @xmath5 ) .
orbital states relevant for magnetic interactions were chosen on the basis of the lda band structure and the respective bands were analyzed using a tight - binding model . such analysis provided an estimate for the magnitudes of all nearest - neighbor and next - nearest - neighbor exchange couplings in ag@xmath0vop@xmath0o@xmath1 .
we found that the values of interest were @xmath28 ( see fig .
[ fig - structure ] ) , in agreement with the empirical structural considerations listed in sec . [ structure ] .
next we turned to the lsda+@xmath29 technique in order to calculate total energies for several patterns of spin ordering and to give an independent estimate of the leading exchange interactions in ag@xmath0vop@xmath0o@xmath1 .
such an estimate requires doubling of the lattice translation in the @xmath30 direction ( see section [ band ] ) .
the doubling of the crystallographic unit cell results in @xmath31 and 104 atoms making the calculations very time - consuming and less accurate .
therefore , we slightly simplified the crystal structure in our lsda+@xmath29 calculations .
the simplification was justified by the similarity of lda band structures and tight - binding models for the crystallographic unit cell and the supercell ( i.e. , the modified cell ) .
the neighboring layers in ag@xmath0vop@xmath0o@xmath1 are not translationally equivalent , but they _ are _ equivalent by symmetry , hence they reveal identical in - layer interactions .
interlayer interactions are very weak ( see sec . [ band ] ) , therefore the unit cell with one layer is actually sufficient for the present computational purpose .
we use a @xmath32 supercell with @xmath33 a , @xmath34 a ( @xmath35 ) , @xmath36 a , and @xmath37 .
the supercell includes one [ vop@xmath0o@xmath1 ] layer that _ exactly _ matches the layer in the real structure .
however , the local environment of silver atoms is changed , and the @xmath38 parameter is slightly elongated as compared to @xmath39 in order to avoid unreasonably short ag o distances .
the modification of the crystal structure is a strong but _
well justified _ approximation , since silver atoms do not take part in superexchange pathways , while the geometry of the v
o framework ( crucial for magnetic interactions ) is unchanged .
these conclusions are supported by similar tight - binding fits of the relevant bands in lda band structures for the crystallographic unit cell and the supercell .
the variation of the respective hopping parameters does not exceed 5% , hence the error for @xmath40 s is below 10% ( i.e. , within the typical accuracy of lsda+@xmath29 ) .
another problematic point of lsda+@xmath29 calculations deals with the choice of the coulomb repulsion parameter @xmath41 [ we use the notation @xmath41 in order to distinguish this parameter of the computational method from the model parameter @xmath42 for the coulomb repulsion in the effective one - band model ( see sec . [ band ] ) ] .
below , we use several physically reasonable values of @xmath41 ( namely , 4 , 5 , and 6 ev ) in order to check the influence of @xmath41 on the resulting exchange integrals .
the exchange parameter of lsda+@xmath29 is fixed at @xmath43 ev , as usually it has no significant influence on the results .
magnetic susceptibility curves for ag@xmath0vop@xmath0o@xmath1 are shown in fig .
[ fig - suscept ] .
the curves reveal a maximum at @xmath44 k and a fast decrease from @xmath45 down to 5 k indicating spin - gap - like behavior .
below 5 k the susceptibility increases ( at low fields ) or remains nearly temperature - independent ( at 5 t ) .
the samples under investigation are powders , therefore the low - temperature part of the data is dominated by the contribution of paramagnetic impurities and defects saturated at high field .
the susceptibility curves above 2.2 k do not show any anomalies that could denote long - range magnetic ordering.@xcite a weak bend is present on the low - field curve below 2.2 k , but the present data ( collected above 2 k ) are not sufficient to reveal the origin of this bend .
thus , the susceptibility data for ag@xmath0vop@xmath0o@xmath1 reveal low - dimensional spin - gap - like behavior consistent with the scheme of magnetic interactions proposed in sec .
[ structure ] .
the alternation of nearest - neighbor interactions @xmath46 and @xmath47 causes the formation of the spin gap .
the high - temperature part of the susceptibility ( above 60 k ) shows the curie - weiss type behavior , although the @xmath48 _ vs. _ @xmath49 curves are nonlinear due to considerable temperature - independent contribution .
we fit the data above 100 k as @xmath50 and find @xmath51 emu / mol , @xmath52 emu@xmath53k / mol , and @xmath54 k. the curie constant @xmath55 corresponds to the effective moment of @xmath56 that is slightly lower than the expected spin - only value of @xmath57 .
the discrepancy may be caused by the presence of nonmagnetic impurities in the samples under investigation , and this explanation is additionally supported by high diamagnetic contribution @xmath58 as well as by the slightly decreased @xmath59 value ( see below ) .
we suggest that the high diamagnetic contribution is caused by the preparation procedure , since the samples are heated in air , i.e. , in oxidative conditions favoring the formation of diamagnetic v@xmath60 .
unfortunately , annealing in vacuum did not result in single - phase samples .
( color online ) magnetic susceptibility of ag@xmath0vop@xmath0o@xmath1 measured at fields @xmath22 of 0.5 t ( empty circles ) and 5 t ( filled circles ) .
solid and dashed lines show the fits with the model of @xmath13 alternating chain according to the first and second lines of table [ table - fit ] , respectively .
the dash - dotted line is the fit with the htse ( ref . ) above 10 k. ] now , we fit the experimental data with model expressions .
basically , one should use the model of the alternating zigzag chain , as suggested by structural considerations in sec .
[ structure ] .
however , only high - temperature series expansion ( htse ) results are available for the susceptibility of this model.@xcite we find a perfect fit of the experimental data above 10 k ( see fig . [ fig - suscept ] ) with @xmath61 , @xmath62 k , @xmath63 k , @xmath64 k , and @xmath65 emu / mol ( temperature - independent contribution ) .
note that we have assumed @xmath66 : at the stage of fitting , this choice is , of course , arbitrary , and one may invert the assumption without any change in the results .
one should not be surprised by the use of the htse for temperatures well below @xmath46 , as the expansion under consideration is valid down to @xmath67 ( see ref . ) .
moreover , the data below the susceptibility maximum ( i.e. , below 20 k ) are crucial for the stability of the fitting and the obtaining of reasonable exchange couplings .
unfortunately , the htse is inapplicable in the low - temperature region . to fit the decrease of the susceptibility between 20 and 5 k
, we have to turn to a simplified model .
if one neglects @xmath17 , ag@xmath0vop@xmath0o@xmath1 becomes an alternating chain system .
the experimental data above 5 k are fitted with the expression : @xmath68 where @xmath58 is the temperature - independent contribution , @xmath69 accounts for the curie - like paramagnetism of impurities and defects , while @xmath70 is the susceptibility of an @xmath13 alternating chain as given in ref . .
the term @xmath70 includes three variable parameters : @xmath71 , and the alternation coefficient @xmath72 .
again , we assume @xmath66 , then @xmath73 and @xmath74 . we find a perfect fit of the experimental data above 5 k ( fit 1 in fig .
[ fig - suscept ] ) with the parameters listed in the first row of table [ table - fit ] . below 5 k ,
the experimental data deviate from the model due to the deviation of the paramagnetic impurity contribution from the curie law .
note that one may skip the @xmath69 term and find a reasonable fit above 8 k ( fit 2 at fig .
[ fig - suscept ] ; the parameters are listed in the second row of table [ table - fit ] ) , but the data between 5 and 8 k are poorly described .
however , the second fit presents a somewhat better @xmath59 value that is close to the expected range of 1.931.96 ( see refs . and ) . on the other hand
, @xmath59 may be decreased due to the presence of nonmagnetic impurities in the sample under investigation ( as shown by the curie - weiss fit above ) or due to the neglect of @xmath17 .
.[table - fit]the results of the fitting of magnetic susceptibility data with an alternating chain model [ eq .
( [ eq - fit ] ) ] .
@xmath58 is a temperature - independent contribution , @xmath75 denotes the curie - like @xmath69 contribution , @xmath59 is the @xmath59-factor , @xmath76 labels the leading in - chain interaction , and @xmath72 is the alternation parameter .
fit 1 and fit 2 are marked according to fig .
[ fig - suscept ] . [ cols="^,^,^,^,^,^ " , ] the hopping parameters ( indicating the width of the half - filled bands ) are two orders of magnitude smaller than the coulomb repulsion ; therefore , we may use the lsda+@xmath29 technique to account for correlation effects within band structure calculations in a mean - field approximation .
four patterns of spin ordering ( see fig . [ fig - patterns ] ) were used in order to estimate @xmath16 , and @xmath17 .
the resulting exchange integrals are listed in table [ table - integrals ] .
the absolute values of @xmath40 s depend on @xmath41 considerably , but qualitatively the results are similar . we find that @xmath46 is indeed the strongest coupling , it exceeds @xmath77 at least by a factor of 6 .
@xmath46 is the interaction corresponding to longer v v separation ( 5.29 a _ vs. _ 4.92 a for @xmath47 ) as shown in fig .
[ fig - structure ] .
the small value of @xmath17 is consistent with the tight - binding results and provides a further justification for the fitting of experimental data with the alternating chain model ( see sec .
[ experimental ] ) .
all the interactions in ag@xmath0vop@xmath0o@xmath1 are antiferromagnetic . as we have mentioned in sec .
[ method ] , the choice of @xmath41 is a problematic point of the lsda+@xmath29 approach .
the @xmath41 parameter is usually fitted to experimentally observed properties ( e.g. , energy gaps , magnetic moments ) or estimated by means of constrained lda calculation ( see , for example , ref . ) .
basically , the choice of @xmath41 remains somewhat ambiguous and depends on the particular experimentally observed quantities used for the fitting as well as structural features and the computational method .
the results for @xmath78 ev show the best agreement with the experimental and tight - binding estimates of @xmath40 s .
note also that the largest energy gap of 2.3 ev found for @xmath78 ev seems to be consistent with the green color of ag@xmath0vop@xmath0o@xmath1 .
these findings are somewhat unexpected since we used @xmath79 ev , and even smaller @xmath80 ev had been applied in lsda+@xmath29 calculations for vanadium oxides before.@xcite however , one should always keep in mind that @xmath42 and @xmath41 are different quantities having different physical meaning .
@xmath42 is an effective repulsion in a mixed vanadium - oxygen band , while the repulsive potential @xmath41 is applied to v @xmath81 orbitals only .
thus , @xmath41 should exceed @xmath42 , since @xmath42 corresponds to `` molecular '' v o orbitals having larger spatial extension as compared to that of atomic v @xmath81 orbitals . as for
the previous lsda+@xmath29 calculations for vanadium oxides,@xcite the difference in computational methods and crystal structures may be relevant .
the linearized muffin - tin orbital ( lmto ) basis is used in ref . , while we apply a different basis of local orbitals . moreover , korotin _ _ et al.__@xcite studied layered compounds containing v@xmath14 in square - pyramidal coordination , while vanadium is surrounded by six oxygen atoms in ag@xmath0vop@xmath0o@xmath1 .
thus , the difference in the optimal @xmath41 value is plausible , although the origin of this difference is not completely clear .
experimental data and band structure calculations provide a consistent microscopic scenario for the magnetic interactions in ag@xmath0vop@xmath0o@xmath1 .
one of the nearest - neighbor couplings ( @xmath46 ) is about 30 k and exceeds the other couplings at least by a factor of 6 .
thus , ag@xmath0vop@xmath0o@xmath1 may be considered as a system of weakly coupled dimers . indeed , a simple model of isolated dimers [ with additional terms @xmath82 similar to eq .
( [ eq - fit ] ) ] with @xmath83 and @xmath84 k perfectly fits the experimental data above 5 k. this fit completely matches fit 1 in fig .
[ fig - suscept ] .
interdimer interactions @xmath47 and @xmath17 may be responsible for the slight variation of the @xmath59 value , and they certainly influence the possible onset of the long - range ordering . the htse fit and band structure calculations indicate that both @xmath47 and @xmath17 are antiferromagnetic , therefore frustration should be expected . however , it will not change the main feature of ag@xmath0vop@xmath0o@xmath1 a spin - gap - type magnetic behavior due to strong intradimer coupling .
( color online ) four patterns of spin ordering employed for the calculation of @xmath40 s with lsda+@xmath29 .
dash - dotted rectangles show the repetition unit ( i.e. the supercell ) .
circles denote vanadium atoms ; full and empty circles mark different spin directions . ]
the identification of weakly coupled spin dimers in ag@xmath0vop@xmath0o@xmath1 is surprising , since two nearest - neighbor interactions @xmath46 and @xmath47 have very similar superexchange pathways .
the differences in interatomic distances and angles for these v o p o v connections do not exceed 0.05 a and @xmath85 , respectively .
nevertheless , the tiny structural changes strongly influence the values of the exchange integrals .
as we have mentioned in the introduction , superexchange pathways in vanadium phosphates are sometimes non - straightforward .
however , some clarity may be achieved by considering two assumptions .
first , v o p o v superexchange pathways are often much more efficient than v o v ones .
second , the magnitude of interactions involving po@xmath3 tetrahedra can be estimated using structural parameters [ via the so - called magnetostructural correlations , see refs . and ] .
these correlations suggest that double phosphate bridges ( i.e. , two po@xmath3 tetrahedra linking two vo@xmath11 octahedra ) are usually more efficient compared to single bridges , while the interactions via double bridges depend on the relative orientation of vanadium and phosphorous polyhedra quantified by several geometrical parameters . the present study may provide a good test for such structural considerations concerning magnetic interactions in systems with complex superexchange pathways . in case of ag@xmath0vop@xmath0o@xmath1 ,
double phosphate bridges are relevant for both @xmath46 and @xmath47 .
most geometrical parameters are similar for these interactions , and the only difference deals with the in - plane displacement of the vo@xmath11 octahedra ( plane implies the equatorial plane of octahedra , i.e. , the plane of the half - filled @xmath15 orbital ) .
such displacement favors @xmath46 rather than @xmath47 , although one would not expect a considerable difference between the two integrals .
however , our experimental and computational results suggest the scenario of @xmath86 resulting in a system of weakly coupled dimers rather than in an alternating chain .
basically , the structural considerations suggest correct trends , but quantitative estimates require more sophisticated techniques . in this paper
, we tried to use an adequate computational approach to the study of vanadium phosphates .
all the computational results available are based on very simplified techniques such as extended hckel @xcite or dft gga calculations @xcite for small clusters ( typically , spin dimers two vo@xmath11 octahedra and bridging po@xmath3 tetrahedra ) .
now , we show that vanadium phosphates are readily accessible to full - potential band structure calculations that are known as a very efficient tool for the study of magnetic properties of transition metal oxides.@xcite full - potential calculations provide reasonable accuracy and set the computational results on much safer grounds compared to cluster calculations , since the crude cluster approximation is avoided and correlation effects ( which are very important for the properties of transition metal compounds ) can be included .
we are looking forward to the further application of full - potential calculations to vanadium phosphates and to the discovery of novel unusual spin systems in these compounds .
it is worth noting that ag@xmath0vop@xmath0o@xmath1 may be an interesting realization of the alternating zigzag chain model .
this model , also known as dimerized zigzag chain or dimerized frustrated chain , has been extensively studied theoretically with a particular interest to its excitation spectrum.@xcite however , the experimental access to the model is still scarce and basically limited by the spin peierls phase of cugeo@xmath18 that corresponds to the weakly dimerized case . as we have stated above ,
ag@xmath0vop@xmath0o@xmath1 is strongly dimerized , and the dominating spin - gap properties are observed .
nevertheless , the frustration of weak interdimer couplings may be manifested in high - field properties of the material or its magnetic excitations . using the representative values of @xmath87
k , @xmath88 k , we find @xmath89 , @xmath90 and estimate the position of ag@xmath0vop@xmath0o@xmath1 on the phase diagram of the model [ @xmath91 is the frustration ratio , while @xmath92 is the dimerization ratio ] . according to refs . and , this position implies the presence of magnetization plateau , while ref . suggests the formation of bound singlet and triplet states above the elementary triplet excitation that corresponds to the spin gap .
thus , a detailed study of the high - field properties of ag@xmath0vop@xmath0o@xmath1 and neutron scattering experiments probing the magnetic excitations of this compound could be interesting . in conclusion , we studied magnetic properties of ag@xmath0vop@xmath0o@xmath1 using magnetic susceptibility measurements and full - potential band structure calculations .
we reach a perfect consistency between experimental and computational results and describe ag@xmath0vop@xmath0o@xmath1 as a system of weakly coupled dimers on the alternating zigzag chain , although the strong dimerization is rather surprising from the structural point of view .
the frustration of interdimer interactions may lead to a number of unusual properties that deserve a further experimental study .
we present the first application of full - potential band structure calculations to vanadium phosphates .
the calculations show reasonably accurate and highly reliable results , hence they should be an efficient tool in the study of related compounds .
the authors acknowledge financial support of rfbr ( project no .
07 - 03 - 00890 ) , gif ( grant no .
i-811 - 257.14/03 ) , and the emmy - noether - program of the dfg .
zih dresden is acknowledged for computational facilities .
we are grateful to evgeny antipov for careful reading of the manuscript and valuable comments .
is also grateful to mpi cpfs for hospitality and financial support during the stay . | magnetic properties of the silver vanadium phosphate ag@xmath0vop@xmath0o@xmath1 are studied by means of magnetic susceptibility measurements and electronic structure calculations . in spite of the layered crystal structure suggesting 1d or 2d magnetic behavior ,
this compound can be understood as a spin dimer system .
the fit of the magnetic susceptibility indicates an intradimer interaction of about 30 k in perfect agreement with the computational results .
our study emphasizes the possible pitfalls in interpreting experimental data on structural basis only and points out the importance of microscopic models for the understanding of the magnetic properties of vanadium phosphates . |
the hot x - ray - emitting coronae around disk galaxies serve as a reservoir where galaxies acquire baryons to build up galactic disks and deposit kinetic energy and chemically - enriched materials .
this x - ray emitting gas has been detected in various types of disk galaxies ( e.g. , @xcite ) , and can be globally reproduced by recent cosmological hydrodynamical simulations ( @xcite ) , supporting the expected origin of galactic corona from feedback close to the galaxy ( @xcite ) , or maybe related to accretion of external gas at large galactocentric radii ( @xmath2 ; @xcite ) . the coronal properties are observed to correlate with various galaxy properties ( @xcite ) . in particular , there is a tight correlation between the coronal soft x - ray luminosity and the total [ core collapsed ( cc)+type ia ] supernova ( sn ) energy injection rate .
the x - ray radiation efficiency ( the fraction of sn energy observed in the coronal x - ray emission ) is quite low ( @xmath3 ) , indicating that most of the feedback material is missing from a clear detection in soft x - ray ( @xcite ) . in galaxies rich in cool gas , a large fraction of the feedback energy
may be radiated through the recombination line of warm ionized gas at a lower temperature ( @xmath4 ; e.g. , @xcite ) .
however , in early - type galaxy bulges which are typically poor in cool gas , the missing feedback material is most likely carried out to the intergalactic space by a galactic bulge wind mainly produced by type ia sne ( e.g. , @xcite ) .
the metal - enriched galactic feedback material is typically hot ( @xmath5 ) and has too low density ( @xmath6 ) to be directly detected in current x - ray observations ( @xcite ) .
it is thus controversial what is the nature of the hot gas detected in soft x - ray around nearby galaxies : the real diffuse hot tenuous corona or just the mixed material at the interface of cool and hot gases ( e.g. , @xcite ) ? in order to understand the nature of this soft x - ray emission and to search for the missing galactic feedback material , we need to measure the metallicity of hot gas around isolated , moderate- or low - mass , early - type galaxies , which are weakly affected by the intra - cluster medium ( icm ) , gravitational heating , or current star formation . however , existing x - ray observations of such galaxies often reveal a low fe ( a primary product of type ia sn ) abundance , typically subsolar or slightly supersolar ( e.g. , @xcite ) , indicating that most of the sn ejecta is not detected .
we herein present a deep _ suzaku _ observation of ngc 5866 .
this galaxy locates in a low density environment , with two companions , ngc 5907 and ngc 5879 , located several hundred kpc away in projection ; their interactions with ngc 5866 should have negligible effects on the coronal properties .
the agn and star formation activities are both weak , unlikely contributing significantly to the x - ray emission .
the edge - on orientation helps the gas - rich galactic disk to block most of the on - disk x - ray - bright point - like sources , reducing the complexity of analyzing the low - resolution _ suzaku _ data . based on our preliminary study of ngc 5866 using _
chandra _ and multi - wavelength data @xcite , we found it has high fe / o abundance ratio as a disk galaxy @xcite , thus well - suited for studying type ia sn metal enrichment .
however , the metal abundances are poorly constrained with _
data only , so we proposed deep _
observation of this galaxy to collect more photons ( @xmath7 ; cycle 6 ; pi : li j. t. ) . in this paper
, we jointly analyze the _ suzaku _ and _ chandra _ data , and further compare this galaxy to other galaxies and discuss how the new observation could help us to understand the type ia sn feedback processes .
the paper is organized as follows : in [ sec : data ] , we describe the calibration of the _ suzaku _ data and the joint analysis of the _ suzaku _ and _ chandra _ spectra . in [ sec : discussion ] , we compare our x - ray measurements of ngc 5866 to similar measurements of other galaxies and further discuss the results and their implications .
out major results and conclusions are summarized in
[ sec : summary ] .
errors are quoted at 90% confidence interval throughout the paper , unless specifically noted .
_ suzaku _ observation of ngc 5866 was taken on 2011 - 05 - 20 with a total exposure time of 101.8 ks .
we reprocess the xis data using the ftool _ aepipeline _ and caldb files of version 2014 october 1 .
we identify the high background flare time intervals using a @xmath8 clip with a lightcurve binned at 200 s. we then remove these background flares and obtain an effective exposure time of 100.8 , 80.9 , 100.2 ks for xis0 , xis1 , xis3 , respectively .
we generate the xis non - x - ray background spectra and images using the ftool _
xisnxbgen_. in imaging analysis , the vignetting is corrected by simulating an event file using _
xissim_. the images are rescaled with bin=8 ( pixel size=@xmath9 ) , trimmed to remove low exposure regions , flat fielded , and exposure corrected .
the resultant tri - color images in 0.3 - 1 kev , 1 - 2 kev , and 2 - 7 kev are shown in fig .
[ fig : image]a , while the 0.3 - 1 kev contours in the central region are presented in fig .
[ fig : image]b .
_ suzaku _ xrt has a typical size of the point spread function ( psf ) of @xmath10 . as shown in fig .
[ fig : image]a , except for the emission from ngc 5866 , the other bright x - ray sources ( the two sources at the edge of the ccd and the two fainter sources marked with the circles ) are all background sources also identified by the _ chandra _ ( [ subsec : datachandra ] ; @xcite ) .
we define the regions used for extracting sky background and source spectra as shown in fig .
[ fig : image]a , b . the smaller circle in fig . [ fig : image]b has a diameter of @xmath11 , matching the xrt psf size ; we herein call it the `` inner halo '' region
. the larger circle in fig .
[ fig : image]b has a diameter of @xmath12 ; we define the annulus between these two circles as the `` outer halo '' region , outside which there is no significant diffuse x - ray emission .
the large annulus in fig .
[ fig : image]a ( with diameters of @xmath13 ) , after removing the two bright background sources , are used to extract the sky background spectrum .
we extract the xis spectra and the corresponding response files of the source and sky background regions using the ftool _
xselect_. the non - x - ray instrumental background spectra are also extracted from the same regions .
there is a weak line feature at @xmath14 of the xis1 spectra , which is not detected in xis0/xis3 or the _ chandra _ spectra .
this feature is not common in thermal plasma spectrum .
we do not find any significant time variation of it .
the feature also presents over the entire xis1 field of view .
it is thus a background feature most likely caused by _
quiescent _ solar wind charge exchange ( e.g. , @xcite ) .
we will use an additional _ gauss _ model to account for it in the following spectral analysis .
_ suzaku _ has too low angular resolution to individually detect x - ray bright point - like sources , such as agns or x - ray binaries , within the optical extent of ngc 5866 .
unlike x - ray faint stellar sources , the contribution from these x - ray bright sources is quite uncertain , and could be the major uncertainty in characterizing the truly hot gas emission .
we therefore need a joint analysis of the _ chandra _ and _ suzaku _ data to better decompose the x - ray emission from hot gas and stellar sources and/or agn .
the archival _ chandra _ acis - s observation of ngc 5866 is presented in @xcite .
it has an effective exposure time of @xmath15 .
we herein directly use the calibrated data , the diffuse x - ray images , and the detected point - like sources ( fig .
[ fig : image]b ) from @xcite .
spectra are extracted from the same regions as the _ suzaku _ spectra .
we do not remove the point sources from the diffuse x - ray emission , in order to be consistent with the _ suzaku _ spectra .
instead , we extract an accumulated _ chandra _ spectrum of all the point sources detected within the inner halo region ( point sources in the outer halo have too few counts to construct a spectrum ) . this spectrum will be jointly analyzed with the _ chandra_/_suzaku _ diffuse x - ray spectra extracted from the inner halo . in general , the x - ray spectra of the inner and outer halos
are comprised of several different components .
we characterize the hot gas emission with an optically thin ionization equilibrium thermal plasma model vapec .
the oxygen and iron abundances are set free , while the abundances of other elements are set to the solar values . in order to better characterize the hot gas emission
, we also need to quantify the various stellar contributions . with the archival _ chandra _ data
, we reach a point source detection limit of @xmath16 in 0.5 - 8 kev .
the accumulated spectrum of all the detected point sources within the inner halo region can be well modeled with a power law , which has a photon index of @xmath17 and an absorbing column density slightly larger than the galactic foreground value of @xmath18 . below the detection limit
, the unresolved stellar source emission is dominated by the emission from cataclysmic variables ( cvs ) and coronal active binaries ( abs ) , as well as the emission from unresolved low mass x - ray binaries ( lmxbs ) and the residual photons spilling out of the discrete source removal circles .
similar as @xcite , we model this component with a 0.5 kev mekal model ( optically thin thermal plasma ) and a @xmath19 power law model .
both components are well calibrated and their normalizations are scaled with the stellar mass ( @xcite ) .
there is still a small residual at @xmath14 after subtracting the sky background .
this feature is only presented in the xis1 spectrum and is most likely a residual feature produced by the _
quiescent _ solar wind charge exchange .
we therefore add into the xis1 spectrum a narrow gauss model with fixed line width of @xmath20 to account for it . .hot gas properties from spectral analysis [ cols="<,^,^",options="header " , ] [ table : earlytypecoabun ] fe / o abundance ratio of hot gas and molecular gas mass ( @xmath21 ) of ngc 5866 and the elliptical galaxies from @xcite and @xcite . for galaxies included in both @xcite and @xcite
, we adopt @xcite s measurements as they use better x - ray data .
@xcite perform x - ray measurements based on different instruments ( _ chandra _ acis , _ xmm - newton _ epic and rgs ) and with different aperture sizes ( diameters of @xmath22 and @xmath23 ) . we herein adopt the _ xmm - newton _ epic @xmath22-diameter measurements , which should be the most similar as our measurement of ngc 5866 , and is less affected by the clustered environment of some galaxies . the fe / o ratio used in the plots
are summarized in the `` fe / o '' column .
we also list @xcite s _ xmm - newton _
epic @xmath23-diameter measurements in the `` fe / o ( @xmath23 ) '' column for comparison .
the differences in these two columns do not affect our results discussed in this paper .
errors of the fe / o ratio are quoted at 90% confidence level in this table , but are plotted at @xmath24 confidence level in figs .
[ fig : feoratiotc ] and [ fig : feoratiomh2 ] to clarify .
the molecular gas mass @xmath21 is based on co observations , with references listed in the last column .
evans05 : @xcite ; sage07 : @xcite ; prandoni10 : @xcite ; young11 : @xcite .
the temperature of the hot gas in the inner halo seems too low to be primarily heated by sne . in gas - poor galactic bulges like ngc 5866
, hot gas is the dominant phase and the radiative cooling of the shocked ism is overall not important in the sn energy budget ( e.g. , @xcite ) .
the specific enthalpy of old stellar feedback ( mass loss of evolved low - mass stars and type ia sne heating ) is @xmath25 per particle .
such a specific enthalpy gives an average gas temperature of @xmath26 , considering the energy equipartition between the hot gas thermal energy and the energy stored in magnetic field and cosmic rays as well as turbulent motion @xcite . on the other hand ,
adopting a central velocity dispersion of @xmath27 obtained from the hyperleda database , the expected temperature of gas thermalized in the gravitational potential of ngc 5866 would be @xmath28 , quite close to the measured hot gas temperature of @xmath1 in the inner halo ( table [ table : specfitresult ] ) .
therefore , the low measured hot gas temperature _ apparently _ indicates that any additional sources of non - gravitational heating , such as type ia sne , play a relatively minor role ( contributes only @xmath29 of the total thermal energy ) .
although located in a very low density environment ( the environmental galaxy number density of ngc 5866 is @xmath30 ; in comparison , @xmath31 is classified as clustered in @xcite ) , it is still possible that there exists a low - density intra - group medium around ngc 5866 .
however , neither of the existing _ suzaku _ or _ chandra _ data as presented in this paper is deep enough to probe the hot gas distribution at large radii .
chandra _ x - ray intensity profiles at small radii , approximately within the spectral analysis regions of the present paper , are presented in @xcite .
these profiles do not reveal any diffuse soft x - ray emission above the stellar contributions at @xmath32 ( or roughly the inner halo region in this paper ; fig .
[ fig : image]b ) . therefore , based on the existing x - ray data , we do not find significant evidence for the existence of an intra - group medium , which , if presents , could help to raise the hot gas temperature in the outer halo ( e.g. , @xcite ) .
it is suggested that the x - ray emissivity of early - type galactic coronae , and also the fraction of sn ejecta detected in soft x - ray , are highly affected by the specific energy of the hot gas ( energy per unit mass , or temperature if the gas is thermalized ; @xcite ) .
if this specific energy is reduced by some processes , there will be a lower temperature of the hot gas and slower galactic outflow . as a result
, the hot gas density and x - ray emissivity will be raised and there will be a larger amount of heavy elements produced by type ia sne ( such as fe ) detected in soft x - ray .
cool - hot gas interaction , as indicated by the relatively large amount of cool gas ( table [ table : earlytypecoabun ] ; fig .
[ fig : feoratiomh2 ] ) and the presence of extraplanar dusty filaments in ngc 5866 ( fig . [
fig : hstfilaments ] ) , could provide a natural explanation of both the low temperature and high fe abundance of the hot gas .
the apparently low type ia sne contribution to the heating of the hot gas is very unlikely , because ngc 5866 shows so many features related to type ia sne feedback , such as the fe enrichment detected in hot gas and the consistency of the gravitational energy of many extraplanar dusty filaments with a single sn energy @xcite .
furthermore , the significant raising of the hot gas temperature in the outer halo ( fig .
[ fig : ktcompare ] ; [ subsec : specanalysis ] ) in such a low - mass isolated galaxy is quite uncommon ( e.g. , @xcite ) .
although we can not absolutely exclude the contribution from a low - density intra - group medium based on the existing _ suzaku _ or _ chandra _ data , it is natural that there exists some additional cooling processes , such as the cool - hot gas interaction .
the presence of cool gas could affect the surrounding hot gas properties via mass loading and enhanced radiative cooling , both processes reduce the specific energy and so the temperature of the hot gas . the unusually strong cool - hot gas interaction in ngc 5866 as an s0 galaxy could also help to explain the high fe abundance detected in soft x - ray .
the cool gas is often less metal - enriched , and the mass loading of cool gas into hot phase will result in a lower average metallicity of the hot gas .
however , most of the sn ejecta contained in the high - temperature ( typically @xmath33 ) hot gas has too low emissivity to be detected in soft x - ray , because the hot gas density of gas - poor galactic bulge is too low ( typically @xmath34 , e.g. , @xcite ) .
therefore , the cool - hot gas interaction will have an opposite effect on the observed gas metallicity , by significantly increasing the soft x - ray emissivity and so the _ emission - measure - weighted _ metallcity as obtained from x - ray observations .
such a difference between the emission - measure- and mass - weighted metallicity is also evidenced by numerical simulations ( e.g. , @xcite ) , indicating that we are missing from the detection in x - ray observations of a large fraction of gas - phase metals , as also being estimated in
[ subsec : hotgasmetallicity ] .
we present detail calibration of the _ suzaku _ observation of an isolated s0 galaxy ngc 5866 . by jointly analyzing this _ suzaku _ data together with an archival _ chandra _ observation
, we obtain the hot gas temperature in the inner ( @xmath35 ) and outer ( @xmath36 ) halos , as well as the o and fe abundances in the inner halo .
these measurements of the hot gas properties are much more accurate than previous measurements with the _ chandra _ data only , allowing for a comparison with other galaxies .
we find a significant supersolar fe / o abundance ratio of @xmath37 at @xmath38 statistical confidence level in the inner halo of ngc 5866 , indicating a clear metal enrichment from type ia sne .
the fe / o abundance ratio of ngc 5866 is also significantly higher than other disk ( spiral+s0 ) galaxies , and even many of the elliptical galaxies .
we also find that ngc 5866 simultaneously has the highest fe / o abundance ratio and molecular gas mass ( @xmath21 ) among a small sample of gas - poor early - type galaxies , although the poor and mostly upper limit measurement of @xmath21 prevents us from claiming a clear positive correlation between fe / o and @xmath21 .
the total amount of fe detected in the inner halo of ngc 5866 is @xmath39 .
in contrast , the total amount of fe injected by type ia sne within the dynamical cross timescale of the galactic outflow is @xmath40 , while that within the age of the galaxy since the last starburst is @xmath41 . considering the uncertainty of the volume filling factor @xmath42 which can not be too low in this non - starburst s0 galaxy , we could conclude that ngc 5866 could preserve a larger than usual fraction , but far from the total amount of fe injected by type ia sne .
ngc 5866 exhibits a lot of extraplanar dusty filaments extending perpendicular to the galactic disk within the inner halo .
these filaments indicate the existence of interaction between sne and cool gas , which probably triggers cool gas mass loading into the hot phase and enhances the radiative cooling .
we find clearly lower hot gas temperature in the inner halo than in the outer halo , which is apparently a result of this cool - hot gas interaction , if this well isolated galaxy is not highly affected by a high - temperature intra - group medium .
this low temperature in the inner halo is also roughly consistent with gravitational heating of the material lost by evolved stars , _ apparently _ leaving little room for additional non - gravitational heating such as by type ia sne .
however , heating by type ia sne is very likely to be important in this galaxy showing clear fe enrichment and cool - hot gas interaction
. therefore , additional cooling processes must exist .
the low temperature and high fe abundance of the hot gas in the inner halo of ngc 5866 strongly suggest that the presence of cool gas in gas - poor galaxy bulges could trigger cool - hot gas interaction , resulting in a lower hot gas temperature and a larger fraction of sn ejecta detected in soft x - ray around the galaxy .
however , most of the metal - enriched sne ejecta is still missing from a clear detection in x - ray .
li jiang - tao would like to acknowledge roberts shawn r. , wang q. daniel , and the _ suzaku _ helpdesk for discussions on the _ suzaku _ background and also the anonymous referee for very helpful comments .
this work is supported by nasa through the grant nnx13ae87 g and nnx15am93 g . | one of the major sources of the x - ray emitting hot gas around galaxies is the feedback from supernovae ( sne ) , but most of this metal - enriched feedback material is often not directly detected in x - ray observations .
this missing galactic feedback problem is extremely prominent in early - type galaxy bulges where there is little cool gas to make the type ia sne ejecta radiate at lower temperature beyond the x - ray domain .
we herein present a deep _ suzaku _ observation of an s0 galaxy ngc 5866 , which is relatively rich in molecular gas as an s0 galaxy and shows significant evidence of cool - hot gas interaction . by jointly analyzing the _ suzaku _ and an archival _ chandra _ data , we measure the fe / o abundance ratio to be @xmath0 relative to solar values .
this abundance ratio is much higher than those of spiral galaxies , and even among the highest ones of s0 and elliptical galaxies .
ngc 5866 also simultaneously has the highest fe / o abundance ratio and molecular gas mass among a small sample of gas - poor early - type galaxies .
an estimation of the fe budget indicates that ngc 5866 could preserve a larger than usual fraction , but far from the total amount of fe injected by type ia sne .
we also find that the hot gas temperature increases from inner to outer halos , with the inner region has a temperature of @xmath1 , clearly lower than that expected from type ia sne heating .
this low temperature could be most naturally explained by additional cooling processes related to the cool - hot gas interaction as being indicated by the existence of many extraplanar dusty filaments .
our results indicate that the large cool gas content and the presence of cool - hot gas interaction in the inner region of ngc 5866 have significantly reduced the specific energy of the sn ejecta and so the velocity of the galactic outflow .
the galaxy could thus preserve a considerable fraction of the metal - enriched feedback material from being blown out . |
recently the studies of @xmath4 collisions with high @xmath5 jets at the tevatron @xcite and of dijet events in @xmath6 collsions at hera @xcite showed that a considerable fraction of these hard scattering events ( of order 1 - 10% ) contain large gaps in the rapidity distribution of particles .
such phenomena had been expected from the exchange of colour neutral objects @xcite . in @xmath3 annihilation
the primary parton process is the production of @xmath7 pairs followed by gluon bremsstrahlung and quark pair production . in this case
the suppression of central gluon radiation would produce a rapidity gap for the partons but the recoiling @xmath7 like systems separated by the gap may become the source of nonperturbative hadron production to neutralize the color field ; then no gap would remain between the observed hadrons .
one way to generate large rapidity gaps between the final hadrons in @xmath3 annihilation is through the production of colour singlet clusters separated by large rapidity gaps .
such configurations can be obtained perturbatively in hard processes leading to at least four partons in the final state , like @xmath8 or @xmath9 @xcite . in this case
the hadrons are produced within each colour neutral @xmath10 or @xmath11 cluster and the large gap persists . as these processes involve a highly virtual intermediate quark or gluon the rate for these large rapidity gap events is rather small though .
one finds that the rate keeps decreasing with increasing gap size , contrary to the case of @xmath4 or @xmath6 collision .
such an unlimited decrease of the gap probability has been seen already some time ago @xcite .
the recent analysis at the @xmath12 resonance by the sld collaboration @xcite has shown the fall of the gap probability over five orders of magnitude ( see also the first results from aleph @xcite ) . at the same time
the observed absolute size of the gap probability exceeds the expectations from the above calculations @xcite by about two orders of magnitude .
this observation excludes this type of hard colour neutralization to be the dominant mechanism for the production of large rapidity gaps . in this paper
we investigate how the fraction of rapidity gap events @xmath13 for the parton cascade ( without hadronization ) compares with the experimental results on hadrons . at first sight
one might expect that the decrease of the gap fraction for hadrons and therefore in the experimental data is steeper than the one of the parton cascade because of the additional suppression from removing the hadrons emitted into the gap during the hadronization phase .
we will see however that this effect strongly depends on the scale @xmath14 for which the parton cascade is terminated .
there are many phenomena in multiparticle production which are in favour of a very simple hadronization picture , called local parton hadron duality ( lphd ) @xcite , which compares the observed sufficiently inclusive particle densities and correlations directly with the calculations at the parton level , assuming in effect the production of a hadron close to the parton in phase space ( for reviews , see @xcite ) . in this approach the parton cascade is evolved down to the low scales @xmath15 of a few hundred mev ; @xmath14 denotes here the cutoff in the transverse momentum of the emitted partons and @xmath16 the qcd scale for the running coupling .
this calculation of the gap rate is similar in spirit to the `` clan '' model analysis within a simplified parton shower interpretation:@xcite the primary independent emission of hadronic objects leads to a poisson distribution with the gap rate related to the probability for no emission by an exponential . in the following
we calculate the gap probability analytically in different approximation schemes of perturbative qcd and compare to the data for the previously determined values of @xmath14 and @xmath16 .
this will teach us whether the parton hadron duality works in the more extreme , quasi exclusive configurations with large rapidity gaps as well ; in this case a confirmation of the duality approach is even less expected than in case of inclusive observables .
a gap in the rapidity distribution occurs if the initial @xmath7 pair does not radiate into the respective rapidity interval .
the probability for no radiation into a certain angular interval is given in field theory by the exponential sudakov form - factor @xcite , originally derived in qed .
a problem very similar to the rapidity gaps considered here occurs in the calculation of multijet rates in @xmath3 annihilation applying the durham/@xmath0 algorithm:@xcite the 2-jet rate is again given by the sudakov form factor which represents the probability for no - parton - production above a certain @xmath0 cutoff ( resolution parameter ) .
in our application we consider the rapidity gap having no gluons inside the angular or rapidity interval above the transverse momentum cutoff @xmath14 which corresponds to a hadronic scale of a few hundred mev as outlined in the introduction .
let us consider specifically the angular interval between @xmath17 and @xmath18 ( @xmath19 ) where @xmath20 is measured with respect to the quark direction .
the rapidity is then obtained from @xmath21 .
let us further denote the probability for emission of a gluon at an angle @xmath22 with the energy @xmath23 off a parent parton @xmath24 ( either a gluon(g ) or a quark(q ) ) as @xmath25 .
then , the sudakov form factor for the angular ordered cascade is given by @xmath26 and this represents the probability that no gluon is emitted within the cone of half angle @xmath20 from the parent parton with the energy @xmath27 with transverse momentum above @xmath14 . in particular
, @xmath28 represents the probabilty that there is no emission of a gluon with emission angle between @xmath17 and @xmath29 .
these rates have been calculated in different approximations which we discuss in the following .
+ _ the double logarithmic approximation ( dla ) _ + the simplest approximation takes into account only the leading contributions from the angle and energy singularities of the gluon emission . in this case the total number of gluons radiated from a primary parton into a forward cone of half angle @xmath30 as in ( [ exactdoubleintegration ] ) is given in a small angle approximation by @xmath31 or , using logarithmic variables @xmath32 with @xmath33 and the dla anomalous dimension @xmath34 = @xmath35 for running coupling @xmath36 @xmath37 where @xmath38 and @xmath39 denote the numbers of colours and flavours respectively .
furthermore @xmath40 for quarks and @xmath41 for gluons .
one finds @xmath42 the probability for a gap without primary gluons is then obtained as @xmath43 for two arbitrary angles @xmath44 measured with respect to the quark direction . in this approximation
the jets evolve independently in both hemispheres and the respective probabilities factorize . in particular , in the case of a symmetric angular interval @xmath45 )
the gap probability is given by @xmath46 in the present small angle approximation the rapidity is related to the angle by @xmath47 ; in case of the symmetric rapidity interval of full width @xmath48 , the relevant gap angle is @xmath49 which yields @xmath50 for the limiting case @xmath51 .
the use of the small angle approximation for such large angles is formally not allowed , but we will see below that the effect is numerically small ( radiation occurs dominantly at small angles ) .
the gap probability @xmath52 following from ( [ fsym ] ) corresponds to an almost exponential decrease for the relevant parameter range .
it is interesting to note that for fixed coupling there would be a flattening of the gap fraction @xmath53 for large gaps @xmath48 as can be derived from @xmath54 this difference comes from the strong radiation near the kinematical limit @xmath55 in case of running @xmath36 which becomes important for small limiting angles @xmath30 . in other words , for large gaps
the strong radiation near the kinematic limit is kinematically allowed and has to be suppressed which makes the gap probability much smaller than in the case of fixed @xmath56 where this radiation is not equally strong .
+ _ the modified leading logarithmic approximation ( mlla ) _
+ besides the double logarithmic terms one also takes into account in this approximation the leading single logarithmic corrections .
in this subsection we first derive the full o(@xmath36 ) emission probability for @xmath3 annihilation from which the various approximations beyond dla can be derived .
we denote the energy fractions of @xmath57 and @xmath58 as @xmath59 and @xmath60 and take the jet ( thrust ) axis along the ( say , left - moving ) @xmath61 . the energy distribution is then given by @xmath62 ( see , for example , ref .
@xcite where also useful kinematic relations are given ) . in order to evaluate the gluon emission probablity from a parent quark @xmath63 we need to work out the jacobian factor to move from @xmath64 to @xmath65 using the kinematical relations @xmath66 with @xmath67 .
this can be done ecomically by noting @xmath68 and we obtain a simple formula @xmath69 the final form of @xmath70 from parent quark is @xmath71 using the kinematical relation @xmath72 one finds the emission density @xmath73 from the exact result ( [ simplest ] ) we can easily obtain the approximation for small gluon transverse energy fraction @xmath74 , i.e. for @xmath75 , @xmath76 this formula contains the dglap splitting function @xmath77 as factor and can be seen to correspond to the expression @xmath78 for , with @xmath79 , the appropriate jacobian factor is @xmath80 the eq .
( [ glapformula ] ) differs from the dla result by the inclusion of the non - singular dglap term and the correct angular dependence .
we have computed @xmath81 by integrating the density as in ( [ exactdoubleintegration ] ) _ numerically _ with the bounds @xmath82 this yields an improved result for the gap probability calculated again as in ( [ fth ] ) or ( [ fsym ] ) which we refer to as mlla - n " .
an _ analytical _ approximation ( mlla - a " ) can be obtained by simplifying the integral over the nonsingular parts of the splitting function @xcite .
one obtains @xmath83 .
\label{wmlla}\ ] ] where @xmath84 and @xmath85 denotes the expression in eq .
( [ wdla ] ) . the results from these calculations will be discussed in the next subsection .
+ _ comparison of analytical results and a monte carlo calculation _ + the result of our computations in the different approximations are displayed in fig . 1 .
for the parameter @xmath86 = 0.1 at @xmath87 = 0.244 gev we show the dla result for the symmetric gap ( [ fsym ] ) and the mlla results in the analytic approximation ( [ wmlla ] ) and in the numerical evaluation of ( [ exactdoubleintegration ] ) with ( [ glapformula ] ) . also shown as data points are the results from the ariadne monte carlo at the parton level @xcite for the same parameters .
this monte - carlo generates the quark gluon cascade as a sequence of dipole emissions and it is terminated as in the analytical calculations by a transverse momentum cutoff @xcite .
furthermore , there is no limitation to the relative magnitude of the parameters ( besides @xmath88 ) . as can be seen from the figure the mlla corrections are not very large : about a factor or two after a decrease of the gap fraction by 5 orders of magnitude .
of the same order is the deviation from the parton monte carlo which takes into account effects not considered here : the correct angular recoil from the primary gluon emission also for large angles and the spillover of secondary gluons into the gap if the primary gluon is produced ouside the gap .
this effect could be responsible for the difference between the mc and the more accurate mlla calculations .
surprisingly , the simple dla and the mc results are quite close to each other .
the figure also shows the very strong dependence on the cutoff @xmath14 for fixed @xmath87 .
indeed , decreasing the transverse momentum cutoff from @xmath14 = 1 gev down to @xmath14 = 0.27 the gap rate at @xmath48 = 6 decreases by 4 orders of magnitude .
this comes from the singular emission probability @xmath89 enhanced by the running @xmath36 : the closer we come to the pole at @xmath90 the more difficult it is to avoid the gluon bremsstrahlung .
apparently the prediction for the gap probability in the parton cascade depends sensitively on the parameter @xmath14 . a determination of both parameters @xmath87 and
@xmath14 within the mlla - lphd approach has been performed in the study of particle energy spectra @xcite applying the moment analysis @xcite .
this study has shown that the two parameters are very close to each other and there is an upper limit to the @xmath91-parameter : @xmath92 whereas the absolute scale was found to be @xmath93 270 mev .
a lower limit on @xmath86 could in principle be obtained from the study of the mean multiplicity in a calculation of yet higher accuracy .
as central value for our prediction we choose therefore @xmath94= 270 mev and @xmath86= 0.1 ( or @xmath87 = 0.244 gev ) but smaller values of @xmath87 are allowed as well . in fig . 2 we compare such results from the dla ( which agree closely with those from the parton mc ) with the experimental measurement of the fraction of rapidity gap events from the sld collaboration @xcite .
these data also contain a contribution from the @xmath95-lepton events which is important for the large gaps @xmath48 .
the data are well described by the jetset monte carlo @xcite including the @xmath95-decays .
the monte carlo predictions without @xmath95-decays are also shown in the figure and indicate a steeper falloff for large @xmath48 .
furthermore it should be noted that the data refer to rapidity gaps between charged particles . from the monte carlo
one would expect a slightly steeper dependence if the gap is calculated from all final state particles.@xcite one can conclude from the figure that the parton cascade with the low values for the scale parameters @xmath14 and @xmath16 obtained from previous analyses gives already enough suppression for large rapidity gaps without any additional hadronization ; values @xmath91 = 0.05 - 0.1 provide an adequate description of the gap distribution ( our predictions refer to gaps in the final state of all charged and neutral particles ) .
the data are consistent with a description of the hadronic final state in terms of a parton cascade terminated at a rather low cutoff of the order of the final state hadron masses without any additional hadronization phase .
this could be interpreted in favour of a dual correspondence of parton and hadron final states and a soft colour neutralization mechanism .
a good description of the data has been obtained @xcite by the jetset monte carlo @xcite . in this model the parton cascade is terminated at a cutoff @xmath96 o(1 gev ) and followed by a hadronization phase . according to fig .
1 the suppression of large rapidity gaps at a scale @xmath97 gev is not so strong and most of the suppression must come from the resonance and particle production in the hadronization phase . in our parton calculation
the strong suppression of large gaps again comes mainly from the last stage of the cascade evolution where the radiation becomes stronger because of the infrared enhancement and the increasing coupling constant @xmath36 .
it appears that the average properties of the hadronization phase can be well represented by the parton bremsstrahlung cascade with running coupling . in the following we note a few further predictions from the perturbative approach to the gap rate . 1 .
there is a difference in the gap fraction of quark and gluon jets according to the respective colour factors in ( [ wdla ] ) .
this can be tested , for example , by studying gaps in jets at high @xmath5 in @xmath4 collisions between angles of , say , @xmath98 using eq .
( [ fth ] ) .
the slope roughly doubles when going from quark to gluon jets .
this measurement would test the connection of the gap probability to single gluon emission .
2 . in the approximation considered the evolution of the parton cascades in both hemispheres is independent , so the corresponding rates factorize .
for example the slope should double when going from the gap ( @xmath99 ) to ( @xmath100 ) .
3 . the eq .
( [ fth ] ) also predicts the energy dependence of the slope . with rising jet energy
the gap distribution gets slightly steeper : at @xmath101=200 gev the slope is larger by 4 % in comparison to @xmath101=90 gev
. it will be interesting to study the gap events further in order to test the proposed perturbative approach with soft colour neutralization .
we thank sergio lupia for useful discussions and help with various problems of this work .
one of us ( w.o . ) thanks the organizers of the workshop l. cifarelli , a. kaidalov and v. a. khoze for the invitation to this lively workshop with good discussions to remember . | we calculate the probability for rapidity gaps in the parton cascade for different approximations within the perturbative qcd and compare the results with recent measurements .
the aim is to find out whether the dual connection between the parton and hadron final states observed so far in various inclusive measurements holds as well for the extreme kinematic configurations with colour sources separated by large rapidity gaps . a description of the data is possible indeed choosing the parameters of the cascade in the range suggested by recent analyses of the energy spectra ( the @xmath0 cutoff @xmath1 qcd scale @xmath2 mev ) . #
1#2 @=11 tempcntc citex[#1]#2@fileswauxout tempcnta@tempcntb@neciteaciteforciteb:=#2citeo#1 citeotempcnta > tempcntbciteacitea , tempcnta = tempcntbtempcnta
@=12
* rapidity gaps in @xmath3 annihilation + and parton - hadron duality * wolfgang ochs + _ max - planck - institut fr physik + ( werner - heisenberg - institut ) + fhringer ring 6 , d-80805 munich , germany _
+ and + tokuzo shimada + _ meiji university , department of physics + higashi mita 1 - 1 , tama + kawasaki , kanagava 214 , japan _ to appear in the proc . of the 33rd eloisatron
workshop `` universality features in multihadron production and the leading effect '' , ( erice , italy , 19 - 25 october 1996 ) |
superconducting / ferromagnetic ( @xmath0 ) hybrid structures give rise to a fascinating interplay between two antagonist electronic orders .
the ferromagnetic exchange field @xmath1 privileges one spin direction while standard superconductivity favors singlet correlations .
this leads to a rich behavior which has triggered an intense activity in the last years ( see e.g. refs . ) .
in particular , the superconducting proximity effect , i.e. the propagation of the superconducting correlations in ferromagnets , has been widely studied .
this propagation is accompanied by spatial oscillations of the superconducting order parameter , because @xmath1 induces an energy shift between electrons and holes . as a result
, one can build electronic devices with new functionalities , such as josephson junctions with negative critical currents@xcite , which could find applications in the field of superconducting circuits@xcite . from a fundamental point of view , it is very instructive to study the density of states ( dos ) of @xmath0 structures .
so far , this quantity has been less measured@xcite than critical temperatures or supercurrents . however , this way of probing the superconducting proximity effect is very interesting because it provides spectroscopic information .
one striking consequence of the spatial oscillations of the order parameter in the @xmath2 layer is that the zero - energy dos can become larger than in the normal state for certain ferromagnet thicknesses@xcite .
the behavior of @xmath0 hybrid circuits depends crucially on the properties of the interfaces between the @xmath3 and @xmath2 materials . in this paper , we focus on the case of diffusive structures .
diffusive @xmath0 interfaces have been initially described with spin - independent boundary conditions@xcite .
it has been found that the amplitude of the superconducting proximity effect directly depends on the tunnel conductance @xmath4 of an interface ( see e.g. ref . ) .
later , spin - dependent boundary conditions have been introduced , in the limit of a weakly polarized ferromagnet@xcite . due to the spin - dependence of interfacial phase shifts ( sdips)@xcite
, one has to take into account new conductance parameters @xmath5 and @xmath6 at the @xmath2 and @xmath3 sides of the interface , respectively .
it has been shown that @xmath5 and @xmath6 can significantly affect the behavior of @xmath0 hybrid circuits .
for instance , @xmath5 can shift the spatial oscillations of the superconducting order parameter@xcite .
more recently , it has been found that @xmath6 can induce an effective zeeman splitting @xmath7 in a superconducting layer with a thickness @xmath8 smaller than the superconducting coherence lengthscale @xmath9@xcite .
this induces a double gap structure ( dgs ) in the @xmath3 and @xmath2 densities of states .
however , in practice , the regime @xmath10 is frequently reached ( see e.g. refs . ) . remarkably , dgss have been recently observed at the @xmath2 side of bilayers with @xmath8 much larger than @xmath9@xcite , although ref .
has found that @xmath7 scales with @xmath11 in the low @xmath8 regime .
whether a dgs persists in the large @xmath8 regime is therefore an important question , especially in the light of this recent experiment . in this paper
, we study how @xmath5 and @xmath6 modify the dos of a @xmath0 bilayer .
we use a numerical treatment to explore a wider parameter range than in previous works .
in particular , we can reach the limit of thick superconductors and larger values of @xmath6 .
we find that @xmath12 shifts the spatial oscillations of the superconducting order parameter in @xmath2 , like @xmath5 .
it can also significantly affect the amplitude of the superconducting gap .
when @xmath8 increases , the sdips - induced dgs becomes narrower , in agreement with ref . .
nevertheless , it can surprisingly persist in the large @xmath8 limit .
indeed , on a distance of the order of @xmath9 near the @xmath0 interface , the resonance energies of the @xmath3 spectrum remain spin - dependent because quantum interferences make the superconducting correlations sensitive to the sdips .
this behavior is transmitted to the whole @xmath2 layer due to the superconducting proximity effect .
we thus obtain , at the @xmath2 side of @xmath0 bilayers , dos curves which have interesting similarities with those of ref . , although @xmath13 .
more generally , our results could be useful for interpreting experiments .
this paper is organized as follows .
section ii defines the @xmath0 bilayer problem studied in this article .
section iii explains the principle of our numerical treatment .
section iv presents a detailed study of the sdips - induced dgs .
section v shows the effects of the sdips on the self - consistent superconducting gap and on the oscillations of the zero - energy dos with the thickness of @xmath2 .
section vi discusses the data of ref . .
section vii concludes .
we consider a diffusive @xmath0 bilayer consisting of a standard bcs superconductor @xmath3 for @xmath14 , and a ferromagnet @xmath2 for @xmath15 . we characterize the normal quasiparticle excitations and the superconducting condensate of pairs with usadel normal and anomalous green s functions @xmath16 and @xmath17 , with @xmath18 the superconducting pairing angle , which depends on the spin direction @xmath19 , the matsubara frequency @xmath20 , and the spatial coordinate @xmath21@xcite .
the usadel equations describing the spatial evolution of @xmath22 write @xmath23 in @xmath3 and @xmath24 in @xmath2 , with @xmath25 } \ ] ] in the above eqs .
, @xmath26 denotes the bulk gap of the @xmath3 material , @xmath27 the superconducting coherence lengthscale , @xmath28 the magnetic coherence lengthscale , @xmath29 the diffusion constant in @xmath30 and @xmath1 the ferromagnetic exchange field of @xmath2 .
the self - consistent superconducting gap @xmath31 occurring in eq .
( [ usadels ] ) can be expressed as @xmath32=\frac{\pi k_{b}t}{2}\sum \limits_{\substack{\sigma\in\{\uparrow,\downarrow\}\\\left\vert \omega _ { n}\right\vert \leq\omega_{d}}}\left ( \sin(\theta_{n,\sigma})-\frac { \delta(x)}{\left\vert \omega_{n}\right\vert } \right ) \label{deltabis}\ ] ] with @xmath33 the debye frequency of @xmath3 , @xmath34 the bulk transition temperature of @xmath3 , @xmath35 the boltzmann constant , @xmath36 the temperature and @xmath37 the euler constant .
the above equations must be supplemented with a description of the boundaries of @xmath3 and @xmath2 .
we use @xmath38 for the external sides of the bilayer . for the @xmath0 interface
, we use the spin - dependent boundary conditions@xcite @xmath39+i\gamma_{\phi}^{f}\sigma\mathrm{sgn}(\omega_{n})\sin[\theta_{n,\sigma}^{f } ] \label{bcright}\ ] ] and @xmath40-i\gamma_{\phi}^{s}\sigma\mathrm{sgn}(\omega_{n})\sin[\theta_{n,\sigma } ^{s } ] \label{bcleft}\ ] ] with @xmath41 and @xmath42 .
these equations involve the reduced conductances @xmath43 and @xmath44 , the barrier asymmetry coefficient @xmath45 , the normal state conductivity @xmath46 of the @xmath30 material , and the junction area @xmath47 .
note that we have used a definition of @xmath48 which differs from that of ref .
, to ensure a symmetric treatment of @xmath49 and @xmath48 in eqs .
( [ bcright]-[bcleft ] ) .
the microscopic expressions of the conductances @xmath4 , @xmath5 and @xmath6 can be found e.g. in ref . .
the term @xmath4 corresponds to the usual tunnel conductance of the interface .
the terms @xmath5 and @xmath6 can be finite only in case of a spin - dependence of interfacial phase shifts ( sdips ) .
the sdips results from the fact that the scattering phases picked up by electrons upon scattering by the @xmath0 interface can depend on spin due to the ferromagnetic exchange field or to a spin - dependent interface potential .
thus , in principle , any kind of @xmath0 interface can have a finite sdips .
however , the exact values of @xmath5 and @xmath6 are difficult to predict because they depend on the detailed microscopic structure of the interface .
one possible approach is to consider @xmath5 and @xmath12 as fitting parameters which have to be determined from proximity effect measurements .
note that the derivation of the boundary conditions ( [ bcright]-[bcleft ] ) assumes a weak transmission probability per channel ( tunnel limit ) , which seems reasonable considering the band structure mismatch between most @xmath3 and @xmath2 materials .
it furthermore assumes that the system is weakly polarized . however , there is no fundamental constraint on the amplitudes of @xmath4 , @xmath5 and @xmath6 because these parameters consist of a sum of contributions from numerous conducting channels . in @xmath0 circuits , long - range triplet correlations ( between equal spins )
can occur when the circuit includes several @xmath2 electrodes or domains with non - colinear magnetizations@xcite .
recently , it has been found that this effect can also arise in @xmath0 circuits with spin - active interfaces , due to spin - flip interfacial coupling terms which are due e.g. to some misaligned local moments at the @xmath0 interface@xcite . in our work
, we consider interfaces which are `` spin - active '' in the sense that the sdips is finite .
however , we assume that there is no interfacial spin - flip coupling and that @xmath2 is uniformly polarized . hence , we do nt obtain any long - range triplet component with our model .
equations ( [ usadels]-[bcleft ] ) have already been solved numerically with a self - consistent procedure in the case @xmath50 ( see e.g. ref . ) . in this paper
, we study the case of @xmath48 and @xmath49 finite , using a numerical treatment based on a relaxation method .
this treatment is divided into two steps .
we first calculate the values of @xmath31 and @xmath22 self - consistently with a relaxation method in imaginary times .
then , we determine the pairing angle @xmath51 corresponding to the calculated @xmath31 by using a similar relaxation method in real times , i.e. we use @xmath52 and @xmath53 in eqs .
( [ usadels]-[bcleft ] ) , with @xmath54 the energy , and @xmath55 a rate which accounts for inelastic processes @xcite .
finally we obtain the dos @xmath56 at position @xmath21 by using @xmath57 $ ] , with @xmath58 the normal dos per spin direction . throughout this numerical treatment , we use a discretized space , with a step of @xmath59 in @xmath60 . in the following ,
we mainly focus on @xmath61 .
has studied analytically @xmath0 bilayers with @xmath62 , @xmath63 , @xmath64 , and @xmath65 .
our approach allows to go beyond this regime .
note that in figures [ fig2 ] to [ fig1 ] , the results are shown for @xmath66 , @xmath67 and @xmath68 .
( top left panel ) and @xmath69 ( top right panel ) at the left and right side of the superconductor respectively , and density of states @xmath70 at the right side of the ferromagnet ( bottom panel ) , plotted versus @xmath54 , for different values of @xmath8 .
the full lines correspond to our numerical results .
the black dashed lines correspond to the analytical predictions of ref . , for @xmath71 .
] the left and right top panels of fig .
[ fig2 ] show the densities of states @xmath72 and @xmath69 at the left and right side of the superconductor respectively , for different values of @xmath8 , and the bottom panel of fig .
[ fig2 ] shows the corresponding dos @xmath70 at the right side of @xmath2 . for @xmath73 ,
the curves calculated with our numerical code ( black full lines ) are in close agreement with the analytical solution given in ref .
( black dashed lines ) .
the dos is similar at the two sides of @xmath3 and displays a dgs which reveals the existence of an effective zeeman splitting of the form @xmath74 with @xmath75 the thouless energy of the @xmath3 layer and @xmath76 its normal state conductance .
the dgs is also visible in @xmath70 due to the proximity effect .
it becomes narrower when @xmath8 increases , in agreement with eq .
( [ heff ] ) which indicates that @xmath7 scales with @xmath11 . for very large values of @xmath8 ,
the dos @xmath72 at the left side of @xmath3 tends to the bulk value @xmath77 , with @xmath78 ( see top left panel , blue full line ) .
however , a dgs remains clearly visible in @xmath70 , a result which is quite counterintuitive considering the low @xmath79 expression eq .
( [ heff ] ) ( see bottom panel , blue full line ) .
note that in the @xmath3 layer , with the parameters of fig .
[ fig2 ] , @xmath13 and @xmath80 [ @xmath81 , @xmath82 decays from its bulk value to @xmath83 on a distance of the order of @xmath9 @xmath84 $ ] near the interface ( not shown ) . in the large @xmath8 limit
, the dos @xmath83 at the left side of the @xmath0 interface does not show a clear dgs for the weak value of @xmath48 used in fig .
[ fig2 ] , because of the strong dos peak at @xmath85 ( see top right panel , blue full line ) .
however , a dgs would appear more clearly in @xmath83 for larger values of @xmath48 , e.g. using @xmath86 ( not shown ) .
the dgs thus seems to persist at large values of @xmath8 due to an effect which involves a @xmath3 area with thickness @xmath87 near the @xmath0 interface and the whole @xmath2 . due to the ferromagnetic exchange field @xmath1 , the zero - energy dos @xmath88 oscillates around its normal state value @xmath89 when @xmath90 increases@xcite . in the large @xmath8 limit
, we find that the dgs can occur at the @xmath2 side for both an ordinary ( @xmath91 ) and a reversed ( @xmath92 ) dos .
however , its visibility varies with @xmath90 , like in the limit @xmath62 of ref . .
figure [ fig4 ] , left panel , shows @xmath93 versus @xmath54 , for different values of @xmath90 and @xmath94 . from @xmath95 to @xmath96 , @xmath97 is positive and the visibility of the dgs increases ( see black , blue and green full lines ) . for @xmath98 ,
@xmath97 is negative and the dgs is not visible anymore ( see red full line ) . for @xmath99 ,
the dgs is visible again , with both the inner and outer peaks of the dos inverted due to @xmath100 ( see pink dashed line ) . at the right side of the ferromagnet ,
plotted versus @xmath54 , for different values of @xmath90 ( left panel ) and different values of @xmath48 ( right panel ) . in the left panel , for @xmath101 , we have multiplied @xmath102 by a factor 50 for visibility of the curve . ] in the parameters range investigated by us ( with in particular @xmath103 , @xmath104 and @xmath105 ) , no dgs occurs when @xmath106 .
the dgs studied in this article thus seems to be a direct consequence of @xmath107 for @xmath108 large as well as @xmath108 small .
figure [ fig4 ] , right panel , shows the variations of @xmath70 with @xmath48 , for a constant value of @xmath90 . for @xmath106 ,
no dgs appears . for a very small @xmath48 ( see red full line , corresponding to @xmath109 or @xmath110 )
, @xmath70 shows a change of slope which corresponds to a smoothed dgs , near @xmath111 ( from the previous section , this dgs occurs only for certain values of @xmath90 ) .
when @xmath48 becomes sufficiently large , @xmath70 shows a clear dgs , i.e. two peaks , one above and one below @xmath85 , while a local minimum is visible for @xmath112 .
the distance between the two peaks of @xmath70 increases with @xmath48 . in fig .
[ fig4 ] , when @xmath48 becomes too large ( @xmath113 ) , the sign of @xmath97 changes .
this suggests that @xmath114 does not only induce dgss but also shifts the oscillations of @xmath70 with @xmath90 .
this last effect will be investigated in more details for @xmath80 in section [ compl ] . with the parameters of fig .
[ fig4 ] , the dgs is not visible anymore when @xmath48 becomes larger than approximately 1 . in the general case ,
this threshold strongly depends on the different parameters characterizing the @xmath0 bilayer . in order to have a better insight on the persistence of the sdips - induced dgs at large values of @xmath8 , we provide in this section an analytic description of the case where @xmath3 is semi - infinite . for simplicity ,
we assume that the superconducting gap is only weakly affected by the presence of the @xmath2 layer , i.e. @xmath115 .
we furthermore assume that the proximity effect is weak , i.e. @xmath116 and @xmath117 , with @xmath118 . in this case , the usadel eqs .
( [ usadels]-[usadelf2 ] ) lead to : @xmath119 k_{n,\sigma}}{\xi_{f}}\right ) /\cosh\left
( \frac { d_{f}k_{n,\sigma}}{\xi_{f}}\right ) \label{thetaf}\ ] ] for @xmath120 and @xmath121 for @xmath122 , with @xmath123 .
we have introduced in the above eqs . @xmath124 and @xmath41 .
the linearization of the boundary conditions ( [ bcright]-[bcleft ] ) with respect to these two quantities leads to : @xmath125 and@xmath126 \frac{\left| \omega_{n}\right| } { \delta_{0 } } } \label{deltathetas}\ ] ] with @xmath127 $ ] .
importantly , eqs .
( [ thetaf]-[deltathetas ] ) are valid provided @xmath128 and @xmath129 , which requires @xmath64 , @xmath63 and @xmath65 .
we have used these hypotheses to simplify eq .
( [ deltathetas ] ) .
the validity of the approximation @xmath115 will be discussed in section [ compl ] . at the right side of the ferromagnet , plotted versus @xmath54 , for @xmath106 ( left panel ) and @xmath48 finite ( right panel ) .
the red lines are calculated from eqs .
( [ thetaf]-[deltathetas ] ) and the black dotted lines correspond to the self - consistent numerical resolution of eqs .
( [ usadels]-[bcleft ] ) . ]
figure [ fig5 ] shows @xmath70 calculated from the analytic continuation of eqs .
( [ thetaf]-[deltathetas ] ) ( black dashed lines ) and from our numerical code ( red full lines ) , for a weak value of @xmath130 , and @xmath106 ( left panel ) or @xmath107 ( right panel ) .
the two calculations are in relatively good agreement@xcite . for the parameters used in fig .
[ fig5 ] , we have checked numerically that the approximation @xmath115 gives results in very good agreement with the full resolution of eqs .
( [ usadels]-[bcleft ] ) . at @xmath131
, small discrepancies arise between the predictions of the numerical code and of eqs .
( [ thetaf]-[deltathetas ] ) , due to resonance effects which make @xmath132 and @xmath133 larger than for @xmath80 or @xmath134 . equations ( [ thetaf]-[deltathetas ] ) allow to recover the fact that a dgs can appear in @xmath70 , due to @xmath107 ( right panel ) . in the limit @xmath13 , the pairing angle of the system
can not be put under the form @xmath135)$ ] , contrarily to what has been found for @xmath62 .
therefore , the notion of sdips - induced effective zeeman splitting is not valid for thick @xmath3 layers . nevertheless , from eqs .
( [ thetas],[deltathetas ] ) , near the s / f interface , the resonance energies of the @xmath3 spectrum can be spin - dependent because quantum interferences make the superconducting correlations sensitive to the sdips on a distance of the order of @xmath136 near the @xmath0 interface . from eqs .
( [ thetaf],[thetaf2 ] ) , this behavior is transmitted to the whole @xmath2 layer due to the proximity effect . from eq . ( [ deltathetas ] ) , the energy scale related to the occurrence of the dgs has the form : @xmath137 with @xmath138 the normal state conductance of a slab of thickness @xmath9 of the @xmath3 material .
interestingly , this expression has a form similar to eq .
( [ heff ] ) , with @xmath8 replaced by @xmath9 ( one has @xmath139 ) .
note that for a ballistic @xmath0 single channel contact , the sdips also produces a spin - dependent resonance effect@xcite .
however , in this case , one does not obtain a dgs but rather a sub - gap resonance in the conductance and the zero - frequency noise of the system .
versus the spacial coordinate @xmath21 , for different values of @xmath130 ( left panel ) and @xmath48 ( right panel ) . ] for completeness , we now discuss the effects of the sdips on the self - consistent superconducting gap @xmath31 and the zero - energy dos @xmath88 versus @xmath90 .
it is already known that the amplitude of @xmath31 decreases when @xmath140 or @xmath130 increase , similarly to what happens in a @xmath141normal metal bilayer@xcite .
figure [ fig0 ] compares the effects of @xmath130 ( left panel ) and @xmath48 ( right panel ) on @xmath31 ( it only shows the effect of @xmath142 , but the effect of @xmath143 is similar ) .
one can see that @xmath31 significantly decreases when @xmath144 increases .
similarly , in a clean superconductor connected to a ferromagnetic insulator ( @xmath145 ) , @xmath31 has been predicted to decrease due to the spin - dependence of the reflection phases against @xmath145@xcite .
in contrast , in the regime of parameters investigated by us , @xmath49 has a negligible effect on the value of @xmath31 because it does not occur directly in the boundary condition ( [ deltathetas ] ) at the @xmath3 side of the interface . from this brief study of @xmath31 ,
we conclude that the approximation @xmath115 used in section [ analytic ] is valid only for sufficiently small values of @xmath140 , @xmath130 and @xmath48 . at the right side of @xmath2 , versus the thickness @xmath90 of @xmath2 .
the density of states calculated numerically is shown with red dots .
the full and empty dots correspond to @xmath146 and @xmath147 , respectively . the density of states given by eqs .
( [ thetaf]-[deltathetas ] ) is shown with black dashed lines .
panel ( a ) corresponds to a case with no sdips ( @xmath148 ) .
panels ( b ) and ( c ) show the effect of a finite @xmath49 .
panels ( d ) and ( e ) show the effect of a finite @xmath48 , in comparison with panel ( c ) where @xmath106 . with the parameters used here , eqs .
( [ thetaf]-[deltathetas ] ) are in agreement with our numerical code only when @xmath106 . ] figure [ fig1 ] presents the effects of @xmath49 and @xmath48 on the variations of @xmath88 with @xmath90 .
the dos calculated numerically is shown with symbols and the dos given by eqs .
( [ thetaf]-[deltathetas ] ) is shown with full lines . in panels ( a ) , ( b ) and
( c ) , we have used @xmath106 , so that the two calculations are in close agreement . in panels ( d ) and
( e ) , the two calculations strongly differ because @xmath48 is too large for the hypotheses leading to eqs .
( [ thetaf]-[deltathetas ] ) to be valid .
we recover the fact that , in the regime @xmath65 , @xmath88 shows exponentially damped oscillations with @xmath90 @xcite .
in the regime @xmath149 , the oscillations of @xmath88 with @xmath90 are less regular . this can be understood from the analytic description of section [ analytic ] . for @xmath65 ,
one has @xmath150 , so that @xmath88 depends on @xmath90 through the @xmath151 term of eq .
( [ thetaf ] ) only . for @xmath149 , @xmath152 and
thus @xmath133 strongly depend on @xmath90 , which complicates the variations of @xmath88 with @xmath90 and leads to more irregular oscillations .
has already shown that @xmath49 can shift the oscillations of @xmath88 with @xmath90 .
panels ( b ) and ( c ) confirm this result and also shows that a positive ( negative ) @xmath49 decreases ( increases ) the amplitude of @xmath88 . from panels ( d ) and
( e ) , @xmath48 can also significantly shift the oscillations of @xmath88 with @xmath90 , in agreement with fig .
[ fig4 ] , right panel .
for the parameters used in fig .
[ fig1 ] , @xmath48 does not modify spectacularly the amplitude of @xmath70 . for larger values of @xmath144 , the amplitude of the superconducting proximity effect would significantly decrease due to a reduction of @xmath31 ( not shown ) .
we now consider the dos measurements realized by sangiorgio et al . for bilayers with @xmath153 @xcite . from ref .
which considers samples fabricated by the same team , one has @xmath154 , so that @xmath155 .
however , double gap structures have been observed by sangiorgio et al . , which motivates a comparison with our model .
figure [ fig6 ] compares the data measured for @xmath156 ( black squares ) with our numerical calculation ( blue full line ) .
our calculation reproduces almost quantitatively the experimental curve .
we have used @xmath156 @xmath94 and @xmath157 , in agreement with refs . .
we have also used the exchange field @xmath158 , estimated by ref . , and the debye temperature @xmath159 of taken from ref . .
we have assumed @xmath160 , @xmath161 , @xmath162 , @xmath163 , and @xmath164 .
note that from ref . , one has @xmath165 @xmath166 and @xmath167 @xmath166 , so that one should have @xmath168 with the above values of @xmath9 and @xmath169 . in figure
[ fig6 ] , we have used a value @xmath170 , which is in relatively good agreement with this estimate .
the values of @xmath140 , @xmath49 , @xmath48 and @xmath130 used in the fit yield @xmath171 and @xmath172 .
a theoretical prediction of these ratios is very difficult because they depend on the detailed microscopic structure of the interface .
however , with a simple delta function barrier model , it already is possible to find situations where @xmath6 and @xmath173 are larger than @xmath4 ( see appendix a and fig . 6 of ref . ) . therefore , we think that the sdips parameters used by us are possible .
we can not reproduce quantitatively the data obtained by sangiorgio et al .
for all values of @xmath90 with the same set of parameters as for @xmath156 .
we think that this might be due to the fact that some characteristics of the samples like e.g. @xmath1 and thus @xmath169 , @xmath49 , @xmath48 , and @xmath130 can vary with @xmath90@xcite .
in the data of sangiorgio et al . , from @xmath156 to @xmath174 , the distance between the two peaks of the dgs increases , like in our model ( see fig .
[ fig4 ] , left panel ) .
however , the outer peak of the dgs remains very close to @xmath131 , which seems difficult to reproduce with our model .
note that in ref .
, for @xmath175 , the outer peaks of the dos are inverted , whereas a sharp dip occurs at low energies , which can give the impression that the innner peaks of the dos persist but are not inverted , in contrast to what we find ( see section iv.b ) . however , we think that the observation of this zero - bias dip is not totally reliable . indeed , sangiorgio et al .
explain that , in the dos of their thickest samples ( for @xmath176 ) , `` the zero bias peak is due to the steep voltage dependence of the background conductance and is therefore a by - product of the data normalization procedure '' .
the zero bias - dip of the @xmath175 sample occurs on the same energy scale as these zero - bias peaks , and it goes together with a strange zero - bias singularity similar to those shown by the thickest samples .
therefore , we are not sure whether a proper interpretation of the data of ref
. must take into account this feature .
suggests that the dgs observed by sangiorgio could be due to triplet correlations .
however , it is difficult to know whether this interpretation can be more satisfying than ours because ref .
does not shown any quantitative interpretation of the data and does not discuss , for instance , the evolution of the inner and outer peak positions with @xmath90 . , and our numerical calculation with @xmath177 , @xmath158 , @xmath178 , @xmath179 , @xmath94 , @xmath161 , @xmath162 , @xmath163 , @xmath170 , @xmath180 and @xmath164 .
in summary , we have calculated the density of states ( dos ) in a diffusive superconducting / ferromagnetic ( @xmath0 ) bilayer with a spin - active interface .
we have used a self - consistent numerical treatment to make a systematic study of the effects of the spin - dependence of interfacial phase shifts ( sdips ) .
we characterize the sdips with two conductance - like parameters @xmath6 and @xmath5 , which occur in the boundary conditions describing the @xmath3 and @xmath2 sides of the interface , respectively .
we find that the amplitude of @xmath31 significantly decreases if @xmath6 is too strong , whereas it is almost insensitive to @xmath5 .
in contrast , both @xmath6 and @xmath5 can shift the oscillations of the zero - energy dos of @xmath2 with the thickness of @xmath2 .
remarkably , we find that the sdips can produce a double gap structure in the dos of @xmath2 , even when the @xmath3 layer is much thicker than the superconducting coherence lenght .
this leads to dos curves which have striking similarities with those of ref . .
more generally , our results could be useful for interpreting future experiments on superconducting / ferromagnetic diffusive hybrid structures .
we acknowledge interesting discussions with a. a. golubov , t. kontos , m. yu . kuprianov and n. regnault .
we thank p. sangiorgio and m. beasley for discussions and for sending us their data .
j.l . was supported by the research council of norway , grants no.158518/431 and no .
158547/431 ( nanomat ) , and grant no .
167498/v30 ( storforsk ) .
d. huertas - hernando , yu . v. nazarov , and w. belzig , cond - mat/0204116 ; d. huertas - hernando , phd thesis , delft university of technology , the netherlands ( 2002 ) , d. huertas hernando , yu . v. nazarov , and w. belzig , phys .
* 88 * , 047003 ( 2002 ) .
the sdips is called by some other authors spin - mixing angle or spin - rotation angle . however , in our case , we consider a single ferromagnet which is uniformly polarized , so that the notions of mixing or rotation are not relevant .
m. eschrig , j. kopu , j. c. cuevas , and gerd schn , phys .
* 90 * , 137003 ( 2003 ) ; y. asano , y. sawa , y. tanaka , and a. a. golubov , phys .
b * 76 * , 224525 ( 2007 ) ; a. v. galaktionov and m.s .
kalenkov , i.e. tamm , a. d. zaikin , phys .
b * 77 * , 094520 ( 2008 ) ; m. eschrig , t. lofwander , nature phys . * 4 * , 138 ( 2008 ) . for the parameters of figs .
[ fig2 ] and [ fig4 ] , equations ( [ thetaf]-[deltathetas ] ) are not in good agreement with our numerical results because the hypotheses leading to these equations are not satisfied ( see section [ compl ] . ) | we calculate the density of states ( dos ) in a diffusive superconducting / ferromagnetic bilayer with a spin - active interface .
we use a self - consistent numerical treatment to make a systematic study of the effects of the spin - dependence of interfacial phase shifts ( sdips ) on the self - consistent superconducting gap and the dos . strikingly , we find that the sdips can induce a double gap structure ( dgs ) in the dos of the ferromagnet , even when the superconducing layer is much thicker than the superconducting coherence lenght .
we thus obtain dos curves which have interesting similarities with those of phys .
rev .
lett . * 100 * , 237002 ( 2008 ) . |
quantum effects in transport phenomena in classical systems represent an interesting fundamental issue in quantum theory that started from a remark of einstein in his celebrated paper on torus quantization @xcite .
one of such effects , widely studied in the context of time - periodic systems , is the quantum suppression of classical chaotic diffusion @xcite , or dynamical localization ( dl ) for short .
remarkably , this effect is a quantum manifestation of the fact that a time - periodic force can stabilize a system , and it is thus expected to play a key role in our understanding of the problem of quantum - classical correspondence in classically chaotic systems @xcite . while it is natural to think that , with the period fixed , this effect must depend on the temporal rate at which energy is transferred from the driving mechanism to the system , i.e. , on the force waveform
, the main target of study up to now has only been their dependence on the force amplitude because of the traditional use of sinusoidal forces .
recent work has provided strong evidence for a different dependence of dl on sinusoidal and square - wave forces @xcite . since there are infinitely many different waveforms , a natural question arises : how can the influence of the shape of a periodic force on dl be quantitatively characterized . in this work
, we demonstrate that for space - periodic systems subjected to a generic ac time - periodic force with equidistant zeros such characterization is well provided by a single quantity : the _ impulse _ transmitted by the force over half - period hereafter called _ force impulse_. this impulse is a quantity that accounts simultaneously for the force s amplitude , period and waveform . the organization of the paper is as follows . in sec .
[ sec.theory ] , we describe our model system of ultracold atoms in a periodically shaken optical lattice as well as our theoretical approach to characterize the effect of the force impulse on dl . section [ sec.results ] provides numerical confirmation of the effectiveness of the force impulse according to the theoretical predictions of sec .
[ sec.theory ] as well as discussing its influence on the quantum dynamics and underlying classical structures .
finally , we conclude by presenting in sec .
[ sec.conclu ] the main conclusions of our work along with some final remarks .
the dynamics of our model system of ultracold atoms interacting with a phase - modulated light field produced using an oscillating mirror is well described by the periodic hamiltonian @xmath0 , \label{eq.htilde}\ ] ] where @xmath1 is the atomic mass , @xmath2 the position , @xmath3 the momentum , @xmath4 the potential height , @xmath5 the wave number , @xmath6 the dimensionless modulation depth , and @xmath7 the ac force given by @xmath8 { \textnormal{dn}}\left [ \frac{4kt}{t}\right ] , \label{eq.f}\ ] ] where @xmath9 and @xmath10 are jacobian elliptic functions of parameter @xmath11 , and @xmath12 is the complete elliptic integral of the first kind .
@xmath13 is the normalization factor shown in fig .
[ fig.1](a ) , and given by @xmath14 } } , \label{eq.n(m)}\ ] ] where the values of the parameters are set equal to @xmath15 , @xmath16 , @xmath17 , and @xmath18 , in order to have the same force amplitude ( equal to unity ) and period , @xmath19 , independently of the waveform , i.e. @xmath20 $ ] , as shown in fig . [ fig.1](c ) . for @xmath21 , one recovers the well known harmonic excitation case previously considered for example in refs .
, since then @xmath22 .
on the other hand , for @xmath23 the waveform has different shapes .
for example , for @xmath24 a nearly square wave pulse is obtained , whereas for the limiting value @xmath25 the force vanishes . as will be shown below
, the force impulse associated with @xmath26 , defined as @xmath27 is a relevant quantity to characterize the effect of the force s waveform . as expected , it is a function of @xmath11 , which has a single maximum at @xmath28 , as shown in fig .
[ fig.1](b ) . also , it tends to zero very quickly as @xmath29 .
, given by eq . , _ vs.
_ @xmath11 .
( b ) normalized force impulse , @xmath30 $ ] , given by eq . , _ vs.
_ @xmath11 .
( c ) force @xmath26 , given by eq . , _ vs. _
@xmath31 , being @xmath19 the period , for four values of the shape parameter : @xmath21 ( red sinusoidal pulse ) , @xmath24 ( green nearly square - wave pulse ) , @xmath32 ( blue double - humped pulse ) , and @xmath33 ( pink sharp double - humped pulse).,title="fig : " ] , given by eq . , _ vs.
_ @xmath11 .
( b ) normalized force impulse , @xmath30 $ ] , given by eq . , _ vs.
_ @xmath11 .
( c ) force @xmath26 , given by eq . , _ vs. _
@xmath31 , being @xmath19 the period , for four values of the shape parameter : @xmath21 ( red sinusoidal pulse ) , @xmath24 ( green nearly square - wave pulse ) , @xmath32 ( blue double - humped pulse ) , and @xmath33 ( pink sharp double - humped pulse).,title="fig : " ] , given by eq . , _ vs.
_ @xmath11 .
( b ) normalized force impulse , @xmath30 $ ] , given by eq . , _ vs.
_ @xmath11 .
( c ) force @xmath26 , given by eq . , _ vs. _
@xmath31 , being @xmath19 the period , for four values of the shape parameter : @xmath21 ( red sinusoidal pulse ) , @xmath24 ( green nearly square - wave pulse ) , @xmath32 ( blue double - humped pulse ) , and @xmath33 ( pink sharp double - humped pulse).,title="fig : " ] it is important to remark that our model accurately describes a system of ultracold atoms in a periodically shaken optical lattice in the absence of a ratchet effect , i.e. with no directed transport due to symmetry breaking of zero - mean forces @xcite .
nevertheless , it has been recently shown that optimum enhancement of ratchet transport is achieved when maximal effective ( critical ) symmetry breaking occurs , which is in turn a consequence of two reshaping - induced competing effects : ( i ) the increase in the degree of symmetry breaking , and ( ii ) the decrease in the ( normalized ) transmitted force impulse @xcite , thus confirming the general relevance of this quantity .
switching now to scaled dimensionless variables @xmath34 the hamiltonian in eq .
is transformed into the following dimensionless one @xmath35 , \label{eq.3}\ ] ] where @xmath36 , @xmath37 , and @xmath38 . by rewriting eq . in the form @xmath39 , with @xmath40 \right\ } , \end{aligned}\ ] ] @xmath41 can be regarded now as the hamiltonian of an effective perturbed pendulum , being @xmath42 the perturbation for @xmath43 , such that expression @xmath44 \right\}\ ] ] accounts for the effect of the perturbation on @xmath45 .
it has been shown for the cases of sinusoidal @xcite and square - wave @xcite forces that the strength of the dl of system is correlated with the chaotic layer width of the aforementioned perturbed pendulum .
for the present case , one can obtain an analytical estimate of the chaotic layer width for the case of small modulation amplitude such that @xmath46 \approx \sin x-\lambda f\left ( \tau ; m\right ) \cos x\ ] ] by calculating the change of the averaged energy along the separatrix @xmath47 \right\ } , \notag \\
p_{0,\pm } \left ( \tau \right ) & = \pm 2\sqrt{\kappa } { \textnormal{sech}}\left [ \sqrt{\kappa } \left ( \tau -\tau _ { 0}\right ) \right]\end{aligned}\ ] ] of the unperturbed pendulum @xmath45 @xmath48 where @xmath49 is an arbitrary initial time @xcite . after some simple algebraic manipulation ,
one straightforwardly obtains @xmath50 where @xmath51 the width function @xmath52 gives a first - order approximation in @xmath53 to the width in energy of the chaotic separatrix layer . as a function of @xmath11 , it presents a single maximum at @xmath54 , which is significantly near @xmath55 , in the sense that the respective wave forms are almost coincident ( see discussion in sec .
[ subsec.calc ] below ) .
also , @xmath56 , as the force vanishes in that limit _ irrespective _ of the ( finite ) value of @xmath6 .
thus , one would expect the dependence of the width function on the shape parameter to remain approximately ( qualitatively ) valid for values of the dimensionless modulation depth which are beyond the first - order perturbative regime .
in this section we compare the classical and quantum dynamics of our system using different tools .
first , by calculating the momentum distributions , we identify the classical footprint of dl in the quantum dynamics .
second , to further demonstrate the influence of the classical phase space structures in the quantum behavior of the system , we compare the ( classical ) poincar surfaces of section ( psos ) and the corresponding husimi based quantum surfaces of section ( qsos ) @xcite , obtaining an excellent concordance between the results yielded by both tools . , and quantum mechanical ( medium red lines ) , @xmath57 , momentum distributions of atoms , and their differences ( thick green lines ) , @xmath58 , as functions of the modulation amplitude , @xmath59 , for @xmath60 , @xmath61 and four modulation waveforms : ( a ) @xmath21 , ( b ) @xmath62 , ( c ) @xmath63 , and ( d ) @xmath64 .
insets : wave forms over a modulation period . ] the classical momentum distribution for our system is numerically calculated as @xmath65 where @xmath66 is the evolution of a uniform distribution over one wavelength with a gaussian momentum distribution characterized by a width of @xmath67 @xcite given by the corresponding liouville equation , defined as @xmath68\right)p\left(x , p;\tau\right)=0,\ ] ] subjected to periodic boundary conditions @xmath69 , @xmath70 , with @xmath71 being the size of the quantization box and @xmath72 . on the other hand , quantum mechanical momentum distributions , @xmath73 , were computed by averaging time propagations with the time - dependent schrdinger equation @xmath74\right\}\psi \label{eq.13}\ ] ] of localized wave packets @xmath75 , \label{eq.gauss}\ ] ] centered at @xmath76 $ ] with @xmath77 and @xmath78 , where @xmath79 is the effective planck constant characterizing the `` quanticity '' degree of the system . the strength of dl is quantified by the difference between the classical and quantum mechanical momentum distributions @xmath80 , being @xmath81 the normalized root - mean - square momentum widths .
recall that the larger @xmath58 , the stronger the dl . in order to demonstrate that the force impulse is a relevant quantity properly controlling dl , we show in fig .
[ fig.2 ] the classical , @xmath82 , and quantum , @xmath83 , momentum distributions as a function of @xmath6 , as well as the difference , @xmath58 , for four different values of @xmath11 , i.e. four different modulation waveforms .
as it can be seen , the numerical results confirm that one needs larger values of the modulation amplitude @xmath6 to obtain a noticeable dl as the force impulse ( as a function of @xmath11 ) decreases from its maximum value at @xmath84 .
( solid red line ) , normalized width of the chaotic layer @xmath85 given by eq .
( dashed green line ) , and the normalized force impulse @xmath86 given by eq .
( dashed - dotted purple line ) _ vs. _ the shape parameter @xmath11 .
the values of the fixed parameters are : @xmath60 , @xmath87 , @xmath88 . ] in fig .
[ fig.3 ] , we show the difference between the classical and the quantum momentum distributions , @xmath58 , as a function of the shape parameter , @xmath11 , keeping the rest of the parameters ( @xmath89 , @xmath53 , @xmath90 ) fixed . in this figure , we have superimposed the chaotic layer width in energy and the force impulse .
as it is shown , @xmath58 presents its maximum values over a range @xmath91 , which is close to the impulse maximum @xmath92 [ cf . fig .
[ fig.1](b ) ] .
although not shown in fig .
[ fig.3 ] , the three plotted quantities vanish in the limit @xmath29 .
moreover , it is also important to remark the excellent agreement that exists between the chaotic layer width and the force impulse , that nearly coincide when adequately scaled . thus , we can unambiguously conclude the correlation existing between these two quantities characterizing , respectively , the degree of chaoticity of the system and the ac force . to further demonstrate the influence of the modulation waveform on dl , we show psoss for three different representative values of @xmath11 in fig .
[ fig.4 ] ( top panel ) .
these psoss have been stroboscopically computed at times multiples of the force period @xmath93 .
one sees that the modulation waveform , and hence the force impulse , does not only have a great influence on the the dl strength , but also on the phase space structures of the system .
thus , for @xmath21 a value of @xmath11 for which dl clearly occurs there are only very tiny islands of regularity ( leftmost panel ) , which are drastically reduced over the range of maximum strength of dl , as for @xmath94 for example ( central panel ) .
finally , for values of @xmath11 which are sufficiently close to 1 , as for @xmath95v(rightmost panel ) , dl is very weak , while the area occupied by regularity islands becomes even larger as @xmath29 .
clearly , this confirms the predicted dl scenario . in this section
we show how the quantum dynamics of our system , as it happens in general , is dramatically influenced by the underlying classical structures and , as a consequence , by the modulation waveform . to demonstrate this effect , we calculate qsoss for some representative values of the shape parameter @xmath11 . similarly to the classical psoss , we stroboscopically computed these qsoss at multiples of the force period , @xmath96 , using the husimi quasi - probability distribution @xcite @xmath97 with @xmath98 .
this expression can be rewritten as a sum of infinite terms , due to the existing periodic boundary conditions , as @xmath99 ^ 2}{2 \alpha}-
i \frac{(z+2\pi n ) p}{\hbar_\textrm{eff}}\right\ } \psi(z+2\pi n , t ) \ ; dz \ ; \right\vert^2 , \label{eq.hus2}\ ] ] although the calculation is effectively limited to the first three or four terms due to the gaussian decay . by averaging
the qsos over the times used in the calculation of the corresponding classical psos , we can quantum mechanically _ unfold _ the phase space structure of the system , which consists of a mix of islands of regularity and regions of stochasticity .
moreover , this averaged qsos can be directly compared with the corresponding classical psos .
some results for different representative values of the parameter @xmath11 are shown in fig .
[ fig.4 ] ( bottom panels ) , where the qsoss after averaging over 50 modulation periods for a coherent state centered at ( @xmath100 are depicted .
this initial condition gives rise to a classically chaotic trajectory .
one sees that , the three qsoss appear all spread out over the corresponding chaotic regions of phase space . also notice how the probability density does not enter into the regions where the motion is regular , since the corresponding tori and resonant islands act as total barriers both for the classical motion and the flow of quantum probability across .
this effect is clearly more noticeable as @xmath11 is increased , i.e. when larger portions of phase space become regular . as a consequence ,
the qsoss mimic to a great extend the corresponding classical psoss .
one point is worth mentioning here .
if we had propagated the initial wave functions for a much longer time , the corresponding qsoss would have entered to some extend inside these regions of regularity due to the usual quantum tunneling .
let us consider now the quantum dynamics associated with other types of classical motion .
when the wave function is initially localized at a point associated with a regular classical trajectory , whose dynamics is then confined to the corresponding invariant tori or chain of islands , the quantum quasiprobability density behaves accordingly , thus remaining well localized in the neighborhood of the tori ( except , again , for very long times when quantum tunneling becomes noticeable ) . for @xmath95 .
the corresponding classical poincar surface of section has been represented superimposed in green dots .
the number at the top right corner of each panel indicates the elapsed time in units of the modulation period @xmath96 . ]
an example is depicted in fig .
[ fig.5 ] , where we show the qsos corresponding to a wave function for @xmath95 and @xmath101 .
this initial condition generates a classical trajectory in the chain of islands associated with a 7:3 resonance ( see the corresponding psos in fig . [ fig.5 ] ) .
one sees that the localized wave function jumps over three of the islands at the instants @xmath102 and returns close ( although with some accumulated delay ) to the initial position at @xmath103 .
note that this is the behavior expected from the dynamics associated with the 7:3 classical resonance . .
all the panels have been ( stroboscopically ) averaged over 50@xmath19 , except panels ( a ) and ( d ) , which have been averaged over @xmath104 and @xmath105 , respectively . ]
it is worth noting that if we add all the qsoss distributions shown in fig .
[ fig.5 ] and renormalize the total , we get a new distribution highly localized on the corresponding classical psos .
this total distribution is depicted in fig .
[ fig.6](a ) , where the described effect is clearly visible .
figures [ fig.6](b)-(f ) show similar results for qsos localized on other relevant regular structures existing in the classical psos of fig .
[ fig.4 ] for @xmath95 . for this purpose ,
we have used wave functions launched at suitable initial conditions in phase space , and we have ( stroboscopically ) averaged the obtained qsos over 50@xmath19 .
one sees that these averaged qsos appear respectively localized on a tori around the central fixed point @xmath106 [ fig .
6(b ) ] , on a tori around the phase space point ( 0,2 ) [ fig .
6(c ) ] , on the 5:1 chain of islands around the same point [ fig .
6(d ) ] , on a tori around the border fixed point at ( @xmath107,0.5 ) [ fig . 6(e ) ] , and on a tori around the other border fixed point at ( @xmath107,2 ) [ fig .
in all cases , the qsoss completely mimic the structure of the classical phase space .
this clearly demonstrates the importance of this classical structures for the quantum motion .
it is worth mentioning that similar results to those presented here have been obtained for other values of the shape parameter @xmath11 .
summarizing , in this paper it has been shown for ac - driven space - periodic hamiltonians that the impulse transmitted by the time - periodic force is an essential quantity to optimally control the wave - packet spreading associated with the phenomenon of dynamical localization . while this result holds for the wide class of ac forces having equidistant consecutive zeros , one expects it to be also valid for even more general class of time - periodic forces , at least in the adiabatic regime @xcite .
this principle , which can be straightforwardly applied to other phenomena , such as field - induced barrier transparency @xcite or quasi - energy band collapse @xcite , paves the way for optimum coherent control of diverse quantum systems .
additionally , it has been shown that the classical invariant structures of phase space have a deep impact on the quantum dynamical behavior of the system , as the force impulse is varied .
this research has been supported by the ministry of economy and competitiveness ( mecc , spain ) under contract nos .
mtm2012 - 39101 , fis2012 - 34902 , and icmat severo ochoa , and junta de extremadura ( jex , spain ) though project n@xmath108 gr10045 . | it is shown that optimum control of dynamical localization ( quantum suppression of classical diffusion ) in the context of ultracold atoms in periodically shaken optical lattices subjected to time - periodic forces having equidistant zeros depends on the _ impulse _ transmitted by the external force over half - period rather than on the force amplitude .
this result provides a useful principle for optimally controlling dynamical localization in general periodic systems , which is capable of experimental realization . |
[ tab_elements ] & period@xmath3 & a@xmath3 & e@xmath3 & i@xmath3 @xmath4 + alpha & @xmath5 kg & @xmath6 km @xmath7 & @xmath8 years & & & + + beta & @xmath9 kg & @xmath10 km@xmath11 & @xmath12 days & @xmath13 km & @xmath14 & @xmath15 + gamma & @xmath16 kg & @xmath17 km@xmath11 & @xmath18 days & @xmath19 km & @xmath20 & @xmath21 + + + + the population of neas ( near - earth asteroids ) is composed by asteroids orbiting the sun with perihelion @xmath22 au . they are classified according to their orbital characteristics as : amor ( @xmath23 au and @xmath24 au ) , apollo ( @xmath23 au and @xmath25 au ) , atens ( @xmath26 au ) and ieo - interior to the earth s orbit ( @xmath27 au ) , where @xmath28 is the semi - major axis and @xmath29 is the aphelion distance of the orbit of the asteroid .
the population of neas is composed mostly by asteroids coming from the main - belt asteroids . according to @xcite , the three most important known sources of neas are the @xmath30 secular resonance with saturn , the 3:1 mean - motion resonance with jupiter , and the mars - crossing region .
once belonging to the neas population the close encounters of those asteroids with the terrestrial planets are quite frequent .
such characteristic makes the neas more accessible and so , interesting objects for exploration through space missions .
example of a successful mission to a nea is the hayabusa mission , performed by the japan aerospace exploration agency and that explored the nea ( 25413 ) itokawa in september , 2005 @xcite . the osiris - rex is a mission developed by nasa designed to return to the earth with a sample of the nea ( 101955 ) bennu .
the mission is planned to be launched in @xmath31 , and reach the nea in 2018 @xcite .
besides those examples , we can also cite the don quijote program @xcite , the ishtar program @xcite , the projects simone @xcite and marco polo @xcite , all of them designed to send space probes to a nea .
the multiple systems in the nea population are especially interesting for a space mission since they increase the observational possibilities . according to @xcite , approximately @xmath32 of the neas
are expected to be multiple system .
currently there are known 48 multiple systems in the neas population , being 46 binaries , and two triple systems : 2001 sn263 @xcite , and 1994 cc @xcite .
the nea 2001 sn263 is the target of the aster mission - first brazilian deep space mission @xcite .
it is a triple system with semi - major axis @xmath33 au , eccentricity @xmath34 and orbital inclination @xmath35 .
the components of the system have diameters of about 2.6 km ( alpha ) , 0.8 km ( beta ) and 0.6 km ( gamma ) .
with respect to the major body alpha , beta has a semi - major axis of 16.633 km - period of about 6.2 days , and gamma has a semi - major axis of 3.804 km - period of about 0.7 days @xcite .
see tab . 1 for more details .
the announcement of the aster mission has motivated a study aimed to characterize regions of stability of this system .
@xcite characterized the stability regions around the components of the triple system for the prograde case .
they performed numerical integrations , for a time span of two years , of a system composed by the sun , the planets earth , mars , jupiter , the three components of the triple system , and thousands of particles randomly distributed within the internal and external regions of the triple system , with relative inclinations between @xmath36 .
for the planar case , it were found stable regions placed very close to alpha and beta .
when the inclinations of the particles were considered , the effects of the kozai mechanism became observable turning the whole region around the three bodies unstable for particles with relative inclination @xmath37 , excepted for region 1 , where some particles survived .
it were also characterized mean motion resonances with beta and gamma in the internal region .
it was obtained that particles with @xmath38 km , experience the effects of a 3:1 resonance with gamma , while the particles with @xmath39 km and @xmath40 km , experience the effects of a 3:1 resonance with beta and of a 1:3 resonance with gamma , respectively . for the external region of the system ( besides gamma ) , it was found that the region is predominantly stable , and that the variation of inclination does not affect the stability of the particles . for a space mission aimed to place the probe in the internal region of the system
those results do not seem to be very interesting , since the stable regions found are tiny and also because those are the regions more likely to have debris .
therefore , knowing that stable regions are greater for retrograde orbits @xcite , here we present a complementary study , considering now particles orbiting the components of the system in the internal and external regions , with inclinations between @xmath41 , i.e. , considering the retrograde case . from the astronomy point of view ,
the analysis made before @xcite for prograde orbits are more relevant because it is believed that the potential dust orbiting the system came from the original disk that formed the bodies , so they orbit in the prograde direction . in this case ,
retrograde orbits are not very common for natural bodies , and they are valid only for particles that are captured by the system , so in a small number . on the other side , if the main goal is to study the system from the perspective of sending a spacecraft to visit the bodies , the retrograde orbits becomes an interesting option .
the spacecraft can easily be placed in prograde or retrograde orbits , with about the same expenditure of fuel .
therefore , it is important to study the stability of retrograde orbits .
the regions of stable retrograde orbits can be useful to place a spacecraft in orbits that does not need major station - keeping maneuvers in order to observe the three bodies of the system .
since the mission is planned to stay around one year in operation , a study of the stability of retrograde orbits are made for two years , that is a time that allows the designed mission with some extra time , in case the equipments last longer than planned .
the idea is to find regions of stable retrograde orbits in locations where the prograde orbits are not stable .
those orbits allow the spacecraft to be in stable orbits in regions unlikely to have particles that may damage the spacecraft .
those regions are identified and shown in detail in the present paper .
we present the regions of stability found for the retrograde case , and compare them with the regions found for the prograde case .
our goal is to complement the results presented by @xcite , given a complete scenario on the dynamic of particles in the neighborhood of the system , and thus , provide important parameters that may guide the design of the aster mission .
only the gravitational perturbations were considered in the present paper and in the previous one .
the solar radiation pressure is a relevant mechanism to be considered in a more complete analysis since it is known that depending on the area - to - mass ratio it may result in significant eccentricity increase , and so , affect the stability .
the analysis of how relevant this mechanism is for the specific problem of a space mission aimed to explore this triple system is the subject of a future work .
such analysis must be guided by the results presented here and by @xcite .
the structure of this paper is as follows . in sec .
[ sec_retrog ] , we present the problem of the stability around the triple system 2001 sn263 for the inclined retrograde case .
we present the initial considerations ( sec .
[ sec_ic ] ) , the adopted method ( sec .
[ sec_method ] ) , and the results obtained for all the regions ( sections [ sec_r1 ] to [ sec_external ] ) . in sec.[sec_compare ] , we compare the results presented here with those presented by @xcite . based on this comparison , we also discuss the regions within the triple system that are more interesting to place a space probe . in sec .
[ sec_conclusion ] , we present the conclusions of the work .
in order to achieve our goal , we first seek characterize the stable and unstable regions in the region around the components of the triple system 2001 sn263 , for the retrograde case .
we based the development of this work on the initial conditions and on the method adopted by @xcite , as follows .
we considered a system composed by the sun , the planets earth , mars and jupiter , the three components of the system 2001 sn263 , and by thousands of particles randomly distributed around alpha , beta and gamma ( internal region ) , and around the whole system ( external region ) .
the internal region was divided into four regions , defined by the approximated hill s radius of each body ( see fig .
[ fig_system ] ) .
the regions 1 and 2 are the regions where alpha is gravitationally dominant .
the region 3 is the region which beta dominates .
the limit of this region was calculated for the problem involving alpha and beta , resulting in a hill s radius of @xmath42 km .
the region 4 is the region which gamma theoretically dominates .
the limit of this region was calculated for the problem involving alpha and gamma , resulting in a hill s radius of @xmath43 km .
the region 1 is limited by the orbit of gamma , and the region 2 is limited by the orbit of beta .
the limits of the external region are better explained in sec .
[ sec_external ] .
we randomly distributed particles within the regions 1 , 2 , 3 , and in the external region .
the region 4 was not considered since it is too small , and would become even smaller if the perturbations from beta were considered .
all the particles started with an eccentricity going from @xmath44 until @xmath45 , and with a random angular distribution : @xmath46 , @xmath47 and @xmath48 , where @xmath49 is the true anomaly , @xmath50 is the argument of the pericentre and @xmath51 is the longitude of the ascending node .
the radial distribution of those particles depends on the region being considered .
regarding the inclination of the particles , we now considered the inclined retrograde case , meaning that all particles were distributed in the internal and external regions with an inclination with respect to the equator of alpha given in the range @xmath52 , taken every @xmath53 .
we considered the oblateness of alpha , with @xmath54 @xcite .
the obliquity of alpha was also considered and was determined through the pole solution , presented by @xcite beta and gamma were assumed to be spherical bodies with radius given in tab .
a non - sphericity of gamma would not affect our results since we did not consider particles around this body .
a non - sphericity of beta may change the results for region 3 , especially for those particles closer to this body .
such exception must be taken into account in further studies or in practical applications .
the method consisted on numerically integrate the equations of motion of the gravitational problem of n - bodies .
the numerical integrations were performed with the gauss - radau numerical integrator @xcite , for a time span of @xmath55 years .
this interval corresponds to about @xmath56 orbital periods of beta , and about @xmath57 orbital periods of gamma .
the time step of the numerical integrations was of @xmath58 days , for the internal and the external regions .
the physical data for alpha , and the orbital elements of beta and gamma are given by @xcite , for the epoch mjd 54509 in the equatorial frame of j2000 ( see tab .
the orbital elements of alpha , earth , mars and jupiter were obtained through the jpl s horizons system for the same epoch , also for the equatorial frame .
the planet saturn was not considered . due to the short time span of the numerical integration its gravitational perturbation , including the effects of the @xmath30 secular resonance ,
is negligible . throughout the numerical integrations ,
the particles could collide with alpha , beta and gamma , or be ejected from the system .
the collisions were defined by the physical radius of the three components .
the distance of ejection @xmath59 was defined for each region . for regions 1 and 2
it is defined by the collision - lines with gamma and beta , being @xmath60 km for region 1 and @xmath61 km for region 2 .
the ejection limit for particles in region 3 is given by the approximated hill s radius of beta , being @xmath62 km .
[ fig_system ] shows those limits .
the ejection distance for the external region is presented in sec .
[ sec_external ] .
the stability and instability are defined by the number of particles that survive throughout the numerical integration .
initial conditions that result in any loss of particles , by collision or ejection , define a stable region .
otherwise , we define the unstable regions .
following , we discuss special features of the method described above , and we present the results obtained for the internal region ( regions 1,2 and 3 ) and for the external region
. in region 1 , we considered particles orbiting alpha with semi - major axis @xmath63 km , taken every @xmath64 km , and with eccentricity @xmath65 , taken every @xmath66 . for each combination of @xmath28 and @xmath67 , we considered @xmath56 particles with random values of @xmath49 , @xmath50 and @xmath51 and with inclination going from @xmath68 to @xmath69 , taken every @xmath53 .
this combination of initial conditions resulted in a total of @xmath70 particles distributed within this region , for each inclination value .
the orbits were numerically integrated according to the method presented in sec .
[ sec_method ] .
the diagrams of fig .
[ fig_reg1 ] show the stable and unstable regions found in region 1 for the retrograde case .
they are composed by a grid of semi - major axis versus eccentricity , with each combination of @xmath28 and @xmath67 represented by small squares .
each of those boxes hold the information of 100 particles that share the same initial values of @xmath28 and @xmath67 .
the colors of the diagram indicate the percentage of survivors , going from @xmath71 of surviving particles ( yellow ) to less than 10% of surviving particles ( dark red ) .
the small black points indicate that @xmath72 of the particles survived ( stability ) .
the limits of the region are represented by the white lines .
they indicate the combination of @xmath28 and @xmath67 such that the pericenter of the orbits of the particles are inside the body alpha , meaning collision ( on the left ) , or such that the apocenter of the orbit is beyond the ejection distance ( on the right ) . in region 1
, the stable region substantially increases for inclination values @xmath73 .
this effect is related to the kozai mechanism @xcite . for the retrograde case _ the critical angle of kozai _
is @xmath74 .
particles with inclination values lower or near this critical angle suffer the action of this mechanism , which is known to generate oscillations of the eccentricity and of the mutual inclinations , leading to the noted instability .
the effects of the kozai resonance is entirely noted comparing the results presented in the diagrams of fig .
[ fig_reg1 ] with the results presented by @xcite for region 1 .
there is a symmetry for the prograde and retrograde cases , where particles with inclination values near the interval of the kozai cycles @xmath75 presented the predicted instability . out of this interval
we have larger regions of stability .
however , we see that for the retrograde cases , the stable conditions are much more prominent , as expected .
the instability appears only for the particles with initial conditions really close to the limits of the regions .
it can also be noted in these diagrams the occurrence of a gap in the stable regions for particles with @xmath76 km .
such characteristic suggests the action of a resonance .
this was investigated and is presented as follows .
we search for a commensurability of mean motion between particles with semi - major axis near @xmath76 km , with beta or gamma .
the mean motion was calculated through @xcite : @xmath77 \label{eq_mean}\ ] ] where @xmath78 is the gravitational constant , @xmath79 is the mass of the primary , @xmath80 is the radius of the primary , and @xmath81 is the gravitational harmonic of the primary related with the oblateness of this body .
we found that particles with @xmath82 km have a 2:1 commensurability of mean motion with gamma .
we then search the resonant angle for retrograde orbits ( see @xcite and @xcite ) .
we found that the resonant angle @xmath83 presents an intermittent behavior ( circulating and librating ) , as can be seen in figs .
[ fig_resonance]a and [ fig_resonance]b . in this equation ,
@xmath84 and @xmath85 is the mean longitude and the longitude of pericenter of the particle , while @xmath86 is the mean longitude of gamma .
therefore , particles with semi - major axis near @xmath76 km suffer the effects of a 2:1 resonance with gamma . as a consequence ,
the particles on this region are perturbed in such way that the semi - major axis remains almost constant while the eccentricity increases ( figs .
[ fig_resonance]c and [ fig_resonance]d ) , and then , the particles cross the collision - line of this region , giving rise to the noticed instability . in region 2
the particles orbit alpha with @xmath87 km , taken every @xmath64 km , and with @xmath65 , taken every @xmath66 . for each combination of @xmath28 and @xmath67
, we considered @xmath56 particles with random values of @xmath49 , @xmath50 and @xmath51 and with inclination going from @xmath68 to @xmath69 , taken every @xmath53 .
this combination of initial conditions resulted in a total of @xmath88 particles distributed within this region , for each inclination value .
the orbits of these particles were integrated following the method presented in sec .
[ sec_method ] .
the stable and unstable regions found for those conditions are presented in the diagrams of fig .
[ fig_reg2 ] .
as presented in section [ sec_r1 ] , we adopted again a grid of semi - major axis versus eccentricity and the color code is also the same .
the limits of the region are represented by the white lines .
they indicate the combination of @xmath28 and @xmath67 such that the particles cross the collision line with gamma , i.e. , the pericenter of the orbits are smaller than @xmath89 km ( on the left ) , or such that the apocenter of the orbit is beyond the ejection distance @xmath90 km ( on the right ) .
we see that for the cases with @xmath91 the region 2 presents significant stable regions .
the particles with @xmath92 have already crossed _ the critical angle of kozai _ or are near to the limit @xmath74 .
thus , the effects of such perturbation are no longer applicable , given rise to the noted stability . from @xcite , we see that , in fact , particles with @xmath93 are under the kozai mechanism action .
in contrast to the results presented by @xcite , showing that the region 2 is predominantly unstable for the planar and prograde cases , here , for the retrograde case , we found that this region is predominantly stable .
for initial conditions out of the kozai mechanism action , almost the entire region is stable .
in fact , the instability appears only for the particles with initial conditions really close to the limits of the region ( escape from the region or collisions due to crossing orbits with gamma or beta ) . in this region
we identified the action of mean motion resonances of particles with beta and gamma .
@xmath55 brings the details of those resonances .
they were found following the procedure described in sec .
[ subsec_resonance ] . for all of them
the resonant angles librate , as exemplified in fig .
[ fig_res_r2 ] for the resonance 3:1 with beta .
the effects of those resonances on the particles placed in region 2 are noted through small gaps within the stable cases , which is more evident for higher values of eccentricities .
[ tab_resonance ] angles@xmath94 + gamma & @xmath95 & @xmath96 & @xmath97 + gamma & @xmath98 & @xmath99 & @xmath100 + beta & @xmath101 & @xmath102 & @xmath103 + beta & @xmath104 & @xmath105 & @xmath106 + + + + in region 3 the particles orbit beta with @xmath107 km , taken every @xmath64 km , and with @xmath65 , taken every @xmath66 .
as previously done for the other regions , we considered @xmath56 particles with random values of @xmath49 , @xmath50 and @xmath51 and with inclination going from @xmath68 to @xmath69 , taken every @xmath53 , for each combination of @xmath28 and @xmath67 .
this combination of initial conditions resulted in a total of @xmath108 particles distributed within this region , for each inclination value .
the orbits of these particles were also integrated following the method presented in sec .
[ sec_method ] .
the stable and unstable regions found for those conditions are presented in the diagrams of fig .
[ fig_reg3 ] .
we adopted the same grid of semi - major axis versus eccentricity and color code for the diagrams .
the limits of the region are represented by the white lines .
they indicate the combination of @xmath28 and @xmath67 such that the pericenter of the orbits of the particles are inside the body beta meaning collision ( on the left ) , or such that the apocenter of the orbit is beyond the ejection distance @xmath109 km for region 3 ( on the right ) .
similar to the results for regions 1 and 2 , we note that the stable regions appear only when the relative inclination is about to reach _ the critical angle of kozai _ @xmath74 .
the effect of the kozai mechanism is more prominent for the particles in this region . from the diagram of fig .
[ fig_reg3 ] and from the diagrams presented by @xcite , for the region 3 and for the prograde case , we see that the stable regions completely vanish for inclination values near the interval @xmath93 .
similar to the results obtained for regions 1 an 2 , we also note a substantial increase of the stable region in region 3 , when comparing the retrograde case with the prograde case @xcite .
as found for regions 1 and 2 , here we also found that the instability appears only for initial conditions really close to the limits of the region .
thus , for inclination values out of the kozai mechanism action , we found that retrograde orbits in region 3 are predominantly stable .
we extend the analysis of the stability of the triple system 2001 sn263 to the external region .
the lower limit of the external region is @xmath110 km from alpha .
this distance corresponds approximately to the distance alpha - beta added to the hill s radius of beta .
the upper limit was determined taking into account the approximated hill s radius of the whole system . considering a body with mass equal to the sum of the mass of alpha , beta and gamma , and the heliocentric orbit of the system
, @xcite calculated that the hill s radius of such body is @xmath111 km at the perihelion of the orbit and @xmath112 km at the aphelion . according to @xcite
the predicted limit of stability for retrograde orbits is of about one hill s radius .
thus , the upper limit of the external region was chosen to be @xmath113 km ( 1 @xmath114 at perihelion , when the system is more perturbed ) . beyond this distance the particles
are considered ejected from the system .
thus , in the external region we have particles orbiting alpha with @xmath115 km , taken every @xmath116 km , and with @xmath65 , taken every @xmath66 .
as previously done , for each combination of @xmath28 and @xmath67 , we considered @xmath56 particles with random values of @xmath49 , @xmath50 and @xmath51 and with inclination going from @xmath68 to @xmath69 , taken every @xmath53 .
this combination of initial conditions resulted in a total of @xmath117 particles distributed within this region , for each inclination value .
the orbits of these particles were integrated following the method presented in sec .
[ sec_method ] .
the stable and unstable regions found for the external region are presented in the diagrams of fig .
[ fig_regext ] .
the limits of the region are represented by the white lines .
they indicate the combination of @xmath28 and @xmath67 such that the particles cross the collision line with beta , i.e. , the pericenter of the orbits are smaller than @xmath118 km ( on the left ) , or such that the apocenter of the orbit is beyond the ejection distance @xmath119 km ( on the right ) . comparing those diagrams with each other , we note that there is no significant difference in the stable and unstable regions .
the instability is observable only in the border of the region .
as previously found in @xcite for the planar and prograde cases , we concluded that the external region of the system is predominantly stable for the retrograde case , and that the stability of the external region of the triple system is not affected by the inclination variation .
for a space mission , the most interesting regions would be those that are unstable for the prograde cases , but stable for the retrograde cases . such configuration provide a stable region to place the mission probe for a period of 2 years with a relative retrograde orbit . at the same time , if we take into account that possible debris located in the neighborhood of the components of the system are more likely to have prograde orbits , we may have a region free of debris since they are expected to be quickly removed from the system due to the instability . the diagrams in fig .
[ fig_compare ] help us to compare the results for prograde and retrograde cases .
they were done considering only the particles in the internal regions of the triple system with @xmath44 , and with @xmath120 ( fig.[fig_compare]a ) , and with @xmath2 ( fig.[fig_compare]b ) .
these diagrams represent the radial distribution of the particles belonging to the set of initial conditions @xmath121 that after the numerical integration resulted in @xmath72 of survival ( yellow region marked with the small black dots ) . for the planar prograde case we may consider the hill stability , i.e. , a stability criterion that consider the restricted three - body problem and that provides a formula to calculate @xmath122 - the mutual orbital separation - such that the orbits remain stable @xcite . being alpha the central body ,
we calculate @xmath122 considering the problems : alpha - beta - particle ( @xmath123 ) and alpha - gamma - particle ( @xmath124 ) , where the particle have the external orbit and @xmath125 . for the alpha - beta - particle problem
, we found that @xmath126 km , meaning that particles beyond beta must have stable orbits for mutual distance beta - particle greater than this value .
for the alpha - gamma - particle problem , we found that @xmath127 km .
although this analytical approach does not take into account a third disturbing body and the effects of ressonances , we found that the results for the prograde case are in agreement with these predictions , as can be seen in the diagram of fig.[fig_compare]a .
we see a tiny stable zone located beyond gamma starting at @xmath128 km ( @xmath129 km ) , and beyond beta the stable orbits appear for particles with orbital radius of @xmath130 km ( @xmath131 km ) . from the diagram in fig.[fig_compare]b it became clear that for the retrograde case the whole region around the triple system 2001 sn263 is essentially stable . differently from the results obtained for the prograde case ,
it is clear here , for the retrograde cases , that the instability appears only for those particles orbiting alpha close to the orbits of gamma and beta .
although it is also possible to have stable particles around beta itself . in the diagram of fig .
[ fig_compare]c , we detached the preferred region with the orange color .
we note that in regions 1 and 3 there are tiny zones around alpha and beta that meets the requirement of being stable for retrograde orbits and unstable for the prograde ones . around alpha
, we have a region with radial length of 0,6 km ( region 1 ) , and of 1,0 km around beta ( region 3 ) .
the region 2 appears as the most interesting region .
it was characterized as unstable for prograde orbits , but is stable when the retrograde orbits are considered .
thus , we have a region with radial length of 8.2 km that is stable for the space probe for a period of 2 years and that must be free of debris .
therefore , we identify three internal regions that are very interesting to place the space probe , since they are stable for retrograde orbits but unstable for prograde orbits and allow the spacecraft to be in stable orbits close to the components of the system .
the first one is around the main body , in region 1 .
it is very useful to place the spacecraft to observe the main body more closely .
the second one is around the second larger body ( beta ) .
it is good to locate the spacecraft if the goal is to better observe this body .
the third one is in the regions between the two satellite bodies ( region 2 ) .
it is a very interesting region to keep the spacecraft to observe the three bodies of the system .
an elliptical orbit with periapsis near the lower limit of this region and apoapsis near the upper limit of this region would allow the spacecraft to pass closer to both secondary bodies in a single stable orbit that stays inside a region that is free of concentration of particles .
the external region are essentially stable for both , prograde and retrograde orbits .
nevertheless , we can detach a region of about @xmath132 km , going from @xmath133 km to @xmath134 km , beyond beta , that meet the _ preferred region _ requirement .
this would be a very interesting region to place a space probe .
there we may have a clean area with stable retrograde orbits around alpha that allow a close observation of the three components of the system .
in the present paper we analyze stability regions of the triple system 2001 sn263 , for retrograde orbits in the internal and external regions of the system . based on those results , we detached the preferred regions , i.e , internal and external regions that are interesting to place a space probe aimed to explore this system .
we applied the method adopted by @xcite , but now considering particles with retrograde orbits , i.e. , particles with @xmath41 relative to the equator of alpha , within and around the entire system .
we analyzed the behavior of such particles to determine the stable regions for the retrograde case , and we compared our results with the planar and prograde cases comparing our results with the planar and prograde cases we noted , as expected @xcite , a substantial increase of the stable region for both , internal and external regions when the retrograde orbits were considered .
we found that nearly the whole internal region is stable , especially for low eccentricities values .
the instability appears only in the border of the regions , or when the inclination of the particles are such that they are under the kozai mechanism action .
the most prominent result was found for region 2 . comparing the simulations for particles with @xmath120 with those with @xmath2
, we see that the region that is essentially unstable for the prograde case , became practically fully stable for the retrograde case .
we found that the external region is also predominantly stable .
the instability appears only in the borders of the region .
the variation of inclination does not affect the stability of the particles .
our results showed that the retrograde orbits are an interesting option for the aster mission .
we detached at least four regions ( three internal and one external ) that can be very useful to this mission .
all of them allow the space probe to closely observe the components of the system in stable retrograde orbits situated in regions expected to be free of debris .
those results provide important information about the stability around the components of the system and substantially contribute to the design of the aster mission .
this work was funded by inct - estudos do espao , cnpq and fapesp ( proc .
2011 - 08171 - 3 ) . this support is gratefully acknowledged .
we also thank ernesto vieira neto for his contribution with the numerical integrator .
everhart , e. an efficient integrator that uses gauss - radau spacings . in dynamics of comets : their origin and evolution , eds .
a. carusi carusi and g. b. valsecchi , d.reidel publishing company ( holanda ) , p. 185 - 202 , 1985 .
morbidelli , a. ; bottke , w. f. , jr . ; froeschl , ch . ; michel , p. origin and evolution of near - earth objects .
asteroids iii , w. f. bottke jr .
, a. cellino , p. paolicchi , and r. p. binzel ( eds ) , university of arizona press , tucson , p.409 - 422 , 2002 . | the nea 2001 sn263 is a triple system of asteroids and it is the target of the aster mission - first brazilian deep space mission .
the announcement of this mission has motivated a study aimed to characterize regions of stability of the system .
@xcite , characterized the stable regions around the components of the triple system for the planar and prograde cases . through numerical integrations they found that the stable regions are in two tiny internal zones , one of them placed very close to alpha and another close to beta , and in the external region . for
a space mission aimed to place the probe in the internal region of the system those results do not seem to be very interesting .
therefore , knowing that the retrograde orbits are expected to be more stable , here we present a complementary study .
we now considered particles orbiting the components of the system , in the internal and external regions , with relative inclinations between @xmath0 , i.e. , particles with retrograde orbits .
our goal is to characterize the stable regions of the system for retrograde orbits , and then detach a preferred region to place the space probe . for a space mission , the most interesting regions would be those that are unstable for the prograde cases , but stable for the retrograde cases . such configuration provide a stable region to place the mission probe with a relative retrograde orbit , and , at the same time , guarantees a region free of debris since they are expected to have prograde orbits .
we found that in fact the internal and external stable regions significantly increase when compared to the prograde case . for particles with @xmath1 and @xmath2
, we found that nearly the whole region around alpha and beta remain stable .
we then identified three internal regions and one external region that are very interesting to place the space probe .
we present the stable regions found for the retrograde case and a discussion on those preferred regions .
we also discuss the effects of resonances of the particles with beta and gamma , and the role of the kozai mechanism in this scenario .
these results help us understand and characterize the stability of the triple system 2001 sn263 when retrograde orbits are considered , and provide important parameters to the design of the aster mission .
[ firstpage ] retrograde orbits - triple system - neas - missions |
voids are some of the most striking large - scale features of the universe .
historically , their study can be loosely grouped into two categories : finding individual voids or using a statistical approach ( see @xcite for a detailed review of the history of void studies ) .
the first focuses on identifying individual voids with sophisticated void - finding algorithms @xcite that allow voids to have any convex shape .
the properties of galaxies in voids can then be studied , including their color distribution , luminosity function , concentrations and star - formation rates @xcite . @xcite
analyze the recently - completed 2df galaxy redshift survey and find that void galaxies are , on average , bluer and show evidence for more recent star - formation than the full 2df galaxy population .
properties of voids themselves can also be studied , including characteristic sizes , mean ellipticities , and radial density profiles .
many of these observational quantities , such as size measurements , can only be interpreted with carefully constructed mock galaxy catalogs .
extensive theoretical work has been carried out concerning void size distributions and density profiles @xcite , void halo properties @xcite , formation of galaxies within voids @xcite , and the properties of galaxies in voids @xcite
. the question of whether simulations can produce voids as large and as empty as voids seen in the observed universe remains unanswered , and can potentially provide powerful constraints on galaxy formation and cosmology @xcite .
the second approach to voids is somewhat more statistical and is in many ways complementary to the first .
studies of this type focus primarily on the void probability function ( vpf ) which is defined as the probability that a sphere of a given size centered on a random point in the survey volume contains no galaxies . statistical void distributions in principle offer a wealth of information , as they can be related to the entire hierarchy of galaxy correlation functions @xcite . yet , due to the difficulties in interpreting void statistics , they have not received as much attention as more conventional statistical measures of large - scale structure such as the two - point correlation function . despite these difficulties , the theoretical framework underlying void statistics has been developed in detail @xcite , and has been extensively studied in simulations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
observationally , voids statistics have been investigated in almost every major galaxy redshift survey , including the cfa @xcite , ssrs @xcite , pscz @xcite , iras @xcite , lcrs @xcite and most recently , the 2df survey @xcite .
results from the latter survey are representative of an emerging consensus ; subsamples of brighter and/or redder galaxies contain larger voids than samples of fainter and/or bluer galaxies .
this is interpreted as brighter and/or redder populations being more strongly clustered than fainter and/or bluer subsamples .
these trends are also reflected in the two - point correlation function of galaxies , @xmath2 , both at low ( e.g. , * ? ? ?
* ) and moderate ( e.g. , * ? ? ?
* ) redshifts .
we note that some studies of voids have not taken into account the fact that galaxy populations have different number densities which will strongly affect void statistics ; hence differences in the vpf across galaxy populations might as easily be attributable to the luminosity function as to varying clustering strengths .
as brighter galaxies are rarer than fainter galaxies , even in volume - limited samples , brighter samples will have higher void probabilities . a critical question we wish to address here
is the extent to which void statistics are governed by low - order clustering statistics .
it is therefore important to separate the effects of clustering from the effects of number density on the vpf .
void statistics can also be used to probe the relation between galaxies and dark matter .
there are strong theoretical arguments that show that the clustering strength of dark matter and galaxies should be different ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
galaxies initially form in high density peaks of the dark matter distribution and hence are `` biased '' tracers of the dark matter at high redshift .
baryonic physics and cosmic evolution also lead one to expect that the biasing between galaxies and dark matter should , in principle , be a function of scale , redshift , and galaxy properties such as color and luminosity .
many of these expectations have been borne out in observations ( see for example * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . by studying voids in observational samples and comparing to dark matter simulations
, one can hope to probe the galaxy bias in a unique way .
the vpf has previously been studied in simulations where a biasing prescription was assumed in order to place galaxies in the dark matter distribution ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite find that the vpf has limited utility in probing the bias , as the relation between dark matter and galaxies changes as a function of the sampling density of a galaxy survey . @xcite and
@xcite however , find that the vpf can be a powerful discriminator between various biasing schemes , but that it is relatively insensitive to the value of the linear bias factor @xmath6 .
for example , the vpf is very different , for fixed @xmath7 , when one statistically identifies galaxies with overdensities in the initial density field ( ` peaks biasing ' ) versus identifying galaxies with overdensities in the final matter distribution ( ` density biasing ' ) .
more recently , the halo model ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) has emerged as a useful prescription for placing galaxies within dark matter simulations . instead of using a biasing scheme such as ` peaks biasing ' or ` density biasing ' to relate galaxies to the dark matter distribution , the halo model places galaxies within virialized dark matter halos as a function of the halo mass .
although in principle the parameters of the halo model can evolve with cosmic time , there is evidence to suggest that the model does not strongly evolve from @xmath0 to @xmath1 @xcite .
@xcite find that the vpf can be used to determine halo model parameters such as the minimum dark matter halo mass , @xmath8 , in which a galaxy of a given luminosity may exist , and can hence provide useful new constraints on the relation between galaxies and dark matter . in this paper
we explore in detail the possibility of using the vpf to constrain the halo model . until now , there has not been sufficient data to study the statistics of voids at intermediate redshift ( @xmath9 ) . from correlation function measurements using the deep2 dataset at @xmath0 @xcite
we know that galaxies were in general less strongly clustered in the past ; one might expect this trend to be reflected in void statistics as well . as more data becomes available at higher redshifts , the importance of self - consistently investigating the evolution of galaxy properties and statistics can not be overemphasized .
this entails using similar selection and analysis techniques at different redshifts .
for this reason , we have chosen to include in this study an analysis of the sdss , which now has data for @xmath10 galaxies at @xmath11 , to allow for robust conclusions concerning the evolution of void statistics .
it should be kept in mind that the two approaches to the study of voids outlined above use the term `` void '' in very different ways .
while the first approach identifies voids as large underdensities in the galaxy distribution , allowing a void to assume any convex shape and allowing galaxies to exist inside voids , the second approach identifies voids as spherical regions in which surveyed galaxies are totally absent .
furthermore , while the first approach only locates `` unique '' voids , insofar as it does not allow smaller `` sub - voids '' to be contained within larger voids , the second approach allows for `` voids '' to overlap ( see @xmath12[s : deep ] for a visualization of this idea ) .
the statistical approach is only interested in the question : what is the probability that a given volume element in the universe is empty ? in this paper we are concerned with the second , statistical approach , and hence we define voids in the latter sense . the rest of this paper is organized as follows . in @xmath12[s : thy ] we outline the theory behind void statistics and point out several problems that arise with the theoretical underpinnings of voids .
@xmath12[s : meth ] describes our methodology .
we statistically analyze voids in mock galaxy catalogs built from n - body simulations in @xmath12[s : mocks ] ; in @xmath12[s : deep ] we present our results from the deep2 galaxy survey at @xmath0 and in @xmath12[s : sdss ] we analyze the sdss galaxy survey at @xmath1 .
@xmath12[s : disc ] discusses the implications and relevance of our results . throughout this paper
we assume a flat concordance @xmath5cdm cosmology with @xmath13 , @xmath14 and @xmath15 @xmath16 km s@xmath17 mpc@xmath17 .
the void probability function ( vpf ) is defined as the probability of finding no galaxies inside a sphere of radius @xmath18 , randomly placed within a sample .
the most common theoretical interpretation of the vpf is as an infinite sum of the hierarchy of correlation functions of the galaxy distribution @xcite .
specifically , for spherical volume elements one may write the vpf as : @xmath19,\ ] ] where @xmath18 is the sphere radius , @xmath20 is the average number of galaxies within the sphere , and @xmath21 is the volume averaged @xmath22-point correlation function , where the volume average is defined by @xmath23 as @xmath24 depends on a recurring factor of @xmath20 , any meaningful comparison of @xmath24 between populations requires a careful handling of their number densities . for many comparisons made in this study , we choose to remove the dependency on @xmath20 by randomly diluting the samples to have similar number densities . in this way we can isolate the effects of clustering on the vpf .
the vpf takes a much simpler form if one uses the hierarchical _ ansatz _ , @xmath25 to relate the hierarchy of correlation functions to the two - point function , @xmath3 .
this allows for a complete and relatively simple description of the entire cosmological density field .
ansatz _ has been formally derived only for an @xmath26 universe with the additional assumptions of stable clustering and self - similarity @xcite , which are both known to be invalid . in the linear regime ( @xmath27 mpc )
perturbation theory seems to validate the _ ansatz _ , although this paper is largely confined to smaller scales .
so far , the _ ansatz _ has not been ruled out by observations ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , although it is worth mentioning that baugh et al . had to remove the largest structure in their sample in order to recover scale invariant @xmath28 values .
there is no reason to believe that the _ ansatz _ should hold in the quasi - linear to strongly nonlinear regimes .
although historically the @xmath28 values were assumed to be scale invariant , nothing in the formalism developed below requires them to be . with the hierarchical _ ansatz _
the vpf becomes : @xmath29.\ ] ] @xcite noted that eqn .
[ eqn : fry ] could be manipulated to isolate the effects of the scaling coefficients ( @xmath28 ) .
fry defined the reduced void probability function , @xmath30 : @xmath31 which , with the substitution of eqn .
[ eqn : fry ] becomes : @xmath32 with this definition , @xmath33 becomes the independent variable , and hence only populations of galaxies with different @xmath28 values will have different reduced vpfs .
although @xmath33 will in general have a different dependence on @xmath18 for each galaxy population , this quantity will always be an increasing function of @xmath18 since @xmath34 and @xmath35 with @xmath36 for all of the populations considered here .
we now briefly summarize several models which predict values for the @xmath28 values . as the dynamical equations governing gravitational clustering can not be solved in the weakly to strongly nonlinear regime using perturbation theory ( or any of its offspring ) ,
various phenomenological models have been proposed to relate the higher order correlation functions to the two - point correlation function .
these models provide a complete description of gravitational clustering , yet each is inadequate either theoretically or observationally .
see fig .
[ fig : models ] for examples of predictions of the vpf from some of the more popular models ; for a detailed treatment see @xcite . in a poisson model ,
all moments with @xmath37 vanish , and the vpf can be described simply by : @xmath38 a gaussian model is almost as simple and equally inapplicable to the statistics of large scale structure in the universe on nonlinear scales at late times .
this model implies that the entire statistical distribution is described by the two - point moment ( in this case @xmath3 ) and that all higher moments identically vanish .
the gaussian and poisson models are obviously not actually hierarchical models ; we include them under hierarchical models for simplicity .
more realistic models rely on the hierarchical _ ansatz _ to specify all possible moments of the galaxy distribution .
these models can be thought of as prescriptions for the @xmath28 scaling coefficients .
the simplest model in this case forces @xmath39 for all @xmath22 and is sensibly called the ` minimal ' model @xcite . yet another model was motivated by an attempt to combine the hierarchical _ ansatz _ with the bbgky kinetic equations @xcite .
the bbgky model becomes unphysical for large void radii , as can be seen in fig .
[ fig : models ] where @xmath30 becomes negative for @xmath40 .
the ` thermodynamic model ' was first proposed by @xcite and arose from a theory which treated galaxy clustering by analogy to statistical mechanics .
it was later extended into a more self - consistent model by @xcite : @xmath41/\bar{n}\bar{\xi},\ ] ] @xmath42 intriguingly , a model developed by @xcite which combines an excursion set approach to the evolution of the halo mass function with a simple model for the spatial distribution of such halos predicts the same values for @xmath28 .
@xcite points out that this theory is only strictly true for large volumes , and that the derivations do not apply for @xmath43 ( @xmath44 @xmath45 mpc ) .
it is also difficult to understand how such large scales could have become thermodynamically relaxed over the age of the universe .
@xcite compared this model prediction to the vpf from the cfa survey , but their results were inconclusive .
they did find , however , that volume limited subsamples of different minimum luminosities were hierarchically related in that samples with different @xmath20 and @xmath3 , had similar @xmath28 .
the final model we consider has had a colorful history .
the negative binomial distribution , also known as the modified bose - einstein distribution , was used early on by @xcite to study the count distribution of charged hadrons resulting from high - energy collisions and by @xcite to describe the observed distribution of zwicky clusters .
the model can be described by : @xmath46 @xmath47 @xmath48^{1/\bar{\xi}}.\ ] ] @xcite analyzed the reduced vpf in the ssrs2 and cfa redshift surveys but could not discriminate between the thermodynamic and negative binomial models because of the size of their errors .
the negative binomial distribution was re - derived in their appendix by considering a sample divided into small cells , with the occupation probability of each cell depending only on @xmath3 , and being independent of the other cells ( see also * ? ? ? * ) .
unfortunately , the derivation contains little insight into the physical mechanisms that might drive point distributions to become negative binomial .
this model has also been derived from thermodynamic arguments @xcite .
@xcite and @xcite independently analyzed the cfa survey and found the data to be more consistent with the negative binomial than thermodynamic model over a range of luminosity thresholds and morphological types , though the agreement between model and data was not conclusive .
most recently , @xcite found that galaxies in the 2df survey follow the negative binomial model over a range of differential luminosity bins .
the astute reader will have noticed two problems with the theoretical models presented above , one related to convergence and the other to the unintuitive nature of the sums in eqns .
[ eqn : vpf ] , [ eqn : fry ] and [ eqn : chi ] .
the form for the @xmath28 values associated with the thermodynamic and negative binomial models ( eqns .
[ eqn : s_thermo ] and [ eqn : s_nb ] ) can be understood as arising from a taylor series expansion for the given @xmath30 values in eqns .
[ eqn : thermo ] and [ eqn : nb ] .
such an expansion is , however , only valid for @xmath49 ; the sum rapidly diverges if @xmath50 .
the conclusion must be that the @xmath28 values often quoted for these models can not be correct for large @xmath33 , i.e. for @xmath51 @xmath45 mpc ( the precise @xmath52 at which @xmath53 depends on the sample ) . realizing this issue , we have explored other forms for the @xmath28 values such that the sums in eqns .
[ eqn : fry ] and [ eqn : chi ] do converge .
simplistic models such as : @xmath54 where a and b are free parameters , are consistent with recent observations of @xmath28 for @xmath55 @xcite and are also consistent with @xmath56 for small @xmath22 .
although the sums converge with these @xmath28 values , they do not converge to a hierarchical model ; in fact , for many values of a and b , the sums converge to non - physical values ( @xmath57 ) .
it is unclear how far one should pursue this , given ( 1 ) that the @xmath28 values are probably scale - dependent , and here we have assumed that they are not , ( 2 ) there is no theoretical reason to suspect that the higher - order moments should be simply proportional to a power of @xmath2 , and ( 3 ) though mathematically valid , the expansion of @xmath24 in terms of @xmath58 is not terribly useful in practice . to illustrate ( 3 ) consider the following .
one might think that increasingly higher - order moments would be increasingly _ irrelevant _ for void statistics .
in fact , we have explicitly computed the sum in eqn .
[ eqn : vpf ] for the first eight correlation functions in large galaxy mock catalogs and find that the resulting partial sum in no way approximates the vpf measured from the same mock catalogs .
it seems that increasingly higher - order moments are , in fact , increasingly _
relevant_. in order to gain a clearer understanding of this we have investigated the sequence of partial sums of eqn . [ eqn : vpf ] using the @xmath28 values given in eqn . [ eqn : sp ] .
we find that the sequence of partial sums for @xmath24 oscillates wildly for @xmath59 , but rapidly converges for @xmath60 .
in other words , it is not the first few , but the first @xmath61 correlation functions that are necessary to accurately describe the void distribution . though it is comforting to know that the sum eventually converges , it is quite puzzling why convergence should require so many correlation functions .
one possible explanation for the importance of higher - order moments is the following : since there exist clusters with twenty objects or more , we might expect the 20-point correlation function to be non - zero .
this function describes the probability , in excess of random , that a region of space which contains 19 objects will contain a 20th .
since this will almost never happen for a random sample , the 20-point function will likely be quite large , at least on small scales .
hence one might expect contributions from @xmath22-point correlation functions as long as there are clusters of objects with @xmath22 members .
we next describe our general methodology for obtaining void statistics from large samples of galaxies .
issues concerning specific survey details , such as the proper handling of angular window functions , are treated as they arise in @xmath62 5 and 6 , where we use data from different surveys .
investigating the vpf and reduced vpf requires the measurement of three quantities : @xmath20 , @xmath3 , and @xmath24 , all of which are functions of the sphere radius , @xmath18 .
each of these quantities is straightforwardly determined by a counts - in - cells ( cic ) approach .
one simply places large numbers ( @xmath63 ) of random spheres within the survey and counts the number of galaxies contained within each sphere .
this is then repeated for many sphere radii .
@xmath20 is the average number of galaxies in a sphere : @xmath64 where @xmath65 to refer to the number of spheres and @xmath4 to refer to the number of galaxies . ]
is the total number of spheres placed and @xmath66 is the number of galaxies in the @xmath67th sphere .
@xmath24 is the number of spheres containing zero galaxies divided by the total number of spheres , @xmath68 where @xmath69 is the number of spheres that contain zero galaxies , and @xmath3 is the variance in the number of galaxies per sphere : @xmath70 in the limit of large numbers of random points , the cic approach for determining @xmath3 is known to be mathematically equivalent to more conventional methods @xcite .
we independently confirm this result by comparing our cic - measured @xmath3 to the volume - averaged correlation function obtained via the popular landy - szalay estimator @xcite .
we find that the two approaches are entirely consistent . to test the sensitivity of these measured quantities to the total number of spheres used , @xmath65 , we have computed @xmath24 , @xmath20 , and @xmath3 using a sphere radius of @xmath71 @xmath45 mpc , as a function of @xmath65 ( fig . [ fig : convergence ] ) , for mock galaxies in a simulation box of length @xmath72 @xmath45 mpc ( see below for simulation details ) .
as expected , for small @xmath65 , these quantities are unstable , but for @xmath73 the quantities converge well .
this rapid convergence holds for both smaller and larger void radii and is due to the fact that the simulation volume is well - sampled once the number of test volumes exhausts the number of independent configurations . to be conservative
, we use @xmath74 to calculate these quantities .
errors for all measured quantities are estimated by `` jackknife '' sampling . before moving on , we should highlight the fact that all quantities used in analyses of void statistics are spherically averaged .
this type of averaging effectively throws away much of the useful information contained in higher order statistics .
for example , @xmath75 measures the skewness of the distribution , and will in general _ not _ be spherically symmetric . hence it is unclear whether or not spherically averaging @xmath75 is even the proper way to average @xmath75 ; these ambiguities make interpretation difficult .
the @xmath3 used here and in other void analyses refers to the correlation function in _ redshift space _ ; the correlation function has not been de - projected into real space , which is the more conventional way of reporting @xmath2 .
this is important because , as we will see , the real space vpf measured in mock catalogs is different from the behavior of the vpf in redshift space , and our main conclusions turn out to be valid only in redshift space .
indeed , @xcite find that , at fixed void radii , @xmath24 in simulations is larger in redshift space compared to real space , with the difference increasing at larger void radii .
these authors also find that redshift space void statistics closely follows one or another hierarchical model ( which model the statistics follow depends on gross cosmological models , i.e. closed v.s open and biased vs. unbiased ) , while the real space void statistics do not .
one example of redshift - space effects relevant to this study is the flattening of the @xmath28 scaling coefficients in redshift space when compared to real space .
this is due to the reduction of clustering strength on small scales caused by peculiar velocities ( known as the fingers - of - god effect ) .
see @xcite for examples of this and other redshift space effects .
finally , we require that the number density of the sample under consideration be independent of redshift , and so construct volume - limited samples for our analysis .
specifically , we require that each galaxy be observable over the entire redshift range we consider .
in this section we analyze void statistics in mock galaxy catalogs constructed from n - body @xmath5cdm simulations in order to help understand and interpret void statistics recovered from observational samples .
first we measure the vpf and reduced vpf for dark matter particles and mock galaxies at @xmath1 and @xmath0 .
we then investigate the reduced vpf for different halo model halo occupation distributions ( hods ) and find that they are similar over a wide range of hod parameters .
next , we determine the effects of redshift space distortions on the vpf and finally we compare vpfs for halo centers alone to vpfs for mock galaxies . throughout this section
we use full simulation boxes at two outputs : @xmath76 ( `` @xmath1 '' ) and @xmath77 ( `` @xmath0 '' ) ; in @xmath12[s : deep ] , where we investigate the effects of survey geometry on the vpf , we extract light - cone geometries from the full simulation . except for @xmath12[s : rse ] , all analyses in this section are performed for galaxies , dark matter particles , and dark matter halos , in redshift space .
the mock galaxy catalogs we use were constructed specifically for the deep2 survey . a complete description of the catalogs is given in @xcite ; we give the relevant details here .
n - body simulations of @xmath78 dark matter particles with a particle mass @xmath79 @xmath80 were run in a @xmath5cdm universe using the treepm code @xcite in a periodic , cubical box of side length @xmath72 @xmath45 mpc .
dark matter halos were identified by running a `` friends - of - friends '' group finder . galaxies
were then inserted via a halo model approach , where galaxies are placed in dark matter halos using a simple prescription ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the main ingredient of the halo model is the halo occupation distribution , hod , which specifies the mean number of galaxies to be placed in a given halo , as a function of the halo mass ( @xmath81 ) . in its most common form
the hod is the sum of a power - law describing the sub - halo ( satellite ) population and a step function above some minimum mass , which describes the host halo ( central galaxy ) @xcite .
luminosities are then assigned to galaxies in a halo according to a conditional luminosity function ( clf ) , @xmath82 , which specifies the luminosity function of galaxies in halos of mass @xmath83 @xcite .
the clf is traditionally chosen to have a schechter form .
the _ same _ hod and clf are used to populate halos at both high and low redshift , with only the underlying dark matter distribution and the characteristic scale of the luminosity function , l@xmath84 , evolving with redshift ( evolution in l@xmath84 was taken to be 1 mag based on combo-17 data @xcite ) .
this `` no evolution '' hypothesis produces catalogs that are in agreement with two - point clustering measurements from deep2 at @xmath0 and the @xmath85-band luminosity function and two - point clustering of 2df galaxies at @xmath1 @xcite . for what we call the ` primary ' simulations ,
the specific parameters of the hod and clf are chosen to best match the two - point clustering measurements at @xmath0 .
this section uses the ` primary ' simulations as described above .
all samples drawn from simulations have been randomly diluted to the same number density ( @xmath86 @xmath87 mpc@xmath88 ) to allow for a direct comparison between vpf measurements ( recall that the vpf , as opposed to the reduced vpf , is quite sensitive to @xmath20 ) .
furthermore , all galaxies in these samples are restricted to have luminosities @xmath89 .
[ fig : mocks_vpf ] shows the vpf calculated for dark matter particles and galaxies at @xmath0 and @xmath1 in redshift space , where errors are estimated using jackknife sampling .
there are two obvious trends .
the first is that , at both redshifts , galaxies have a larger @xmath24 at all radii compared to dark matter , indicating that they are more highly clustered and that galaxies included here are biased relative to the dark matter .
this simply results from the fact that these galaxies live in massive dark matter halos that are large overdensities and hence are more highly clustered than the overall dark matter distribution .
this is not in conflict with the realization that at @xmath1 the galaxy distribution accurately traces the dark matter on large scales @xcite because ( 1 ) we are probing different scales and ( 2 ) these studies have only measured the linear and quadratic bias terms , whereas we are in principle sensitive to all non - linear biasing terms . the second result in fig . [
fig : mocks_vpf ] is that both galaxies and dark matter become more clustered with time , as seen by the larger voids at @xmath1 relative to @xmath0 ; this is due to the effects of gravity . in fig .
[ fig : mocks_rvpf ] we show the reduced vpf separately for low- and high - redshift mock samples as well as for two hierarchical models , the negative binomial ( dashed ) and thermodynamic ( dash - dot ) . here as elsewhere , the upper limit on @xmath33 in the reduced vpf is set by the radius at which @xmath90 ; at that point @xmath30 becomes undefined .
the error on the mean in each point ( computed by dividing the cube into octants and measuring void statistics within each octant ) is smaller than the plotted symbols ( @xmath91 ) , although errors are highly covariant between different sphere radii . in principle
there should be errors on @xmath33 in addition to @xmath30 , as @xmath33 will vary across the octants at a specific @xmath18 . in practice however , the error on @xmath33 is negligible as it is smaller than the binsize .
the dark matter vpf is well described by the thermodynamic model at all scales for both @xmath0 and @xmath1 , while the galaxy vpf fits the negative binomial model to a precision better than the achievable observational errors .
in fact , the dark matter matches the thermodynamic model well to at least @xmath92 , the highest redshift simulation output available .
this directly shows that the vpf can be entirely parameterized in terms of two variables , @xmath20 and @xmath3 , since , for a given hierarchical model , @xmath30 is a function of only of their product ( see eqn .
[ eqn : nbvpf ] ) .
the differences seen in the vpf for galaxies at low and high redshift ( fig .
[ fig : mocks_vpf ] ) are described entirely by the evolution of @xmath3 with cosmic time .
below we find that this dependence holds in the data as well , not only as a function of redshift but also for samples with varying galaxy properties at the same redshifts .
@xcite do not find this good agreement between the thermodynamic model and dark matter particles drawn from the hubble volume simulations .
we speculate that the much larger particle mass ( @xmath93 @xmath94 ) of these simulations does not afford an accurate measurement of the dark matter particle reduced vpf .
there is a small but perceptible up - turn in the reduced vpf for all samples in fig .
[ fig : mocks_rvpf ] at the largest @xmath33 measured .
a much larger up - turn has been seen in the cfa survey @xcite at similar void radii ( @xmath95 @xmath45 mpc , depending on the luminosity threshold ) .
this may be due to fluctuations in the initial conditions of the simulations , as both the galaxies and dark matter at low and high redshift were derived from the same simulation .
furthermore , these large scales ( corresponding to void diameters of @xmath96 @xmath45 mpc ) are approaching the limit at which a simulation with box size of @xmath97 @xmath98 mpc@xmath99 becomes unreliable due to the fact that large - scale modes , which become increasingly important on larger scales , can not be included in simulations with periodic boundary conditions .
future studies with different realizations of the initial conditions and larger box sizes will be able to address the significance of this deviation .
regardless , we are unable to probe these largest void scales with the currently available galaxy surveys . finally , we find that the good agreement between the reduced vpf in simulations and the negative binomial model is insensitive to random dilutions of the mock galaxies .
specifically , we randomly dilute the mock galaxy catalogs to @xmath100% , @xmath101% , and @xmath102% of the nominal number density and find no differences in the resulting reduced vpfs , as expected since removing dependencies on @xmath20 was a major reason for introducing the reduced vpf in the first place . defining the vpf as a sum of higher order correlation functions and then inserting the hierarchical _ ansatz _
provides a simple explanation for this insensitivity of the reduced vpf on @xmath20 . as mentioned above
, the halo model can be very useful for placing galaxies in cosmological dark matter simulations .
it also provides a way of understanding the vpf without resorting to an infinite sum of higher - order correlation functions .
specifically , @xmath103 where @xmath104 is the mass density contrast smoothed over a volume @xmath105 , @xmath106 is the probability of having a given smoothed mass density contrast , and @xmath107 is the probability of finding a halo with mass less than @xmath8 in a volume @xmath105 with smoothed density contrast @xmath104 .
here we have written @xmath108 explicitly as a function of the smoothing volume , but note that it is the same @xmath24 that occurs elsewhere .
the only input from the halo model is the value for @xmath8 , the mass above which halos contribute to @xmath24 .
@xmath106 is variously assumed in the literature to have a lognormal , negative binomial , or gaussian distribution , and @xmath107 is fixed by the relation between halos and dark matter ( the halo bias ) , which can be determined from simulations .
we would like to test the ability of the vpf to constrain the halo model , as suggested by @xcite .
their main results are ( 1 ) that the vpf is sensitive to m@xmath109 , the minimum mass that a dark matter halo must have to host a galaxy , and ( 2 ) that the vpf is entirely insensitive to the spatial distribution of galaxies within halos . to investigate these claims we calculate void statistics , in redshift space , for mock galaxy catalogs constructed from various hods .
the top panel in fig .
[ fig : hod ] shows five different hods with the simple form @xcite : @xmath110^{\alpha } & \textrm{for $ m > m_{min}$}\\ 0 & \textrm{for $ m < m_{min}$ } \end{array } \right.\ ] ] where @xmath111 and m@xmath112 and @xmath113 ( in units of @xmath114 @xmath45 m@xmath115 ) . for each m@xmath109
, m@xmath116 is determined by fixing the overall number density at 0.01 @xmath87 mpc@xmath88 .
these five models are referred to as b1 , b3 , b5 , b7 , and b9 , respectively . in order to clearly show small variations between the models , in fig .
[ fig : hod ] ( bottom panel ) we plot differences between the negative binomial model and the measured reduced vpf for each galaxy catalog .
the reduced vpf remains essentially unchanged for these different hods , leading us to conclude that the vpf can be entirely determined by @xmath3 , i.e. , by eqn .
[ eqn : nbvpf ] .
the scatter between the five models is @xmath117 , much smaller than the attainable accuracy with even the largest current galaxy surveys ; we therefore conclude that over a wide range of physically meaningful hods , the vpf can not constrain m@xmath109 beyond constraints attainable from @xmath3 alone . in the figure ,
the reduced vpf for the b1 hod is the most deviant from the negative binomial model .
however , observations of the two - point correlation function at @xmath0 @xcite rule this model out .
these results are consistent with the claims of @xcite , as they were concerned with void statistics in _
real space_. the halo model formalism becomes much more complex analytically in redshift space , due to the presence of peculiar velocities .
we find using simulations , that redshift - space effects essentially ` wash out ' any differences in the vpf due to different hods .
qualitatively , one would expect that placing more galaxies in halos would increase the @xmath28 values on small scales in real space .
however , in redshift space , small - scale power is more smeared out because there are now more galaxies residing in massive halos with higher velocity dispersions .
it is only a surprise that these two effects should act to cancel each other out exactly .
we now investigate the effects of redshift - space distortions on void statistics in detail .
the effects of redshift space distortions on clustering statistics have been studied in detail ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and have been probed with void statistics @xcite .
@xcite found that voids appear somewhat larger in redshift space and that the reduced vpf in real space does not agree with any of the hierarchical models .
we explore the extent to which redshift - space effects wash away information by calculating the vpf resulting from several hods with fixed @xmath8 and varying @xmath118 ( for @xmath119 ) , i.e. varying the number of galaxies placed in halos with mass above @xmath8 , while still keeping the overall number density fixed at 0.01 @xmath87 mpc@xmath88 .
one might imagine that , by fixing @xmath8 and varying @xmath118 , @xmath24 would remain unchanged , but @xmath3 would vary because adding more galaxies to a halo will strongly affect the small scale clustering . if this were true , then @xmath24 would not simply be a function of @xmath3 alone . by analyzing a suite of mock galaxy catalogs constructed with these various hods ,
we find , as expected , that @xmath24 remains unchanged .
yet surprisingly , but in accordance with the idea that redshift - space distortions conspire to ` wash out ' differences due to different hods , @xmath3 remains unchanged for @xmath120 @xmath45 mpc and is only mildly different for @xmath121 mpc . by showing that @xmath3 remains the same across different hods
when @xmath24 and @xmath20 remain the same , we further solidify our claim that @xmath24 is solely a function of @xmath122 , and hence that @xmath24 does not contain additional information beyond what is available in @xmath3 . to quantify the differences between real and redshift space
we calculate void statistics for our five hod models ( b1-b9 ) in real space and compare them to their redshift space analogs .
the top panel in fig .
[ fig : hod_rz ] displays the fractional probability increase of the vpf when comparing redshift space ( @xmath123 ) to real space ( @xmath124 ) .
two competing effects are at work here : ( 1 ) peculiar velocities smear out small scale clustering , and ( 2 ) kaiser infall , due to the coherent infall of structures on larger scales @xcite , which has the effect of increasing the size of voids on large scales in redshift space
. the general trends in this figure are consistent with the underlying hod of each sample .
for example , model b1 preferentially places galaxies in higher - mass halos compared to other hods , and more massive halos have larger velocity dispersions .
the larger dispersions decrease the size of voids as seen in redshift space , explaining why there are fewer voids in redshift space compared to real space for the b1 model .
this model also has the smallest differences on large scales .
this is in part because velocity dispersions are small compared to the largest voids , but also because of kaiser infall , which increases the apparent size of voids in redshift space on large scales , is important primarily for lower mass halos that are falling into forming structures .
since @xmath8 is so large in the b1 model , we expect that most halos in this model are not falling onto forming superstructures . in the bottom panel of fig .
[ fig : hod_rz ] we see that the reduced vpf can in fact uniquely constrain the halo model when the analysis is performed in real space .
the vpf for a set of catalogs can be entirely described by a single function @xmath125 , and hence can not be used to uniquely constrain models , if the reduced vpf for those models are all the same . yet for our five hod models , this is not the case in real space .
instead , there is a definite trend of @xmath30 with @xmath8 ; samples with smaller @xmath8 are less like the negative binomial model . on larger scales ( @xmath126 @xmath45 mpc ) the hods
are again indistinguishable , as is expected since hod parameters are relevant only on halo scales ( i.e. @xmath127 @xmath45 mpc ) .
although it seems intuitive that the vpf should be capable of uniquely constraining the hod ( specifically @xmath8 ) , redshift - space effects make this extremely difficult in practice .
we can better understand the insensitivity of the vpf to particular parameters of the hod by considering the vpf for the _ centers _ of dark matter halos ( restricted to have @xmath128 ) alone .
[ fig : halo ] compares the reduced vpf for dark matter particles , dark matter halo centers , and mock galaxies .
as the figure indicates , it is primarily the clustering properties of dark matter halos , not the number of galaxies within them , that generates agreement with the negative binomial model .
we find the similarity in reduced vpf for halo centers and mock galaxies intuitive for two reasons .
first , as most dark matter halos in our simulations with @xmath129 contain a single isolated galaxy , and galaxies in most halo models are placed first in the center of the halo , it is reasonable that the clustering of halo centers has more impact on the galaxy reduced vpf than @xmath118 .
second , when more than one galaxy is placed in a halo , it is typically within @xmath130 @xmath45 kpc of the halo center , as very few halos have larger radii .
hence a sphere containing a satellite galaxy will typically already include the galaxy at the halo s center .
this also explains why void statistics are insensitive to the spatial distribution of galaxies within halos , as noted by @xcite . in conclusion
, the vpf does not seem to be capable of providing competitive constraints on the major parameters of the hod .
this is due both to redshift space distortions washing out information on small scales and to the fact that the clustering of dark matter halos alone dominates the vpf .
although one might have thought that the vpf for galaxies and halos would be uniquely sensitive to @xmath8 , it turns out that any changes in @xmath8 are likewise reflected in @xmath3 , in a way that can exactly account for changes in the vpf .
we now measure void statistics in state - of - the - art redshift surveys that are mapping the three - dimensional positions of thousands of galaxies . in this section
we probe void statistics when the universe was roughly half its present age ( @xmath0 ) . in the following section we focus on voids in the local universe ( @xmath1 ) .
the deep2 galaxy redshift survey @xcite is an ongoing project that is gathering optical spectra for @xmath131 galaxies at @xmath0 using the deimos spectrograph on the keck ii 10-m telescope .
the completed survey will span a comoving volume of @xmath132 mpc@xmath99 , covering @xmath133 deg@xmath134 over four widely separated fields ; observations began july 2002 and are expected to be completed in mid-2006 . due to the high dispersion and excellent sky subtraction provided by the deimos spectrograph ( @xmath135 ) , our rms redshift errors , determined from repeated observations , are @xmath96 km s@xmath17 , minuscule compared to void scales .
target galaxies are selected using @xmath136 imaging from the cfht telescope down to a limiting magnitude of @xmath137 ( all magnitudes in this paper are in the ab system ; see coil et al .
2004 for photometric details ) . in three of the four fields we also use apparent colors to exclude objects likely to have @xmath138 .
this pre - selection greatly enhances our efficiency for targeting galaxies at high redshift @xcite . a fourth field
, the extended groth strip ( egs ) , has no redshift pre - selection , and is not used in the present analysis . for galaxies with a successfully identified redshift , absolute @xmath139-band magnitudes ( @xmath140 ) and restframe @xmath141 colors , denoted @xmath142 , have been derived @xcite . in the discussion
below , we define `` red '' and `` blue '' galaxies according to whether @xmath143 or @xmath144 ; this roughly corresponds to the saddle point of the color bimodality observed in deep2 data @xcite .
the current study uses @xmath145 redshifts with @xmath146 in three fields covering @xmath147 deg@xmath134 .
the three fields we use correspond to the deep2 pointings 21 , part of 22 , 31 , 32 , 33 , 41 , and 42 , as defined in @xcite .
an important aspect in analysis of any large - scale galaxy survey is the proper handling of selection effects that may vary as a function of galaxy property , such as color and luminosity .
deep2 is an @xmath18-band limited survey , which corresponds to restframe uv at @xmath148 , and results in fewer red galaxies being targeted at higher redshift compared to blue galaxies @xcite . due to this selection effect and the generally lower sampling density beyond @xmath0 , we limit our analysis in this study to @xmath149 .
each deep2 field is much longer in the redshift direction than on the sky ; the @xmath150 deg@xmath134 fields used for this work span @xmath151 @xmath45 mpc in transverse comoving extent , while the range @xmath152 corresponds to @xmath153 @xmath45 mpc comoving in the redshift direction .
the number of possible independent spheres contained within the survey , as indicated by @xmath154 , varies from @xmath155 for @xmath156 @xmath45 mpc spheres to @xmath157 for @xmath71 @xmath45 mpc spheres .
one might be concerned that the survey geometry would skew void statistics when probing voids with diameters comparable to the short dimension of the survey .
we have tested this effect by comparing the vpfs from a full mock galaxy simulation box at @xmath0 , constructed from the ` primary ' simulation ( see @xmath12[s : sims ] ) , to mock galaxy catalogs with a `` lightcone '' geometry similar to deep2 .
the lightcone geometry is constructed by stacking together several simulation box outputs and slicing through them diagonally , so as not to pick up the same structures at different times .
we find that , even at the largest radii tested , the vpf in the mock lightcone and larger simulation box agree to within @xmath158 ( fig .
[ fig : mock_ccm ] ) . of course , due to the decrease in the number of independent volume elements at large void radii , cosmic variance increases , and hence the scatter ( computed across three mock light cones ) increases .
we conclude that the geometry of the deep2 survey allows us to accurately probe voids to radii of at least 7 @xmath45 mpc comoving .
another selection effect is the survey completeness .
the deep2 survey spectroscopically targets @xmath15960% of objects that pass the apparent magnitude and color cuts mentioned above . of those targeted objects , we are able to secure redshifts for @xmath160% .
follow - up observations have shown that @xmath161% of the targets are objects at @xmath162 ( c. steidel , private communication ) .
we therefore have successful redshifts for @xmath163% of all galaxies at @xmath164 in the surveyed fields with apparent magnitude of @xmath165 .
this sampling rate is effectively scaled out when we dilute the mock galaxy catalogs to compare to the deep2 data .
further complications arise because the sampling rate is non - uniform due to the necessities of slitmask design .
spectra of objects are not allowed to overlap on the ccd , such that objects that lie near each other in the direction on the sky that maps to the wavelength direction on the ccd can not be simultaneously observed ; this results in modest ( @xmath166% ) under - sampling in regions with the highest density of targets on the plane of the sky ( each point on the sky is covered by multiple deep2 slitmasks to ameliorate this effect ) .
we model this effect by applying the actual deep2 maskmaking algorithm to the mock galaxy catalogs and then computing the vpf .
we find that the impact of the target selection algorithm on the vpf is negligible to within @xmath158 ( see fig .
[ fig : mock_ccm ] ) .
one would have expected this , as the effect is most relevant on small scales and in overdense regions .
there are also selection effects due to bright stars and other regions that were not observed .
these result in inhomogeneities on much larger scales ( @xmath167 mpc ) that could potentially be more relevant for void studies . to take account of these effects for the deep2 sample ,
we generate an angular window function that has , for each right ascension and declination , the completeness at that point determined from the bad - pixel masks used in making our photometric catalogs and the observed redshift success for slitmasks covering that point .
this window function therefore masks unobservable regions such as areas around bright stars ( they are given a completeness of @xmath168 ) .
we then convolve this window function with a circular kernel proportional to the path length through a sphere of radius @xmath18 projected on the sky ( i.e. , @xmath169 , where @xmath170 is the projected radius of the sphere in arcseconds , and @xmath171 and @xmath172 are the separations on the sky from the center of the kernel in the right - ascension and declination directions , respectively ) .
this convolution has a simple physical consequence : we treat a test sphere completely inside the survey volume at a region of @xmath101% completeness as equivalent to a sphere only @xmath101% in the survey volume at a region of @xmath173% completeness . our final product is a window function that contains , at each point , the projected completeness averaged over a sphere of a given radius , weighting by the path length through the sphere .
we then throw down random spheres only at points above a minimum convolved completeness .
this allows us to robustly avoid regions of bright stars , regions of low completeness ( due , for example , to bad weather during observations ) and the edges of the survey . for deep2
, we set the completeness threshold at @xmath174% so that the allowable completeness range varies over the survey by @xmath166% .
it is more important for our purposes to be _ uniformly _ complete than highly complete , as an overall dilution of the sample does not effect the reduced vpf .
we test the accuracy of this method by dividing the measured average number of galaxies in a sphere of radius @xmath18 , @xmath20 , by the sphere volume for each radius .
we find that this inferred number density is constant with sphere radius , providing good evidence that the larger spheres do not tend to lay farther outside the survey geometry than smaller spheres .
finally , we have spot - checked by eye the locations of the largest voids to ensure that they fall within the survey geometry ( see fig .
[ fig : deep2_show ] ) .
we investigate the vpf and reduced vpf for deep2 galaxies in three ways .
first we consider an `` overall '' sample consisting of all galaxies with @xmath89 and @xmath164 .
then we compute the vpf as a function of galaxy color and of luminosity , using volume - limited samples .
errors for all deep2 measurements were derived from jackknife sampling using subvolumes of the three widely separated deep2 fields .
we find excellent agreement between the overall deep2 sample vpf and the mock galaxy catalog vpf ( fig .
[ fig : deep2_overall+colors ] , top panel ) .
the mock catalogs were randomly diluted to have the same number density ( @xmath175 @xmath87mpc@xmath88 ) as the deep2 overall sample .
the dashed line is a prediction of the negative binomial model .
we show the reduced vpf , plotted as differences from the negative binomial prediction , for the overall sample in the top panel of fig .
[ fig : deep2_nb ] , top panel .
the good agreement between the data and the model implies that the vpf can be described entirely by @xmath20 and @xmath3 .
the largest deviations on large scales seen here and below are of low statistical significance ; the data are highly covariant from bin to bin , and hence our errors are underestimates of the true error .
the excellent agreement between the overall sample and mock galaxy vpf reflects the fact that the mock galaxy catalogs were constructed to match the observed @xmath3 in the deep2 data .
stated differently , since these samples have the same number density ( by construction ) , the fact that they have similar vpfs implies that they will have similar @xmath3 values , and vice versa .
we next divide the deep2 sample into galaxies with @xmath144 or @xmath176 , roughly matching the observed saddle point in the color bi - modality .
we find that samples of red galaxies have more and/or larger voids than samples of blue galaxies ( fig .
[ fig : deep2_overall+colors ] , bottom panel ; both samples were randomly diluted to @xmath177 @xmath87mpc@xmath88 ) .
this dilution is critical ; without it , the differences between the red and blue galaxy vpf would be dominated by differences in the observed number density of red versus blue galaxies .
it would be very difficult to separate the effects of number density from clustering strength on the number and size of voids if we simply compared the vpfs of undiluted samples . as with the overall sample , we find that the reduced vpf of both the blue and red galaxy populations follows the negative binomial model within the errors ( fig . [
fig : deep2_nb ] , top panel ) .
again , this implies that the differences between the vpfs for blue and red galaxies ( fig .
[ fig : deep2_overall+colors ] , bottom panel ) are solely due to differences in their two - point correlation functions .
the similarity in reduced vpfs is somewhat surprising .
we know that blue and red galaxies are biased differently relative to the dark matter ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) , and that the @xmath28 values are dependent on the bias .
we might have expected the reduced vpfs , which according to eqn .
[ eqn : chi ] should depend on the @xmath28 values , to be different for these two populations .
the halo model affords a more direct interpretation of these results .
we have seen in @xmath12 [ s : constrain ] that the reduced vpf is insensitive to parameters of the hod . from studies of galaxy correlations at @xmath1
, it is becoming apparent that different populations of galaxies can have very different hods .
hence the result that red and blue galaxies have similar reduced vpfs is consistent with our tests using mock catalogs ; very different hods will still produce the same reduced vpf .
we also investigate the deep2 vpf dividing the sample into three luminosity bins : @xmath178 , @xmath179 , and @xmath180 .
we take a slightly different approach for this analysis .
since the number densities of the three luminosity subsamples vary by roughly a factor of four , the dilution required for a direct comparison of vpfs can discard a great deal of information and unnecessarily increases poisson errors .
for this reason , we begin by analyzing the reduced vpf for each luminosity sample ( fig . [
fig : deep2_nb ] , bottom panel ) , again finding good agreement with the negative binomial model . using this model , we can then use eqn . [
eqn : nbvpf ] to predict the vpf at any radius based solely on our knowledge of @xmath20 and @xmath3 at that radius . in fig .
[ fig : deep2_mag ] we plot the vpf for each of the three magnitude bins ( without diluting the number density ) and also show the predictions based on the negative binomial model ; again , the agreement is within the errors .
knowledge of the form of the reduced vpf allows us to separate the effects of number density from clustering on the vpf .
we note that the three magnitude bins cover slightly different redshift intervals . while the faintest bin ( @xmath181 ) spans @xmath182 , the brightest bin ( @xmath183 ) spans @xmath184 .
each redshift interval is chosen such that the sample being considered should have a roughly constant number density as a function of redshift .
this ensures that there is no artificial redshift dependence in the void distribution ; i.e. , the samples are volume limited .
the less luminous galaxies have a constant number density only over a limited redshift range because they have @xmath185 at higher redshifts .
the reduced vpf of all populations of galaxies at @xmath0 is well fit by the negative binomial model .
hence the vpf , a statistic that in principle relies on the infinite hierarchy of correlation functions , can in fact be accurately described solely by the number density and the volume - averaged two - point correlation function of the sample .
we conclude that , at @xmath0 , the vpf provides no constraints on either the halo model or on cosmological parameters that can not be gleaned from studies of correlation statistics .
the sloan digital sky survey ( sdss ; * ? ? ? * ; * ? ? ?
* ) is an extensive photometric and spectroscopic survey of the local universe .
imaging data exist over @xmath186 deg@xmath134 in five bandpasses , @xmath187,@xmath188,@xmath52,@xmath67 , and @xmath189 .
we use color conversions provided by m. blanton ( private communication ) to derive @xmath139-band magnitudes in the ab system , with typical 1-@xmath190 errors in the conversion of 0.2 magnitudes . approximately @xmath191 objects are being targeted for follow - up spectroscopy as part of the sdss ; most spectroscopic targets are brighter than @xmath192 @xcite .
automated software performs all the necessary data reduction , including the assignment of redshifts .
redshift errors are @xmath96 kms@xmath17 , similar to deep2 .
the spectrograph tiling algorithm ensures nearly complete sampling @xcite , yet the survey is not @xmath173% complete due to several effects : ( 1 ) fiber collisions that do not allow objects separated by @xmath193 to be simultaneously targeted , affecting @xmath194% of targetable objects , ( 2 ) a small fraction ( @xmath195% ) of targeted galaxies fail to yield a reliable redshift , and ( 3 ) bright galactic stars block small regions of the sky . unlike the deep2 survey , ( 3 ) is not as important for sdss because a bright star will block out a much smaller comoving volume at @xmath1 compared to @xmath0 .
the first of these effects is only important on small scales , and is likely negligible for void statistics since an undersampled , intrinsically high density region will not be counted as a void . for this analysis
we make use of the hybrid nyu value added galaxy catalog ( vagc ) @xcite .
this catalog combines the sdss data release 2 with a multitude of other publicly available catalogs ( 2df , 2mass , iras pscz , first , and rc3 ) , and includes a variety of derived parameters including @xmath196-corrections @xcite and structural parameters . from the vagc
we have selected the two largest contiguous regions of the sdss .
we will call @xmath197 the region at @xmath198 , @xmath199 , which includes @xmath200 galaxies with spectroscopic @xmath11 and @xmath201 the region centered on @xmath198 , @xmath202 with @xmath203 galaxies at @xmath11 .
we divide red and blue galaxies in the sdss data at the valley visible in the rest - frame @xmath204 color distribution at @xmath205 .
in addition to the ` main ' sdss sample there is a secondary targeting algorithm designed to identify large numbers of luminous red galaxies ( lrgs ) at moderate redshifts via photometric color cuts .
these lrgs , though low in spatial number density , cover an enormous volume , and are hence ideal objects for measuring very large scale clustering in the universe ( see * ? ? ?
* ; * ? ? ?
* for a more detailed description of the sdss lrg sample ) .
the lrg sample we study here contains @xmath206 galaxies and spans the redshift range @xmath207 and luminosity range @xmath208 with @xmath209 ( note that this is the only case where we use sdss absolute magnitudes ) .
since the sdss is a low - redshift survey , the angular size of a test void sphere of the same comoving radius varies enormously over the redshift range @xmath210 . paralleling our analysis of the deep2 sample
, we account for geometry and completeness effects in the following way .
we generate an angular window function for the sdss with the aid of _ mangle _ @xcite over a dense grid in right ascension and declination ; the resulting resolution is @xmath211 degrees in right ascension and declination .
completeness values of either 0 or 1 were assigned to each right point of this grid depending on the spectroscopic coverage and locations of bright stars ( as sdss spectroscopy is highly uniform in depth ) .
we then convolve this window function with a kernel that represents the depth through the test void projected on the sky ( see @xmath12 [ sec : deep_compl ] for details ) and only place random spheres in regions above a minimum convolved completeness .
this allows us to avoid placing test spheres in poorly - sampled regions .
we set this threshold at @xmath212% , so that the completeness within spheres placed down varies over the survey by @xmath166% .
the net result of the strong scaling of angular size with redshift is that larger - volume test spheres can not be placed in the lowest redshift bins because they would span a region of the sky comparable to the entire survey .
hence at larger void radii we are restricted to higher redshifts .
this does not cause a detrimental bias , however , because we only consider galaxies over a redshift range such that their number density is approximately constant , i.e. , volume - limited samples .
hence a bias towards slightly higher redshifts ( @xmath213 ) for larger voids will not skew our statistics , assuming that there is not strong void statistic evolution from @xmath1 to @xmath213 .
indeed no significant evolution has been detected in the vpf out to @xmath214 @xcite .
there is one final effect in the sdss data that , if not properly treated , could significantly bias our results .
there is a massive structure in @xmath201 at @xmath215 dubbed the `` sloan great wall '' @xcite , which will strongly affect any clustering measurements .
this structure has been removed in the correlation function studies of @xcite , and we do the same here by defining our samples to avoid @xmath216 in the @xmath201 sample .
the structure extends across the entire angular extent of @xmath201 ; if included it would cause gross underestimates of the cosmic error in any large - scale structure measurement .
based on many realizations of large cosmological simulations with gaussian initial conditions , it is expected that a structure of this size occurs in a volume the size of the sdss approximately @xmath217% of the time @xcite . with the full sdss dataset
, it should be possible to accurately account for this enormous structure in the error budget without relying on simulations . in order to accurately measure quantities via our counts - in - cells approach ,
it is important that the number of randomly placed test spheres exhausts the number of independent volumes in the survey .
the second data release of the sdss over the interval @xmath218 spans a volume of @xmath219 @xmath98 mpc@xmath99 .
this corresponds to @xmath220 independent volumes for @xmath156 @xmath45 mpc spheres and @xmath157 for @xmath221 @xmath45 mpc spheres , a factor of @xmath217 more than in the deep2 survey .
the number of spheres we use always exceeds the number of independent volumes .
we now present void statistics in the local universe using the sdss dataset . as in our analysis of deep2 data ,
we investigate void statistics for three sets of sdss samples : an `` overall '' sample with @xmath89 and @xmath222 , two subsamples split according to @xmath204 color ( at @xmath205 ) , and three subsamples differing in luminosity . in this section , as before , we plot reduced vpfs as differences from the negative binomial model in order to clearly show small deviations . as for the deep2 luminosity subsamples , all comparisons made in this section
are made without requiring that samples have similar number densities , and by using the negative binomial model to predict the vpf ( via eqn .
[ eqn : nbvpf ] ) .
errors for all quantities were derived from jackknife subsamples .
the reduced vpfs for the overall , red , and blue samples ( fig .
[ fig : sdss_rvpf ] , top panel ) are all well described by the negative binomial model . as in the previous sections
, we can use this measured agreement to predict the vpf at any radius based solely on @xmath20 and @xmath3 for that radius , ( e.g. as estimated from counts - in - cells ) , for each sample .
[ fig : sdss_vpf ] ( top panel ) plots the vpf for these samples and compares them to the predictions from the negative binomial model ( lines ) .
the trend that the red galaxy sample has more and/or larger voids than the blue sample is due to the fact that it is both more strongly clustered and lower in number density than our sample of blue galaxies . that the measured data agree with the negative binomial model within errors implies that , just as at @xmath0 , the vpf can not be used to uniquely constrain the halo model or cosmological parameters . as with the overall deep2 vpf
, the blue sdss data do not agree within @xmath158 errors on the largest scales ; this is likely insignificant , as the data are highly covariant , and hence our errors are underestimates of the true error .
next , we compare void statistics for samples of sdss galaxies as a function of luminosity . the reduced vpf for galaxies in all luminosity bins we consider ( @xmath183 , @xmath223 , and @xmath181 ) , is in good agreement with the negative binomial model ( fig . [ fig : sdss_rvpf ] , bottom panel ) .
we again make this agreement explicit by computing the vpfs for these samples and comparing to that predicted by assuming the negative binomial model from the values of @xmath20 and @xmath3 ( fig .
[ fig : sdss_vpf ] , bottom panel ) .
since these samples are volume limited , each luminosity sample spans a slightly different redshift range , with the brightest spanning the largest range ( @xmath224 ) ( since bright galaxies can be detected at greatest distances ) and the faintest spanning the smallest range ( @xmath225 ) .
finally , we compute the vpf for the lrg population , defined to be bright ( @xmath208 ) and red ( @xmath209 ) , spanning the redshift range @xmath207 . their low number density ( @xmath226 @xmath87 mpc@xmath88 ) and large redshift range allow for a measurement of the vpf out to unprecedented scales ( r@xmath227 @xmath45 mpc ) .
[ fig : sdss_lrg ] shows that the lrg population is well described by the negative binomial model out to the largest void radii tested .
unlike our analysis of other sdss data , we plot the vpf separately for the two regions used in this study ( denoted @xmath197 and @xmath201 ) .
the difference between the two regions reflects the impact of cosmic variance on void statistics .
the reduced vpfs of all galaxy populations explored at @xmath1 are in good agreement with the negative binomial model .
it is quite surprising that this agreement , which provides a simple mapping from @xmath20 and @xmath3 to @xmath24 , is equally valid from @xmath0 to @xmath1 .
what this tells us is that the evolution in the vpf for a given population from @xmath0 to @xmath1 must proceed in lockstep with evolution in the number density and two - point correlation function of that population .
finally , it is encouraging that these conclusions at @xmath1 agree with @xcite , who measure void statistics for the low - redshift 2df survey , and find good agreement between the galaxy reduced vpf and the negative binomial model for a range of galaxy luminosities .
we have presented measurements of the void probability function ( vpf ) and reduced vpf for galaxies as a function of color and luminosity at @xmath0 using deep2 data and at @xmath1 with sdss data .
we find that all samples are well described by the negative binomial model .
this agreement implies that the vpf for a given sample is determined entirely by the sample number density and volume - averaged two - point correlation function , @xmath3 .
in particular , evolution of the vpf for a population of galaxies from @xmath0 to @xmath1 is governed by the evolution in @xmath3 and @xmath20 for that population .
we have furthermore shown , using mock catalogs , that this simple relation between the vpf and @xmath3 holds for a wide range of halo models in redshift space , but breaks down in real space .
we now discuss the relevance and implications of these findings .
the vpf we measure at @xmath0 is in good agreement with the vpf measured in mock galaxy catalogs created from @xmath5cdm simulations .
although this is encouraging , our result that the two - point correlation function and number density determine the vpf implies that this agreement found between data and simulations is simply a consequence of the fact that the simulations were constructed to match the number density and two - point correlation function of deep2 galaxies
. stated differently , the vpf currently can not provide new constraints either on cosmological parameters or on the method in which galaxies are placed into dark matter n - body simulations .
previous void studies have interpreted the vpf according to its mathematical expansion as an infinite sum of higher - order correlation functions , using the hierarchical _ ansatz _ to relate the higher - order functions to lower - order functions .
these studies then attempt to use the reduced vpf to test the validity of this _
ansatz_. strictly speaking , however , finding that the vpf for various populations of galaxies can be characterized entirely by @xmath20 and @xmath3 does _ not _ provide evidence for the validity of this _
ansatz_. in fact , the interpretation of the vpf in terms of higher - order correlation functions implies that the reduced vpf depends only on the scaling coefficients ( @xmath228 ) , which suggests that populations with different biases ( and therefore different scalings between the three - point and two - point functions ) should actually not have the same reduced vpf .
we find just the opposite in the data , namely that populations with different biases have the same reduced vpf .
the interpretation of the vpf as a sum of correlation functions affords little insight into these results . despite this , the interpretation predicts that the reduced vpf of a sample should be independent of that samples number density , and this is indeed observed .
the halo model provides a more tractable theoretical framework .
we have shown that the reduced vpf of mock galaxies is quite insensitive to particular parameters of the hod when voids are measured in redshift space . using measurements of @xmath2 to constrain the halo model
, it appears that blue and red galaxies have very different hods @xcite .
the samples we investigate here hence likely have a range of hods .
yet , based on the result that different hods generate the same reduced vpf , it is not surprising that different samples have the same reduced vpf .
furthermore , an analysis of the dark matter _ halo _ reduced vpf has led us to conclude that it is the redshift - space distribution of halos themselves that is primarily responsible for the agreement between the measured reduced vpf and the negative binomial model .
the reduced vpf changes very little when one populates dark matter halos with galaxies .
it is only surprising that the minimum halo mass , @xmath8 , has very little effect on the reduced vpf as well .
this insensitivity to @xmath8 implies that the changes in @xmath3 caused by changes in @xmath8 are enough to completely account for the changes in the vpf .
we would like to stress that the strict dependence of the vpf on @xmath33 does not rely on any theoretical interpretation , including the hierarchical _ ansatz _ , and , in particular , does not depend on the reduced vpf following the negative binomial model .
our result that all galaxy populations studied here have consistent reduced vpfs immediately implies that the vpf can be entirely described by @xmath33 .
this , in turn , implies that the vpf is currently incapable of uniquely providing constraints on either cosmological parameters or particular aspects of the halo model , as any useful information provided by the vpf is likewise provided by two - point correlation function analyses ; the vpf is of little use in understanding the large scale structure of the universe .
we would like to thank andreas berlind , michael blanton , darren croton , ofer lahav , chung - pei ma , roman scoccimarro , and simon white for useful discussions .
this project was supported in part by the nsf grants ast00 - 71198 and ast00 - 71048 .
the deimos spectrograph was funded by a grant from cara ( keck observatory ) , an nsf facilities and infrastructure grant ( ast92 - 2540 ) , the center for particle astrophysics , and gifts from sun microsystems and the quantum corporation .
c.c . acknowledges support from an nsf reu grant .
j.n . acknowledges support from nasa through hubble fellowship grant hst - hf-01165.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 .
some of the data presented herein were obtained at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california , and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w.m . keck foundation .
in addition , we wish to acknowledge the significant cultural role that the summit of mauna kea plays within the indigenous hawaiian community ; we are fortunate to have the opportunity to conduct observations from this mountain .
funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the korean scientist group , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , the university of pittsburgh , the university of portsmouth , princeton university , the united states naval observatory , and the university of washington . | we present measurements of the void probability function ( vpf ) at @xmath0 using data from the deep2 redshift survey and its evolution to @xmath1 using data from the sloan digital sky survey ( sdss ) .
we measure the vpf as a function of galaxy color and luminosity in both surveys and find that it mimics trends displayed in the two - point correlation function , @xmath2 ; namely that samples of brighter , red galaxies have larger voids ( i.e. are more strongly clustered ) than fainter , blue galaxies .
we also clearly detect evolution in the vpf with cosmic time , with voids being larger in comoving units at @xmath1 .
we find that the reduced vpf matches the predictions of a ` negative binomial ' model for galaxies of all colors , luminosities , and redshifts studied .
this model lacks a physical motivation , but produces a simple analytic prediction for sources of any number density and integrated two - point correlation function , @xmath3 .
this implies that differences in the vpf across different galaxy populations are consistent with being due entirely to differences in the population number density and @xmath3 .
we compare the vpf at @xmath0 to @xmath4-body @xmath5cdm simulations and find good agreement between the deep2 data and mock galaxy catalogs .
interestingly , we find that the dark matter _ particle _ reduced vpf follows the physically motivated ` thermodynamic ' model , while the dark matter _ halo _ reduced vpf more closely follows the negative binomial model . the robust result that all galaxy populations follow the negative binomial model appears to be due to primarily to the clustering of dark matter halos .
the reduced vpf is insensitive to changes in the parameters of the halo occupation distribution , in the sense that halo models with the same @xmath3 will produce the same vpf .
for the wide range of galaxies studied , the vpf therefore does not appear to provide useful constraints on galaxy evolution models that can not be gleaned from studies of @xmath3 alone . |
deep inelastic lepton - nucleon scattering experiments have provided precision information about the quark distributions in the nucleon .
however , there has been a long - standing discrepancy between the @xmath1 structure functions extracted from neutrino and muon experiments in the small @xmath2 range .
recently , a new analysis of differential cross sections and structure functions from ccfr @xmath3-fe and @xmath4-fe data was presented ; in this study , the neutrino - muon difference is resolved by extracting the @xmath3 structure functions in a physics model independent way.@xcite in previous analyses of @xmath3 data,@xcite structure functions were extracted by applying a slow rescaling correction to correct for the charm mass suppression in the final state .
in addition , the @xmath0 term ( used as input in the extraction ) was calculated from a leading order charm production model .
these resulted in physics model dependent ( pmd ) structure functions . in the new analysis,@xcite slow rescaling corrections are not applied , and @xmath0 and @xmath1 were extracted from two parameter fits to the data .
the extracted physics model independent ( pmi ) values for @xmath5 are then compared with @xmath6 within the framework of nlo models for massive charm production ; these are found to be in agreement , thus resolving the long - standing discrepancy between the two sets of data . however , the first measurements of @xmath0 are higher than current theoretical predictions .
the objective of this paper is to investigate the sensitivity of @xmath0 upon a variety of factors including renormalization scheme and scale , quark mass effects , higher twist , isospin violation , and pdf uncertainties . in
we have plotted the quantity @xmath0 for an isoscalar target computed to order @xmath7 . for comparison
, we also display data from the ccfr analysis.@xcite while there is much freedom in the theoretical calculation , the difference between these calculations and the data at low @xmath8 values warrants further investigation .
we will discuss and compare the different theoretical calculations , and examine the inherent uncertainty in each with respect to different input parameters .
we will also examine the experimental input , and assess uncertainties in this data .
the structure functions are defined in terms of the neutrino - nucleon cross section via : @xmath9 \nonumber
\\ \label{eq:}\end{aligned}\ ] ] where @xmath10 is the fermi weak coupling constant , @xmath11 is the nucleon mass , @xmath12 is the incident energy , @xmath13 is the fractional energy transfer , and @xmath14 is the final state hadronic energy .
the sum of @xmath3 and @xmath4 differential cross sections for charged current interactions on an isoscalar target is then : @xmath15 \frac{(1-\epsilon)\pi}{y^2g_f^2me_\nu } \nonumber\\ & = & 2xf_1 [ 1+\epsilon r ] + \frac{y(1-y/2)}{1+(1-y)^2 } \delta xf_3 \label{eq : feps}\end{aligned}\ ] ] where @xmath16 $ ] is the polarization of the virtual @xmath17 boson . in the above equation ,
we have used the relation : @xmath18 where @xmath19 is the ratio of the cross - sections of longitudinally- to transversely - polarized @xmath17-bosons , @xmath20 is the square of the four - momentum transfer to the nucleon , and @xmath21 is the bjorken scaling variable . for @xmath22 , @xmath23 in neutrino scattering
is expected to be somewhat larger than @xmath23 for muon scattering because of the production of massive charm quarks in the final state for the charged current neutrino production . using and ,
@xmath1 and @xmath0 can be extracted separately . because of the positive correlation between @xmath23 and @xmath0 , the extracted values of @xmath1 are rather insensitive to the input @xmath23 .
if a large input @xmath23 is used , a larger value of @xmath24 is extracted from the @xmath25 distribution , thus yielding the same value of @xmath1 .
in contrast , the extracted values of @xmath0 are sensitive to the assumed value of @xmath23 , which yields a larger systematic error , shown on the data . for @xmath23
the qcd - inspired fit in @xcite is used , but corrected for charged current neutrino scattering using a leading order slow rescaling model .
this gives precisely the same type of correction as a full nlo calculation including the massive charm quark , as shown in ref .
@xcite , but leads to a somewhat higher normalization than the perturbative correction .
it is arguable which prescription for @xmath23 leads to the better fit to existing data , but the difference between the two leads to an uncertainty in @xmath0 of the order of the systematic error shown . clearly a further reduction in the assumed value of r ( even down to zero ) , as suggested by models including @xmath26 resummations , would still leave a discrepancy between the lowest @xmath20 data points and the theoretical predictions .
now that we have outlined the experimental method used in the extraction of the @xmath27 structure functions , it is instructive to recall the simple leading - order correspondence between the @xmath27 s and the pdf s : @xmath28 therefore , the combination @xmath0 yields : @xmath29 note that since this quantity involves the parity violating structure function @xmath30 , this measurement has no analogue in the neutral current photon - exchange process .
also note that since , at leading - order , @xmath0 is directly sensitive to the strange and charm distributions , this observable can be used to probe the heavy quark pdf s , and to understand heavy quark ( charm ) production .
we discuss these possibilities further in the following subsection .
we have illustrated in how @xmath0 is closely tied to the heavy quark pdf s .
the questions is : given the present knowledge base , should we use @xmath0 to determine the heavy quark pdf s , or vice verse . to answer this question
, we briefly review present measurements of heavy quark pdf s , and assess their uncertainty .
the precise measurement of @xmath17 plus heavy quark @xmath32 events provides important information on heavy quark pdf s ; additionally , such signals are a background for higgs and squark searches.@xcite unfortunately , a primary uncertainty for @xmath31 production comes from the heavy quark pdf s . given that @xmath0 is sensitive to these heavy quark pdf s , we see at least two scenarios .
one possibility is that new analysis of present data will resolve this situation prior to run ii , and provide precise distributions as an input to the tevatron data analysis .
if the situation remains unresolved , then new data from run ii may help to finally solve this puzzle . in the far future , a neutrino experiment from the neutrino factory at a linear collider would be an ideal tool to measure any neutrino structure function.@xcite the strange distribution is directly measured by dimuon production in neutrino - nucleon scattering .
the basic sub - process is @xmath33 with a subsequent charm decay @xmath34 .
dis dimuon measurements have safely established the breaking of the @xmath35 flavor symmetry , @xmath36/2 } \right|_{q^2={{\cal{o}}(10{\rm gev^2 } ) } } \lesssim \frac{1}{2 } \label{eq : kappa}\ ] ] in the nucleon sea .
still , there remain large uncertainties for the @xmath37-quark distribution in the kinematic regions relevant for @xmath0 , even though the @xmath38-range of the ccfr dimuon measurements @xcite is comparable .
ccfr recorded 5044 @xmath3 and 1062 @xmath4 @xmath39 events with @xmath40 , @xmath41 , @xmath42 , and @xmath43 .
the more recent nutev experiment recorded a similar sample of events , and these are presently being analysed through a mc simulation based on nlo quark- and gluon - initiated corrections at differential level@xcite .
a complete nlo analysis of this data together with a global analysis may help to further constraint the strange quark distribution . at present ,
pdf sets take strangeness suppresssion into account by imposing the constraint in as @xmath44/2 $ ] at the pdf input scale @xmath45 , @xcite or by evolving @xmath46 from a vanishing input at a lower scale.@xcite the residual uncertainty can be large , as can be seen from the collection of strange seas in . in the following , we will take the uncertainty in the strange pdf into account by relaxing the experimental constraint in as implemented in the mrst partons.@xcite this will be discussed in .
the strange distribution can also be extracted indirectly using a combination of charged ( @xmath5 ) and neutral ( @xmath6 ) current structure functions ; however , the systematic uncertainties involved in this procedure make an accurate determination difficult.@xcite the basic idea is to use the ( leading - order ) relation @xmath47 to extract the strange distribution . here
, @xmath48 represents a sum over all quark flavors .
this method is complicated by a number of issues including the @xmath24 component which can play a crucial role in the small-@xmath2 region
precisely the region where we observe the discrepancy . from the corresponding relation : @xmath49 we see that these problems are not independent ; however , this information , together with the exclusive dimuon events , may provide a more precise determination of the strange quark sea , and help to resolve our puzzle . prior to the dis dimuon data , the 1992 cteq1 analysis @xcite found that a combination of nc structure functions from nmc@xcite and the physics - model - dependent charged current structure functions from ccfr @xcite seem compatible with approximate @xmath35-symmetry , _ i.e. _ , @xmath50 in .
recent dimuon measurements now exclude an @xmath35-symmetry @xmath51 .
we shall explore the effect of @xmath52 on @xmath53 in .
we now systematically investigate the sensitivity of the theoretical predictions of @xmath0 upon a variety of factors including : renormalization scheme and scale , quark mass effects , higher twist , isospin violation , and pdf uncertainties . to simplify this analysis
, we first examine the influence of these factors on the lo expression : @xmath54 ; after using this as a toy model , " we will then return to the full nlo calculation in the next section .
asymmetry @xcite would average out in @xmath0 . ] for most variables , the simplified lo is sufficient to display the general behavior of the full nlo result .
there are two exceptions : 1 ) the scheme dependence , and 2 ) the pdf dependence .
these factors depend on the interplay of both the quark - initiated lo contributions as well as the nlo gluon - initiated contributions . for this reason
, we will postpone discussion of these effects until the following section .
we start by examining the effect of the charm mass , @xmath55 , on @xmath0 . in
, we plot the leading - order expression @xmath54 _ vs. _ @xmath20 for values of @xmath55 in the range @xmath56 $ ] gev using steps of 0.2 gev . here
, we define @xmath57 which is a slow - rescaling " type of correction@xcite which ( crudely ) includes mass effects by shifting the @xmath2 variable .
`` slow rescaling '' naturally arises at lo from the single - charm threshold condition(@xmath58 ) and is required at nlo for consistent mass factorization @xcite .
note , the result of this correction is most significant at low @xmath20 . to isolate the kinematic @xmath55 dependence ,
we have chosen the scale @xmath59 .
( we will separately vary the scale in the following subsection . )
note that this exercise is only altering the charm mass in one aspect of the calculation ; to be entirely consistent it would be necessary to obtain parton distributions ( particularly the charm quark ) using fits with different charm masses . for charm masses in the range
1.2 to 1.8 gev , this will be a small effect ; for charm masses below @xmath60 gev , ( which is below the experimentally allowed range@xcite ) , such issues become important and the curves of will be modified , i.e. lower @xmath55 will lead to longer evolution for charm , a larger charm distribution , and a lowering of the curves in .
while taking @xmath61 does raise the theoretical curves in the regions where we observe the discrepancy ( namely , the low @xmath8 region ) , varying @xmath55 even within the wide range of @xmath62 $ ] gev ( lower 4 curves ) does not give us sufficient flexibility to match either the shape or normalization of the data .
next , we investigate the variation of @xmath54 with the renormalization / factorization scale @xmath63 .
we use three choices of the @xmath64 scale : * @xmath65 , * @xmath66 , * @xmath67 . of course , @xmath0 with @xmath64 scales of @xmath20 and @xmath68 differ only at lower values of @xmath20 ; @xmath0 with @xmath64 scale of @xmath69 is comparable to @xmath20 and @xmath68 at larger @xmath2 , but lies below for smaller @xmath2 .
the scale choice @xmath68 leads to an improvment over @xmath20 by providing a lower bound on @xmath63 to keep the scale in the perturbative region .
the choice of @xmath70 is motivated , in part , by some observations by collins.@xcite to display the relationship between @xmath70 and @xmath8 , in we plot both @xmath70 and @xmath8 _ vs. _
@xmath2 for 4 choices of @xmath25 ; note@xcite that the @xmath2-dependence of @xmath70 is opposite that of @xmath8 .
while tying the scale choice of @xmath63 to @xmath70 has some interesting intuitive interpretations , of the partonic subprocess , but this is unobservable ; therefore @xmath70 is used instead.@xcite ] for small @xmath2 this scale clearly becomes too large for the relevant physics . in any case , it can not help us with our @xmath71 problem as the scale choice of @xmath70 moves the theory curves _
away _ from the data .
we now illustrate the potential effects due to higher twist contributions .
we parameterize such contributions by multiplying the leading - twist terms by a correction factor that will increase at low @xmath20 .
more specifically we use:@xcite @xmath72 where @xmath73 .
we vary @xmath74 over the range [ 0,10 ] @xmath75 in steps of @xmath76 .
( for the purposes of our simple illustration , it is sufficient to take @xmath74 to be independent of @xmath2 . )
we will find that this range is well beyond what is allowed by experiment ; however , we display this exaggerated range to make the effect of the higher twist contributions evident . to normalize this choice with the allowable range consistent with data ,
we compare with the mrst higher twist fit which extracted a limit on the function @xmath77 .
we see from the table of @xmath77 that the allowed contribution from the higher twist terms is quite small .
in addition , we note that the sign for @xmath77 obtained for the relevant small-@xmath2 region tends to be negative exactly the opposite sign that is needed to move the theory toward the data .
as we have shown constraints on @xmath77 , the obvious question is should we expect @xmath74 to be substantially different ?
a calculation of the power corrections using renormalons@xcite suggests that while both @xmath78 and @xmath74 are of the same order of magnitude as @xmath77 and have similar @xmath2-dependence , that @xmath78 and @xmath74 are even more negative than @xmath77 ; again , this trend would move the theory farther from the data .
therefore , we conclude that that @xmath77 represents a conservative limit for the @xmath74 power corrections . examining
, we note that it would take an enormous higher twist contribution to bring the normalization of the curves in the range of the data points , and still the shape of the @xmath20-dependence is not well matched ; hence , we conclude that this is not a compelling solution .
the naive parton model identity in is modified if the full ( non - diagonal ) ckm structure , nlo qcd radiative corrections , and qcd based charm production are taken into account .
this expression is also modified _ even in leading - order _ if we have a violation of exact @xmath79 isospin - symmetry , ( or charge symmetry ) ; _ e.g. _ , @xmath80 . in deriving ,
isospin - symmetry was necessary to guarantee that the @xmath81-contributions cancel out in the difference , thereby leaving only the @xmath46 and @xmath82 contributions .
the validity of exact _ charge symmetry _ ( cs ) has recently been reexamined.@xcite residual @xmath81-contributions to @xmath0 from _ charge symmetry violation _ ( csv ) would be amplified due to enhanced valence components @xmath83 , and because the @xmath84 transitions are not subject to slow - rescaling corrections which strongly suppress the @xmath85 contribution to @xmath0 , as illustrated in .
we will examine possible contributions to @xmath0 by considering a `` toy '' model to parameterize the cs - violations .
in isospin space , we can parameterize a general transformation as a rotation : @xmath86 where @xmath23 is a rotation matrix , and @xmath87 is the normalization factor .
for example , in this model the @xmath88-distribution in the neutron would be related to the proton distributions via the relation : @xmath89\end{aligned}\ ] ] for @xmath90 , we recover the symmetric limit @xmath91 . note that we define the normalization @xmath92 such that this model preserves the sum rule : @xmath93 note that should not be considered a serious theory , but rather a simple one - parameter ( @xmath94 ) model which is flexible enough to illustrate a range of csv effects for @xmath0 . in
we vary @xmath94 over the its maximum range @xmath95 $ ] in steps of @xmath96 . the exact charge symmetry limit ( @xmath97 ) of corresponds to the lowest curve in .
note the effect of the csv contribution monotonically increases as @xmath94 deviates from the charge symmetry ( cs ) limit @xmath98 . from the plot we observe that a violation of @xmath99 is required to bring the theory in the neighborhood of the data .
at the relevant @xmath2 values of interest for @xmath0 , this translates ( via and its analogue for @xmath100 ) into a @xmath101 symmetry violation for @xmath102 , and a @xmath103 symmetry violation for @xmath100 .
specifically , @xmath104 since the electroweak couplings are flavor - independent , ( @xmath105 ) , @xmath0 is in principle insensitive to a re - shuffling of the csv contributions between @xmath102 and @xmath100 .
in particular , if we define the shift due to csv in @xmath0 as @xmath106 , we find : @xmath107 where @xmath108 the expression for @xmath109 of is evaluated at @xmath110 from with @xmath111 and is plotted in .-shape from should _ not _ be taken as a serious model prediction . ]
although this level of isospin violation certainly improves the description of @xmath0 , it is necessary to consider precisely what level of violation is actually allowed by other experimental data .
for instance , it has previously been suggested that the discrepancy between @xmath1 from neutrino and muon data itself may be due to isospin violation.@xcite this type of violation required that @xmath112 be very different to @xmath113 in the region of interest .
the measurement of the lepton charge asymmetry in w decays from the tevatron @xcite places tight constraints on the up and down quark distributions in the range @xmath114 , constraining them to be approximately as specified in the parton sets obtained by the global analyses . while only strictly telling us about parton distributions in the proton , this data rules out an isospin violation of this type to about @xmath115 as demonstrated in @xcite .
however , there are other strong constraints on isospin violation . for example
, we note that while the toy model above leaves the neutron singlet combination @xmath116 invariant at the @xmath117 level in the region @xmath118 $ ] , it would lower the nc observable : @xmath119_n \ \right|_{x\epsilon [ 0.01;0.1]}\ ] ] by about 10% .
an effect of this size would definitely be visible in nmc @xmath120 data which has an uncertainty of order a few percent in this kinematic region , and acts as a major constraint.@xcite at this point , one could play clever games to evade the constraints of specific experiments .
for example , a re - shuffling of csv contributions between the individual @xmath121 in according to : @xmath122 would keep invariant .
however , this would in turn raise the cc observable by @xmath123 ; though it would help to explain the excess in @xmath124 , it would spoil the new - found compatibility between neutral current and charged current data .
in addition , there are also fixed - target drell - yan experiments@xcite such as na51 and e866 which precisely measure @xmath125 in the range@xcite @xmath126 , and are also sensitive to isospin violating effects .
we therefore conclude that the many precise data sets which constrain different combinations of the pdf s probably leave no room for csv contributions of the magnitude necessary to fully align the theory curves with the @xmath0 data .
however , we should add the caveat that an exhaustive investigation of the interplay of these different data sets and their influence on @xmath0 will only be possible within a global pdf analysis
now that we have used the leading - order expression for @xmath0 to systematically investigate dependence of this observable on various parameters , in this section we now turn to the full nlo calculation . in , we have plotted the lo and nlo calculations for @xmath53 _ vs. _ @xmath20 on an isoscalar target in 3 and 4 flavor schemes .
the 3-flavor lo calculation ( f=3 , @xmath128 ) involves primarily the strange quark contribution , @xmath46 , as the charm distribution is excluded in this case . when the higher order terms are included ( f=3 , @xmath7 ) , this result moves ( substantially ) toward the predictions of the 4-flavor scheme .
we note that while the 3-flavor lo calculation ( f=3 , @xmath128 ) appears consistent with the data , we can not take this result as a precise theoretical prediction as this simplistic result is highly dependent on scheme and scale choices ; a result that is verified by the large shift in going from lo to nlo .
the pair of curves in the 4-flavor scheme ( using the cteq4hq distributions ) nicely illustrates how the charm distribution @xmath129 evolves as @xmath130 for increasing @xmath20 ; note , @xmath129 enters with a negative sign so that the 4-flavor result is below the 3-flavor curve . for the scale choice , we take @xmath131 .
while the scale choice @xmath132 is useful for instructive purposes such as demonstrating the matching of the 3- and 4-flavor calculations at @xmath133 , the choice @xmath131 is more practical as it provides a lower bound on @xmath63 which is important for the pdf s and @xmath134 .
_ , , and ref . @xcite . )
additionally we note the stability of the 4-flavor scheme in contrast to the 3-flavor scheme .
the shift of the curves when including the nlo contributions is quite minimal , particularly when compared with the 3-flavor result.@xcite this suggests that organizing the calculation to include the charm quark as a proton constituent can be advantageous even at relatively low values of the energy scale . in
, we show the variation of nlo calculation of @xmath135 on the strange - quark pdf . to obtain a realistic assessment of the @xmath46 dependence , we have use the nlo calculation with pdf s based on the mrst set which are re - fit with the value of @xmath136 constrained to be @xmath137 .
note , by re - fitting the pdf s with the chosen value of @xmath52 we are assured to have an internally consistent set of pdf s with appropriate matching between the quarks and gluon , and with the sum - rules satisfied .
is certainly @xmath8-dependent , and the values for @xmath52 quoted above correspond to the @xmath138 of the evolution .
while @xmath52 compares the integral of @xmath46 to the seq - quarks , there is also the possibility of an @xmath2-dependent variation .
this has been studied in the fits of the strange - sea@xcite ; we shall find that such subtle effects can play no role in resolving the @xmath0 issue . ]
the choice @xmath139 is in line with the many experimental determinations of @xmath52 , _ cf .
_ , ; as expected , this prediction lies farthest from the data points .
the choice @xmath140 is taken as an extreme upper limit given the experimental constraints ; actually , in light of the results of , this is arguably _ beyond _ present experimental bounds .
this prediction is marginally consistent at the outer reach of the systematic + statistical error bars . finally , we take an @xmath35 symmetric set ( @xmath141 ) purely for illustrative purposes .
it is interesting to note that even this extreme value is still below the central value of the data points at the higher @xmath2 values .
in conclusion we note that increasing the strange quark distribution does succeed in moving the theory toward the data ; however , our consistent nlo analysis presented here suggests that the we have only limited freedom to increase @xmath46 , and that this alone is not sufficient to obtain good agreement between theory and data . in our final section , we present the best theoretical predictions presently available to demonstrate the scheme dependence of @xmath142 .
specifically , in we show predictions for : * nlo ffs grv.@xcite * nlo vfs tr calculation.@xcite * nlo vfs acot calculation with cteq4 pdfs.@xcite * nlo vfs acot calculation with cteq5 pdfs.@xcite all these calculations use nlo matrix elements , and are matched with appropriate global pdf s which are fitted in the proper scheme .
the first observation we make is how closely these four predictions match , especially given the wide variation displayed in previous plots such as .
in hindsight , this result is simply a consequence of the fact that while different renormalization schemes can produce different results , this difference can only be higher order .
thus , the difference between these curves is indicative of terms of order @xmath143 which have yet to be calculated .
, see ref . @xcite .
] when terms of order @xmath144 are included , the span of these predictions will be systematically reduced to order @xmath145 . in
, we note the very close agreement among the vfs calculations , particularly the tr calculation and the acot calculation with cteq4 pdf s . the acot calculation with the two cteq curves
show primarily the effect of the charm distribution , as cteq4 uses @xmath146 and cteq5 uses @xmath147 .
the grv calculation shows the effect of using yet a different scheme , in this case a ffn scheme , with its appropriately matched pdf .
were we to use mrst or cteq pdf s , the spread of these theory curves would decrease ; however , this would most likely represent an underestimate of the true theoretical uncertainties arising from both the hard cross section and pdf s .
while we consider it a triumph of qcd that different schemes truly yield comparable results ( higher order terms aside ) , we should be cautious and note that the spread of these curves can only underestimate the true theoretical uncertainty .
note that grv has a rather different strange distribution due to a different philosophy of obtaining this distribution rather than due to a different scheme .
comprehensive analysis of the neutrino data sets can provide incisive tests of the theoretical methods , particularly in the low @xmath20 regime , and enable precise predictions that will facilitate new particle searches by constraining the pdf s .
this document serves as a progress report , and work on these topics will continue in the future .
theoretical predictions for @xmath0 undershoot preliminary fixed target data at the @xmath148-level at low @xmath2 and @xmath8 .
the neutrino structure function @xmath0 is obviously sensitive@xcite to the strange sea of the nucleon and the details of deep inelastic charm production .
a closer inspection reveals , however , considerable dependence upon factors such as the charm mass , factorization scale , higher twists , contributions from longitudinal @xmath149 polarization states , nuclear shadowing , charge symmetry violation , and the pdf s .
this makes @xmath150 an excellent tool to probe both perturbative and non - perturbative qcd .
we have explored the variation of @xmath150 on the above factors and found none of these to be capable of resolving the discrepancy between the data and theory .
although we have not eliminated the possibility that the entire set of parameters conspires to align the theory with the data , we have demonstrated this to be extremely unlikely .
of course , a definitive answer can only be obtained by a global analysis which combines the neutrino data for dimuons , @xmath0 , @xmath151 , and @xmath152 .
as the situation stands now , this @xmath150 puzzle poses an important challenge to our understanding of qcd and the related nuclear processes in an important kinematic region .
the resolution of this puzzle is important for future data analysis , and the solution is sure to be enlightening , and allow us to expand the applicable regime of the qcd theory .
this work is supported by the royal society , the u.s .
department of energy , the national science foundation , the lightner - sams foundation , and the ` bundesministerium fr bildung , wissenschaft , forschung und technologie ' , bonn .
t. adams _ et al . _
[ nutev collaboration ] , hep - ex/9906037 .
p. vilain _ et al . _
[ charm ii collaboration ] , eur .
j. * c11 * , 19 ( 1999 ) .
p. astier _ et al .
_ [ nomad collaboration ] , phys . lett . *
b486 * ( 1900 ) 35 . c. boros , f.m .
steffens , j.t .
londergan and a.w .
thomas , phys .
* b468 * , 161 ( 1999 ) ; c. boros , j.t .
londergan , a.w .
thomas , phys .
rev . * d59 * , 074021 ( 1999 ) ; c. boros , j.t .
londergan , a.w .
thomas , phys .
* 81 * , 4075 ( 1998 ) and related references therein .
s. alekhin , eur .
j. * c10 * , 395 ( 1999 ) ; w. t. giele and s. keller , phys .
d * 58 * , 094023 ( 1998 ) ; w. t. giele , s. keller and d. a. kosower , `` parton distributions with errors , '' _ in * la thuile 1999 , results and perspectives in particle physics * 255 - 261 _ ; j. pumplin , d. r. stump and w. k. tung , hep - ph/0008191 . | the first measurements of @xmath0 are higher than current theoretical predictions .
we investigate the sensitivity of these theoretical predictions upon a variety of factors including : renormalization scheme and scale , quark mass effects , higher twist , isospin violation , and pdf uncertainties . 9.5 in 10.0 in 9.5 in @=11 citexr[#1]#2@fileswauxout citeaciteforciteb:=#2#1 @=12
@=11 maketitle @box= # 1preprint#1
@=12 # 1eq .
( [ eq:#1 ] ) # 1eqs .
( [ eq:#1 ] ) # 1fig .
[ fig:#1 ] # 1table .
[ tab:#1 ] # 1 # 1sec .
[ [ sec:#1 ] ] # 1 # 1 |
almost all ultraluminous infrared galaxies [ uligs : @xmath5 m ) @xmath6 l@xmath3 ] have optical / near - infrared morphologies indicative of galaxy - galaxy mergers ( e.g. joseph & wright 1985 ; armus , heckman , & miley 1987 ; sanders et al .
1988a ; murphy et al . 1996 ; kim 1995 ) .
their large quantities of dust and molecular gas ( e.g. , sanders , scoville , & soifer 1991 ; solomon et al .
1997 ) , as well as evidence of abundant young star clusters in many uligs ( e.g. surace et al . 1998 ) , are strong evidence of recent / ongoing star formation .
molecular gas is also a likely source of fuel for active galactic nuclei ( agn ) such as quasars and radio galaxies , many of which have high @xmath8 and disturbed optical morphologies ( e.g. , stockton & mackenty 1983 ; mackenty & stockton 1984 ; heckman et al . 1986 ; smith & heckman 1989a , b ) , thus providing a possible connection between mergers and the building of supermassive nuclear black holes .
pks 1345 + 12 ( iras 13451 + 1232 : @xmath9 l@xmath3 ) is a prime candidate for the link between the ulig phenomenon and radio galaxies . with a radio luminosity of @xmath10 w hz@xmath11 ,
it is the most powerful radio galaxy detected in co(@xmath0 ) to date .
it also belongs to a family of `` warm '' ( @xmath12 , similar to the colors of seyfert galaxies : de grijp , miley , & lub 1987 ) infrared galaxies believed to be in a transition state between the `` cold '' ( @xmath13 ) ulig phenomenon , when rampant star - formation is occuring and the accretion disk is forming around the nuclear black hole , and the optical quasar phase ( sanders et al 1988a , b ) .
it is observed to have two nuclei with a projected separation of @xmath14 ( @xmath154 kpc : heckman et al .
1986 ; smith & heckman 1989a , b ; kim 1995 ) , a very compact radio jet ( @xmath16 pc ) , and an extremely high molecular gas mass ( @xmath17 m@xmath3 : to @xmath18 conversion factor of 4 m@xmath3 ( k km s@xmath11 pc@xmath19)@xmath11 ; see 3 . ]
mirabel , sanders , & kazes 1989 ) .
the ratios of the narrow optical emission lines in pks 1345 + 12 indicate that it contains a seyfert 2 nucleus ( sanders et al .
1988b ; veilleux et al .
further , recent near - infrared spectroscopic observations have detected broad ( @xmath20 2600 km s@xmath11 ) pa@xmath21 emission , indicating the presence of a quasar nucleus which is obscured at optical wavelengths ( veilleux , sanders , & kim 1997 ) .
the near - infrared spectroscopy of pks 1345 + 12 illustrates the importance of probing the nature and distribution of obscured nuclear energy sources in uligs at near - infrared wavelengths .
such studies also complement co interferometry , which provides spatial and kinematic information of the molecular gas reservoirs in these dusty systems . in this _ letter _ , the superior resolution of the hst near - infrared camera and multi - object spectrometer ( nicmos : 0.10.2@xmath4 resolution at 12 @xmath1 m ) and the owens valley millimeter array ( @xmath14 resolution at 3 mm ) are used to show for the first time the spatial coincidence of the molecular gas and the radio - loud nucleus of pks 1345 + 12 .
these observations provide further support for the relationship between mergers , molecular gas , and agn activity .
an @xmath22 km s@xmath11 mpc@xmath11 and @xmath23 are assumed throughout such that 1@xmath4 subtends @xmath15 2.0 kpc at the redshift of the galaxy ( @xmath24 ) .
hst nicmos observations of pks 1345 + 12 were obtained as part of a larger program to image infrared - luminous galaxies ( scoville et al .
1999 ; evans 1999a ) .
observations were obtained in a single orbit on 1997 december 5 using camera 2 , which consists of a @xmath25 array with pixel scales of 0.0762@xmath4 and 0.0755@xmath4 per pixel in @xmath26 and @xmath27 , respectively , providing a @xmath28 field of view ( thompson et al .
images were obtained using the wide - band filters f110w ( 1.10@xmath29 m , @xmath30 m ) , f160w ( 1.60 @xmath1 m , @xmath31 m ) , and f222 m ( 2.22 @xmath1 m , @xmath32 m ) , which provide a resolution ( fwhm ) of 0.11@xmath4 , 0.16@xmath4 , and 0.22@xmath4 , respectively .
the basic observation and data reduction procedures are the same as those described in scoville et al .
( 1999 ) ; the total integration times per filter setting for these observations were 480 sec ( 1.1 and 1.6 @xmath1 m ) and 600 sec ( 2.2 @xmath1 m ) .
flux calibration of the images were done using the scaling factors @xmath33 , and @xmath34 jy ( adu / sec)@xmath11 at 1.10 , 1.60 , and 2.22 @xmath1 m , respectively .
the corresponding magnitudes were calculated using the zeropoints 1775 , 1083 , and 668 jy ( rieke 1999 ) .
aperture synthesis maps of co(@xmath0 ) and 2.7 mm continuum emission in pks 1345 + 12 were made with the owens valley radio observatory ( ovro ) millimeter array during five observing periods from 1996 september to 1997 may .
the array consists of six 10.4 m telescopes , and the longest observed baseline was 242 m. each telescope was configured with @xmath35 mhz digital correlators . during the observations , the nearby quasar
hb89 1413 + 135 ( 1.5 jy at 103 ghz ; 14@[email protected]@xmath38 + 13@xmath39:[email protected]@xmath4 [ b1950.0 ] ) was observed every 25 minutes to monitor phase and gain variations , and 3c 273 and 3c 454.3 were observed to determine the passband structure .
finally , flux calibration observations of uranus were obtained .
the ovro data were reduced and calibrated using the standard owens valley data reduction package mma ( scoville et al .
the data were then exported to the mapping program difmap ( shepherd , pearson , & taylor 1995 ) .
the reduced 1.1 , 1.6 , and 2.2 @xmath1 m images are shown in figure 1a c .
the galaxy consists of two nuclei with a projected separation of @xmath41 ( 4 kpc ) .
low level surface brightness emission envelops both nuclei , with an east - west extent of 11@xmath4 ( 22 kpc ; full width at 0.5% the maximum flux density at 1.1 and 1.6 @xmath1 m ) , and a 5.5@xmath4 ( 11 kpc ) southern extent beyond the southeastern nucleus ( hereafter pks 1345 + 12se ) .
the radial surface brightness profile does not constrain the nature of the progenitor galaxies or the type of galaxy they are evolving into ; both an @xmath42 law and an exponential disk give reasonable fits to the profile ( see also scoville et al .
1999 ) .
pks 1345 + 12se is observed to be extended at all three wavelengths with a fwhm of 0.15@xmath4 ( 300 pc ) at 1.1 @xmath1 m .
the measured 1.1@xmath4 aperture magnitudes are 16.86 , 15.85 , and 15.31 at 1.1 , 1.6 , and 2.2 @xmath1 m , respectively , and the derived colors are thus @xmath43 = 1.01 and @xmath44 = 0.54 . in contrast ,
the northwestern nucleus ( hereafter pks 1345 + 12nw ) is unresolved with a fwhm of 0.11@xmath4 ( 220 pc ) at 1.1 @xmath1 m and magnitudes of 16.66 , 15.45 , and 13.96 at at 1.1 , 1.6 , and 2.2 @xmath1 m , respectively .
the near - infrared colors of pks 1345 + 12nw are extremely red ; @xmath43 = 1.20 and @xmath44 = 1.49 .
the red nature of pks 1345 + 12nw relative to pks 1345 + 12se is also evident in the 1.1@xmath4-aperture flux density ratios of the two nuclei ; @xmath45 decreases from a value of 0.82 at 1.1 @xmath1 m to 0.29 at 2.2 @xmath1 m .
both the co(@xmath0 ) emission and underlying 2.7 mm continuum in pks 1345 + 12 are unresolved .
the continuum flux density is 0.31 jy and is consistent with a power - law extrapolation of the radio flux density ( e.g. steppe et al . 1995 ) .
the co emission ( figure 2 ) has a @xmath46 km s@xmath11 , a flux density of 14@xmath47 jy km s@xmath11 , and a co luminosity of @xmath48 k km s@xmath11 pc@xmath19 .
thus , the line profile and luminosity are consistent with those derived from nrao 12 m telescope observations of pks 1345 + 12 ( i.e. , @xmath49 k km s@xmath11 pc@xmath19 : mirabel et al .
1989 ) , and confirms that the flux measured in the single - dish observations ( i.e. , 81@xmath4 fwhm beam ) is entirely recovered with ovro ( @xmath50 synthesized beam ) . assuming a standard ratio ( @xmath21 ) of co luminosity to h@xmath51 mass of 4 m@xmath3 ( k km s@xmath11 pc@xmath19)@xmath11 , which is similar to the value determined for the bulk of the molecular gas in the disk of the milky way ( scoville & sanders 1987 ; strong et al . 1988 ) , the molecular gas mass is calculated to be @xmath52 m@xmath3 , or 14 times the molecular gas mass of the milky way .
ranges from 25 m@xmath3 for a reasonable range of temperatures and densities , thus the molecular gas mass of pks 1345 + 12 may actually be as low as 1.6@xmath53 m@xmath3 . ] finally , using the fwhm beam of the co map ( 2.2@xmath4 ) , the molecular gas concentration is calculated to be @xmath542000 m@xmath3 pc@xmath55 , or @xmath54151500 times that observed in local early - type spiral galaxies ( e.g. , young & scoville 1991 ) and comparable to the concentrations observed in a sample of luminous infrared galaxies observed by scoville et al .
( 1991 ) and bryant & scoville ( 1999 ) .
the interpretation of the data presented in this _ letter _ depends on accurate astrometry of the multiwavelength images . to determine the coordinates of the two nuclei of pks 1345 + 12
, the positions of stars within 3.5@xmath40 of the galaxy were first retrieved from the usno - a1.0 database . a plate solution ( world coordinate system ) was then derived for a @xmath56 _ r_-band image ( kim 1995 ) , which also shows the two nuclei of pks 1345 + 12 , using the iraf task pltsol .
the coordinates of the two nuclei , along with the positions of the co and radio emission from the galaxy , are listed in table 1 .
the radio and co emission appear to be spatially coincident with pks 1345 + 12nw - the measured near - infrared peak of pks 1345 + 12nw is displaced 0.3@xmath4 ne of the radio emission , but is within the uncertainties associated with the positions of the stars used to derive the position of the pks 1345 + 12 nuclei .
likewise , the measured near - infrared peak of pks 1345 + 12nw is displaced @xmath57 ne of the co emission and continuum centroids , consistent with the measured ovro beamsize ( see figure 1d ) .
the bulk of the activity in pks 1345 + 12 is associated with pks 1345 + 12nw .
the derived nicmos colors of the two nuclei provides further support that the agn resides in pks 1345 + 12nw .
while the colors of pks 1345 + 12se are consistent with starlight reddened by 15 magnitudes of dust , pks 1345 + 12nw has colors similar to other warm uligs observed with nicmos - i.e. , similar to optically - selected quasars with a 5001000 k dust component ( see also scoville et al .
1999 and the summary in evans 1999a ) .
similar results are derived from near - infrared ground - based observations of pks 1345 + 12 ( e.g. , surace & sanders 1999 ) , and recent , high - resolution near - infrared spectroscopy ( veilleux & sanders 1999 ) confirm that pks 1345 + 12 is the source of the broad - lines detected by veilleux , sanders , & kim ( 1997 ) .
the presence of a large and concentrated reservoir of molecular gas in pks 1345 + 12nw is consistent with the notion that this gas is a source of fuel for the radio phenomenon ( e.g. , mirabel et al .
1989 ) . in this scenario ,
the molecular gas in the western galaxy is driven inward via gravitational instabilities induced by interactions with pks 1345 + 12se .
of all of the radio galaxies detected in co(@xmath0 ) to date ( phillips et al .
1987 ; mirabel et al . 1989 ; mazzarella et al . 1993
; evans 1999b ; evans et al .
1999 ) , pks 1345 + 12 is the most molecular gas - rich , and the only one that clearly has two nuclei .
thus , while pks 1345 + 12 is an advanced merger in terms of its relatively small nuclear separation , the fact that the stellar nuclei are 4 kpc apart implies that this system is dynamically younger than the other radio galaxies observed . if the assumption is made that the nuclei of pks1345 + 12 are gravitationally bound ,
the relative velocity of the merging galaxies is @xmath58 km s@xmath11 , where @xmath59 m@xmath3 and @xmath60 kpc , and thus the merger has at least an additional @xmath61 years before the nuclei coalesce . from the extent of the radio emission
, it is also clear that the jet activity commenced fairly recently . in the 2.3 and 8.5 ghz radio maps shown in fey ,
clegg , & fomalont ( 1996 ) , the radio jet has a maximum extent of 0.10@xmath4 ( @xmath62 pc ) .
thus , if the jet propagates at a speed of 0.1@xmath63 , it can be no more than 7000 years old .
for comparison , the linear extent of the jets associated with single - nuclei radio galaxies detected in co are 10200 kpc , but their estimated jet ages ( @xmath64 years ) are significantly less than the timescale of the merger process ( @xmath65 years ) .
the creation of the radio jets so late in the life of the merger can be understood in terms of merger dynamics - there will be a natural offset in the time at which the merger begins and the agn activity occurs because of the time it takes for the molecular gas to agglomerate in the nuclear regions of the galaxy .
further , depending on how typical such a delay is in radio galaxies , the consumption of molecular gas by extended star formation may be well underway prior to the onset of the radio activity , and may continue for another @xmath66 years or so .
this provides a natural explanation of why radio galaxies with older , extended jets are not observed to have large reservoirs of molecular gas ( i.e. , @xmath67 m@xmath3 : mazzarella et al .
1993 ; evans 1999b ) .
we thank the staff of the owens valley millimeter array and the nicmos gto team for their support both during and after the observations were obtained , and the referee sylvain veilleux for many useful suggestions .
ase also thanks m. shepherd , d. frayer , and j. surace for useful discussion and assistance .
ase was supported by nasa grant nag 5 - 3042 .
j.m.m . and
d .- c.k . were supported by the jet propulsion laboratory , california institute of technology , under contract with nasa .
( a c ) _ hst _ nicmos 1.1 , 1.6 , and 2.2 @xmath1 m images of pks 1345 + 12 .
the images have peak intensities of 7.5 , 11 , and 29 @xmath1jy for the 1.1 , 1.6 , and 2.2 @xmath1 m images , respectively .
the speckle pattern surrounding pks 1345 + 12nw at 2.2 @xmath1 m is a psf artifact .
( d ) continuum - subtracted co(@xmath0 ) emission superimposed on the false - color nicmos image .
the nicmos data are displayed with blue as 1.1 @xmath1 m , green as 1.6 @xmath1 m , and red as 2.2 @xmath1 m .
the co data are plotted as 50% , 60% , 70% , 80% , 90% , and 99% , where 50% corresponds to 3.9@xmath69 rms and 100% corresponds to a peak flux of 0.0339 jy / beam .
the co emission is unresolved , with a beam fwhm of [email protected] at a position angle of -70.5@xmath39 . | millimeter co(@xmath0 ) interferometry and high resolution , _ hubble space telescope _ ( hst ) 1.1 , 1.6 , and 2.2 @xmath1 m imaging of the radio compact galaxy pks 1345 + 12 are presented . with an infrared luminosity of @xmath2 l@xmath3 , pks 1345 + 12
is a prime candidate for studying the link between the ultraluminous infrared galaxy phenomenon and radio galaxies .
these new observations probe the molecular gas distribution and obscured nuclear regions of pks 1345 + 12 and provide morphological support for the idea that the radio activity in powerful radio galaxies is triggered by the merger of gas rich galaxies .
two nuclei separated by 2@xmath4 ( 4.0 kpc ) are observed in the near - infrared ; the extended southeastern nucleus has colors consistent with reddened starlight , and the compact northwestern nucleus has extremely red colors indicative of an optical quasar with a warm dust component .
further , the molecular gas , 3 mm continuum , and radio emission are coincident with the redder nucleus , confirming that the northwestern nucleus is the site of the agn and that the molecular gas is the likely fuel source . |
for simplicity , we will use one dimensional maps for the chaotic units .
the results generalize to other dynamical systems @xcite .
we consider a model system of @xmath0 dynamical units , each one of which , when isolated , satisfies @xmath1 , where @xmath2 is the value of unit @xmath3 at time @xmath4 and @xmath5 is a parameter vector for system @xmath3 . the systems , when coupled , are taken to satisfy ( e.g. , @xcite ) @xmath6 where @xmath7 is a function such that @xmath8
, @xmath9 is a symmetric laplacian matrix ( @xmath10 ) describing the network connections , and @xmath11 is a function independent of @xmath3 and @xmath12 . [ in our examples , we will take @xmath13 . ]
the constant @xmath14 determines the strength of the coupling .
if the systems are identical ( i.e. , @xmath15 for all @xmath3 ) , there is an exactly synchronized solution of eqs .
( [ eq : coupled ] ) , @xmath16 , whose time evolution is the same as the uncoupled dynamics of a single unit , @xmath17 , where @xmath18 .
the stability of the synchronized state can be determined from the variational equations obtained by considering an infinitesimal perturbation @xmath19 from the synchronous state , @xmath20 , @xmath21 let @xmath22 $ ] , and define the vector @xmath23 $ ] by @xmath24 , where @xmath25 is the orthogonal matrix whose columns are the corresponding real orthonormal eigenvectors of @xmath9 ; @xmath26 , @xmath27 , where @xmath28 is the eigenvalue of @xmath9 for eigenvector @xmath29 . then eqs . ( [ eq : linearized ] ) are equivalent to @xmath30\eta_{n}^{k}.\ ] ] the quantity @xmath31 is the weight of the @xmath29th eigenvector of @xmath9 in the perturbation @xmath32 .
the linear stability of each ` spatial ' mode @xmath29 is determined by the stability of the solution of eq .
( [ eq : componentwise ] ) . by introducing a scalar variable @xmath33 , the set of equations given by eq .
( [ eq : componentwise ] ) can be encapsulated in the single equation , @xmath34\eta_{n}.\ ] ] the master stability function @xmath35 @xcite is the largest lyapunov exponent for this equation .
this function depends only on the coupling function @xmath11 and the dynamics of an individual uncoupled element , but not on the network connectivity .
the network connectivity determines the eigenvalues @xmath28 ( independent of details of the dynamics of the chaotic units ) .
the stability of the synchronized state of the network is determined by @xmath36 , where @xmath37 indicates instability .
if the systems are slightly different , one gets instead of eqs .
( [ eq : linearized ] ) the equations @xmath38 where @xmath39 , and @xmath40 represents the effect of the mismatch and is assumed to be small .
terms of order @xmath41 were neglected . defining @xmath42 $ ]
, we obtain an equation analogous to eq .
( [ eq : componentwise ] ) , @xmath43\eta_{n}^{k } + ( q l)^{k},\ ] ] where @xmath44 is the kth element of the vector @xmath45 . the lyapunov exponent for the solution of eq .
( [ eq : componentwise ] ) is @xmath46 . assuming a solution of eq .
( [ eq : misma ] ) to have the same average damping as that for eq .
( [ eq : componentwise ] ) , then , if @xmath47 is negative for all modes , the amplitude of @xmath48 can be estimated as @xmath49 ( for example , if we model eq .
( [ eq : misma ] ) by the simple system @xmath50 , then @xmath51 satisfies , as @xmath52 , @xmath53 .
see @xcite . )
the largest lyapunov exponent @xmath47 above corresponds to a typical trajectory in the chaotic attractor . however , the lyapunov exponent for unstable periodic orbits embedded in the attractor might be larger .
assume that one of these periodic orbits has a positive lyapunov exponent and the attractor has a negative lyapunov exponent .
in this case , most of the time the amplitude of @xmath48 will be very small and given approximately by eq .
( [ eq : machete ] ) .
eventually the trajectory @xmath54 will get very close to this transversally unstable periodic orbit . while it is close to this orbit , @xmath55 in eq .
( [ eq : misma ] ) is no longer damped and gets exponentially amplified with the lyapunov exponent of the unstable periodic orbit .
the deviation from the synchronized state becomes large , producing a desynchronization burst .
if there are no other attractors , the system returns to the synchronized state and the process repeats .
desynchronization bursts can be expected when the master stability function for a typical trajectory is negative for all modes , and there is an embedded unstable periodic orbit which has a positive master stability function for at least one mode ( i.e. , @xmath56 for some @xmath29 , where @xmath57 is the master stability function of one of the embedded periodic orbits ) .
we will now use the fact that modes with the same eigenvalue have the same stability . for simplicity , assume that the coupling is all to all , so that all the modes ( except the mode in the synchronization manifold , which has zero eigenvalue ) have the same eigenvalue , @xmath58 .
( [ eq : machete ] ) implies that the coefficients in the eigenvector decomposition of the deviations from synchrony ( @xmath31 ) are , on average , proportional to those for the deviations of the mismatch parameters from their mean [ @xmath59 .
it follows that the mismatch vector @xmath60 is proportional to the vector @xmath61 while these approximations are valid .
if the vector @xmath61 is measured , the deviations of the mismatch from its mean can be determined approximately up to an unknown scaling factor . for this method to work ,
the measurements need to be made when the system is still in the linear regime .
since it is assumed that there is a limitation in the measurement accuracy , @xmath61 needs to be small enough to guarantee linear behavior , but large enough to be measured .
one can thus set up the system so that desynchronization bursts are expected and make measurements while a desynchronization burst is developing .
if the system allows continuous tuning of the coupling strength , one could also increase it so that the synchronous state becomes unstable and take measurements as the system desynchronizes .
to illustrate our method , we use the _ circle map _ , described by the equation @xmath62\mbox { mod } 1.\ ] ] we choose the parameters to be @xmath63 and @xmath64 .
these parameters produce a chaotic attractor in @xmath65 $ ] .
we found the embedded periodic orbits up to period four . to determine the orbits of period @xmath66 we used newton s method to find the roots of @xmath67 , where @xmath68 is described by eq .
( [ eq : map ] ) and @xmath69 denotes the @xmath66 times composition of @xmath70 .
eliminating all the orbits outside of the attractor , we found one period 1 orbit , two period 2 orbits , and one period 4 orbit .
we show in fig .
[ fig : msf ] the master stability functions of the orbits found , and the master stability function of the attractor .
here @xmath71 , where @xmath72 is the @xmath29th eigenvalue of the coupling matrix and @xmath14 is the global strength of the coupling . for definiteness
, we assume a network that is coupled _ all to all_. this means that for a network of @xmath0 systems , an element in the coupling matrix @xmath73 is given by @xmath74 this matrix has two distinct eigenvalues , @xmath75 and @xmath76 for @xmath77 .
we ignore the 0 eigenvalue since this corresponds to a perturbation in which all of the systems are displaced by the same amount ( thus they remain synchronized ) . due to the lack of other distinct eigenvalues , it is easy to pick an @xmath78 that produces desynchrozination bursts . for our map
the region where bubbling is expected is where @xmath79 ( note that for our example @xmath72 is the same for all @xmath29 ) .
we now present an example of the method where noise is negligible .
the mismatch is chosen to be in @xmath80 since it has a more complicated effect than mismatch in @xmath81 .
the coupled systems can then be described by the general equation @xmath82 \mbox { mod } 1,\ ] ] where @xmath83 is the mismatch in system @xmath3 , @xmath84 , and @xmath85 are indices representing the @xmath3th and @xmath12th system in the network [ cf .
( [ eq : coupled ] ) ] .
the term @xmath86 determines the coupling of the network .
we chose @xmath87 systems , @xmath88 as the global coupling , and @xmath89 \times 10^{-6}$ ] as the values for our examples .
it should , however , be noted that this method works for any number of systems ( with @xmath14 being adjusted accordingly ) and mismatch of any size , although , if the mismatch becomes small , the waiting time for a desynchronization burst becomes large .
the waiting time can be adjusted by changing the values of @xmath0 , @xmath14 , and @xmath90 . defining @xmath91 , where @xmath92 , we plot @xmath93 versus @xmath4 and look for desynchronization bursts in the network . in fig .
[ fig : bub1 ] we show the time evolution of @xmath93 near a desynchronization burst .
our interest is in the vector @xmath94 where @xmath95 is the first time that @xmath96 is in the _ sampling region _ , defined as @xmath97 ( see fig . [
fig : bub1 ] ) .
this region is determined by the limitations on the ability to accurately measure @xmath98 and the dynamics of the system considered .
the latter exists because the master stability function method relies on the systems being close to synchronization . during the desynchronization burst
, the difference in the systems can be so large that the linearization used in the analysis of sec .
i no longer applies .
we determined that the upper bound to this region in the circle map is @xmath99 or @xmath100 .
the lower bound was arbitrarily chosen as representing the accuracy of the measurements , which we assume is not enough to measure the mismatch directly .
generally the method becomes more effective the smaller the lower bound is .
according to the previous section , at time @xmath95 we should have approximately @xmath101 , where @xmath102 .
we can then obtain the relative deviations of the mismatch parameters , @xmath83 , by measuring the much larger values of @xmath103 . in fig .
[ fig : secii ] we show a superimposed plot of @xmath104 and @xmath105 versus @xmath3 where @xmath106 , the scaling factor , minimizes @xmath107 ^ 2 $ ] . in fig .
[ fig : secii ] we calculated @xmath108 and this corresponds to the amplification of the mismatch .
it should be noted that the sign of @xmath106 is undetermined unless we have knowledge of @xmath83 .
we see from the figure that @xmath109 .
the definition of the sampling region is somewhat arbitrary , and it may occur that nonlinear effects still play a role in the resulting spatial pattern of the burst .
in fact , in fig . [
fig : secii ] we observe that there are still small deviations from the real mismatch pattern . in order to take this into account
, we can take the average over various bursts . in the next section
, we will discuss how to appropriately take the average , and we will also deal with the effects of noise .
after learning from the simpler model in sec .
ii , we can now analyze a more realistic situation .
the method proposed and explained in the previous section applies to a similar network with noise , but there are a few adjustments to be made .
we use the same model described by eq .
( [ eq : nonoise ] ) , except we modify it to @xmath110\mbox { mod } 1,\ ] ] where @xmath83 , @xmath111 are defined in the same way as before , and @xmath112 is a random variable uncorrelated at different @xmath3 and @xmath4 and simulates the noise . in our example , we choose @xmath112 uniformly from the interval @xmath113 $ ] ( note that the noise and mismatch @xmath114 are of comparable size ) . as mentioned in the previous section , nonlinear effects might produce deviations from the simple relation @xmath115 .
we assume that the effects of the nonlinearity and the noise can be represented by a random variable @xmath116 , such that @xmath117 .
we furthermore assume that @xmath116 has zero mean . under these assumptions ,
the mismatch is given by @xmath118 where the brackets represent an average over realizations of @xmath119 . because of the definition of the sampling region , the scaling factors for different samples will have similar magnitude , but possibly different sign .
we thus get approximately , assuming the sign of @xmath106 is independent of @xmath116 , @xmath120 since the sign of @xmath106 is unknown , we use a _ least - squares optimization _ to find the signs which minimize the dispersion from the mean .
more precisely , if we have @xmath121 samples of @xmath94 s , we can define an average to be @xmath122 , where @xmath123 is a sequence of @xmath124 s and @xmath125 s and @xmath126 is the vector @xmath127 $ ] for the @xmath128th sample .
note that @xmath129 is an @xmath0-dimensional vector and that its @xmath3th component is an average of the @xmath3th component of the @xmath121 samples of @xmath94 s .
if we then minimize the error , defined by @xmath130 we can find an optimal sequence of @xmath124 s and @xmath125 s , which we shall call @xmath131 , that ensures most of the @xmath94 s are oriented the same way . to minimize this error and optimize @xmath132
it is impractical to test all possibilities since there are @xmath133 different @xmath132 s .
we can , however , follow an algorithm starting with a randomly generated @xmath132 . at each iterate
, we generate three new @xmath132 s .
the first , @xmath134 , is a new random sequence , @xmath135 is @xmath132 altered such that the signs of 1% of the sequence are changed , and @xmath136 is defined in a similar way but with 5% of the signs changed .
we can then compare @xmath134 , @xmath135 , @xmath136 , and @xmath132 and determine which one has the smaller error determined by eq .
( [ eq : err ] ) . the one with the smallest error
is then redefined as @xmath132 and the process is repeated until an approximation to @xmath131 , which we denote as @xmath137 , is found .
we can then define @xmath138 .
we show in fig .
[ fig : seciii ] a superimposed plot of @xmath139 and @xmath140 versus @xmath3 where @xmath141 minimizes @xmath142 ^ 2 $ ] . to obtain @xmath143 we repeated the process of optimization for @xmath144 @xmath145 times
according to the discussion above , we should have approximately @xmath146 where @xmath147 .
indeed we see that @xmath148 even when the noise was comparable to the mismatch [ @xmath149 in eq .
( [ eq : noisy ] ) ] .
we have presented a method to use large deviations from synchronization in order to determine the characteristics of the parameter mismatch in a collection of nearly identical chaotic dynamical systems .
it has been noted that knowledge and manipulation of the mismatch patterns can be advantageous in order to improve the quality of the synchronization @xcite .
the main advantage of our method is that it only requires direct knowledge of the synchronization error when it is large enough to be measured .
furthermore , in principle , there are no limitations on the number of systems it can handle . on the other hand
, the method only provides the relative deviations from the mean and has yet to be extended to systems with comparable mismatch in different parameters .
we have demonstrated our method by determining the relative parameter mismatch in an ensemble of @xmath150 circle maps . by measuring the large deviations from the synchronized state that occur during a desynchronization burst
, we were able to determine the very small relative differences in parameters ( see fig .
[ fig : secii ] ) .
we considered the presence of noise , and dealt with it by suitably averaging the measurements taken for various desynchronization bursts . for a noise comparable to the mismatch we were able to determine the relative parameter mismatch by averaging @xmath151 realizations ( see fig .
[ fig : seciii ] ) . for both situations
, we were able to determine the relative parameter mismatch from measurable values even when the mismatch itself was assumed to be immeasurable .
a. pikovsky , m.g .
rosenblum , and j. kurths , _ synchronization : a universal concept in nonlinear sciences _ , ( cambridge university press , cambridge , 2001 ) .
e. mosekilde , y. maistrenko , and d. postnov , _ chaotic synchronization : applications to living systems _
( world scientific , singapore , 2002 ) .
u. parlitz , phys .
lett . * 76 * , 1232 ( 1996 ) . | we present a method to determine the relative parameter mismatch in a collection of nearly identical chaotic oscillators by measuring large deviations from the synchronized state .
we demonstrate our method with an ensemble of slightly different circle maps .
we discuss how to apply our method when there is noise , and show an example where the noise intensity is comparable to the mismatch .
the study of networks of coupled dynamical systems is an important area of research with applications in diverse fields , ranging from biology to laser physics @xcite-@xcite .
the synchronization of coupled oscillators has been under extensive study in recent years and , in particular , the synchronization of identical oscillators has received considerable interest @xcite . since it is impossible in practice to obtain identical oscillators , the effect of the difference in the parameters of the oscillators , or _ parameter mismatch _ , might be relevant in some applications .
it might be desired to have dynamical units as similar to each other as possible , or to know the characteristics of the parameter mismatch in a collection of nearly identical systems . in this paper
we propose a method to use deviations from synchronization to extract information on the parameter mismatch of the coupled dynamical units .
existing methods for parameter estimation ( see , for example , refs .
@xcite-@xcite ) usually rely on knowledge of the typically small synchronization error .
our method depends on relatively large deviations from the synchronized state , and might be useful in cases in which the small synchronization error can not be measured accurately .
when a number of identical systems are appropriately coupled in a network , a solution exists in which the state of all oscillators at all times is the same .
this is referred to as _
identical _ synchronization .
this concept is useful only when the systems are identical .
we will deal with systems that are nearly , but not exactly , identical .
we will refer to a situation in which the states of the systems are very close to each other as _ nearly identical _ synchronization .
a method to determine the stability of the synchronous state when the systems are identical , the _ master stability function _ , has been proposed by pecora and carroll @xcite . in the case of
nearly identical chaotic systems , the nearly synchronized state might be interrupted by relatively short periods of desynchronization ( _ desynchronization bursts _ ) .
these bursts develop with spatial patterns on the network .
these spatial patterns , and the parameters for which they can be expected , can be predicted from the laplacian matrix describing the network connections , the master stability function of the attractor , and the unstable periodic orbits embedded in it @xcite .
the spatial patterns of the bursts depend on the parameter mismatch of the different systems .
we use this fact to infer the relative deviation of the parameters of the individual units with respect to their mean from the desynchronization bursts .
the proposed method is as follows .
the oscillators are connected in such a way that the parameter mismatch determines the spatial patterns of the desynchronization bursts . as we will see later
, one such way is all - to - all coupling .
the system is set up in a parameter region in which desynchronization bursts are expected .
while a burst is developing , measurements are taken of the deviations of the different systems from the synchronous state . from these observations ,
the relative deviations of the parameters from the mean are deduced .
some limitations of this method are the following .
it is assumed that measurements can be taken with enough precision such that the deviations from the synchronous state can be measured in the linear regime .
also , the presence of noise affects the spatial patterns of the bursts .
although we will describe how to deal with the noise , the effectiveness of the method decreases as the ratio of noise to mismatch increases .
it is also assumed that unavoidable small differences in the way in which the systems are connected to each other does not introduce a difference between the systems which is of the same order of magnitude or larger than the parameter mismatch being measured . in sec .
i we briefly describe the master stability function method and its extension to deal with nearly identical systems . in sec .
ii we present and illustrate our method with an example in the case where the noise is negligible . in sec .
iii we discuss how to deal with the noise and show an example . in sec .
iv we present our conclusions . |
a good model of a supernova remnant s ( snr s ) blast wave propagating through the ambient medium allows us to derive , from the observed quantities , important properties of the medium , as well as the explosion energy .
the sedov solution ( 1959 ) for adiabatic expansion into an isotropic medium usually serves as the prototype , and further , more realistic models also assume isotropy ( see e.g. jun , jones & norman 1996 ) . nevertheless , in most cases of astrophysical interest ( stellar winds , garcia - segura & mac low 1995 ; snrs in vicinity of molecular clouds , dohm - palmer & jones 1996 ; galactic supershells , maciejewski et al . 1996 ; and starburst galaxies , jogee , kenney & smith 1998 ) , the ambient medium is not isotropic . at the next level of approximation , kompaneets ( 1960 ) found an implicit solution for a blast wave propagating in an exponentially stratified ambient medium , under the assumption that the post - shock pressure is uniform over the surface of the remnant .
an explicit form can be obtained when one additional assumption is made , that the shape of the shock wave can be approximated by a prolate ellipsoid .
after an alternative presentation of the kompaneets solution in 2 , we derive explicit analytical formulae for the size , velocity and the onset of cooling in 3 .
a prolate ellipsoid is a good approximation for all evolutionary times prior to blowout ( kompaneets 1960 ) , but we show in 4 that the kompaneets approximation of uniform post - shock pressure loses validity much earlier . in
5 we show how our explicit approximation can be used in exploring the parameter space to identify values of interest in making 2-dimensional numerical models of snrs .
before we introduce our assumption , let us show a simple way of re - deriving the basics of the original kompaneets solution , using one of his conclusions , namely that in the direction perpendicular to the density gradient , the remnant is widest halfway between the high and low ambient density extremes of the shock , a location we shall refer to as the equator . with the resulting formulae for the remnant size parallel and perpendicular to the gradient , our approximation for the shock shape immediately provides the volume and pressure needed to learn the time evolution .
we consider the shock expansion in cylindrical coordinates @xmath0 : the explosion occurs at @xmath1 ( fig.1 ) .
the ambient density @xmath2 is an exponential function of @xmath3 : @xmath4 where @xmath5 is the density at the explosion site and @xmath6 is the stratification scaleheight . denoting @xmath7 and @xmath8 as the location of the dense and the tenuous end of the remnant , respectively , we have @xmath9 , where @xmath10 is the extent of the remnant in the direction along the density gradient . for a strong , non - radiative shock in a @xmath11 gas , the post - shock pressure @xmath12 , pre - shock density @xmath2 and shock velocity @xmath13 are related by @xmath14 assuming after kompaneets that the post - shock pressure @xmath12 is constant over the surface of the remnant , the shock speed at a particular time can be written as a function of @xmath3 : @xmath15 comparing the values of shock velocities at the ambient density extremes and at the equator , we can eliminate the unknown @xmath16 and calculate the evolution of the remnant shape as a function of its semi - major axis @xmath17 .
the rate of expansion along the density gradient is an average of shock velocities at two density extremes : @xmath18 and @xmath19 @xmath20 where @xmath18 and @xmath19 are velocity magnitudes , @xmath7 is negative , and @xmath21 . the center of the remnant is displaced from the explosion site by @xmath22 . at @xmath23 ( the equator ) ,
the remnant expands only in the lateral direction and the expansion rate is @xmath24 although the equator moves along the @xmath3 axis ( fig.1 ) as the remnant expands , the rate of change of the remnant s semi - minor axis @xmath25 is exactly equal to @xmath26 : @xmath27 .
the evolution of the displacement @xmath28 can be written as @xmath29 after substituting @xmath30 and @xmath31 and simple manipulations , we can eliminate the constant @xmath16 and get @xmath32 integrating the equations above , we can readily find @xmath25 and @xmath28 as functions of @xmath17 @xmath33 these results are independent of the assumed shape of the blast wave and the same as found by kompaneets ( 1960 ) , after his auxiliary variable @xmath34 is recognized as @xmath35 .
in particular , the equations above allow us to estimate the flattening of the remnant @xmath36 and the relative displacement of the center @xmath37 .
it appears that even remnants extending several scale - heights retain a nearly spherical shape : for example , if @xmath38 , the ambient density contrast between the ends of the shock is almost 30 , but the remnant remains barely flattened ( size ratio @xmath36=0.9 ) . on the other hand ,
the ellipsoid s center is displaced from the explosion site by @xmath39 in the low density direction .
the same conclusions about sphericity and the center shift can be drawn from the analytical solution based on sectoral approximation ( gnatyk 1988 ) .
hydrodynamical models constructed by dohm - palmer & jones ( 1996 ) arrive to these conclusions as well .
one can also notice that @xmath40 : in the blowout case , when @xmath41 , it is 0.5 .
this is kompaneets famous result , that as blowout occurs and @xmath8 becomes large , @xmath7 goes to a constant @xmath42 .
the lateral size at blowout also agrees with kompaneets prediction : for @xmath41 , we get @xmath43 .
the implicit solution by kompaneets shows that the shock has the shape of a somewhat boxy ellipsoid .
in fact , its shape is virtually indistinguishable from a true ellipsoid within limits of accuracy set by kompaneets assumption of uniform post - shock pressure .
we use this fact in order to introduce time as an independent variable instead of @xmath17 , when we assume that the post - shock pressure @xmath12 has the same volume relationship as in the 1d sedov solution ( see e.g. bisnovatyi - kogan & silich 1995 ) @xmath44 where @xmath45 , @xmath46 is the explosion energy and @xmath47 is the volume of the remnant approximated by an ellipsoid . combining equations ( [ e : a0 ] ) ,
( [ e : a1 ] ) and ( [ e : a8 ] ) , and identifying @xmath23 along the minor axis of the ellipsoid we have @xmath48 after expressing @xmath28 and @xmath25 as functions of @xmath17 , we get @xmath49 which , after integrating gives @xmath50 where @xmath51 is the nominal time at which a sedov remnant in a homogeneous medium would reach a radius of @xmath52 , and the integral @xmath53 is defined in table 1 for numerical evaluation .
the numerical form of @xmath51 is @xmath54 where @xmath55 is @xmath6 in parsecs , @xmath56 is @xmath46 in units of 10@xmath57ergs , and @xmath58 is the nuclear number density of the ambient medium in @xmath59 . in this notation ,
the mass density is @xmath60 , where the average mass per nucleus is @xmath61 , @xmath62 being the mass of hydrogen .
the shock velocity at any @xmath3 can be easily calculated by combining equations ( [ e : a1 ] ) and ( [ e : a8 ] ) and substituting the local ambient density for @xmath2 . in the particular case of top , bottom and equator expansion velocities
, one can use equation ( [ e : a3 ] ) , which after combining with equation ( [ e : a2 ] ) gives @xmath63 . after substituting @xmath64 from equation ( [ e : a10 ] )
, one gets explicit formulae for shock velocities at two density extremes : @xmath18 and @xmath19 .
obviously , the shock expansion velocity on the equator is the geometrical average of @xmath18 and @xmath19 these results are given in table 1 .
the ratio of expansion velocities at the two ends is @xmath65 . in snr evolution in a uniform medium ,
the adiabatic era is brought to close by the onset of significant radiative cooling , at remnant radius @xmath66 , closely followed by formation of a dense shell .
when a remnant is evolving in a density gradient of scaleheight @xmath6 , cooling occurs first on the dense end , and then spreads over the remnant surface .
the evolution has two characteristic scales : the kinematic @xmath6 and the radiative @xmath67 , with a history of shell formation that is dependent on their ratio . in order to estimate cooling times
, we used kahn s cooling law approximation ( 1976 ) , in which the cooling coefficient takes the form @xmath68 .
it is a reasonable approximation to the actual cooling coefficient for the temperature range of cooling snrs ( smith 1996 ) , and it has the special property that in the absence of thermal conduction , the time for a parcel of hot gas to cool from an initial state is independent of its history and equals @xmath69 for a newly shocked parcel of gas , the post - shock values of pressure @xmath12 , temperature @xmath70 , and density @xmath71 are known , so we can find @xmath72 .
this incremental cooling time @xmath73 is added to the time at which a parcel is shocked , to find the actual cooling time of that parcel , and then the sum minimized over all parcels to find the earliest parcel to cool .
for a homogeneous medium , @xmath72 is proportional to @xmath74 and therefore to @xmath75 or @xmath76 .
numerically , the minimum of @xmath77 occurs at @xmath78^{5/14}$ ] such that the first gas to cool was initially a distance @xmath79 from the explosion site , and was shocked at time @xmath80 ( see cox & anderson 1982 , cox 1986 ) .
the expression for @xmath81 can be obtained by substituting equation ( [ e : a1 ] ) to ( [ e : deltat ] ) , and using formulae for the sedov radius and velocity from table 1 .
one can get then @xmath82 \ ; \left ( e^{0.6}/n_0^{1.6 } \right)$ ] . at the minimum , @xmath83 , for other radii , @xmath84 .
first parcel s cooling is complete at @xmath85 , when the remnant radius is @xmath86 .
we define @xmath87 as the cooling time of the remnant
. in our approximation for the remnant in a density gradient , a combination of equations ( [ e : a0 ] ) , ( [ e : a8 ] ) and ( [ e : deltat ] ) gives @xmath88 where we substituted @xmath89 k@xmath90@xmath91erg s@xmath92 in kahn s formula for cooling .
noting that @xmath93 yields @xmath94 a measure of the likehood of finding a remnant with a cold shell only on the dense end is provided by the ratio @xmath95 of cooling times between the tenuous and dense ends .
these cooling times can be obtained by minimizing the value of the sum @xmath96 of times given by equations ( [ e : a12a ] ) and ( [ e : a11 ] ) over @xmath97 @xmath98 \ , \ ] ] where the function @xmath99 is given in table 1 , and the upper ( lower ) sign corresponds to the dense ( tenuous ) end .
the formula to derive values of @xmath34 at minimum , @xmath100 , is given in table 1 .
one can see that @xmath100 is a function of only one variable : @xmath101 .
the ratio of cooling times between the tenuous and dense ends of the remnant is numerically close to the density contrast between the two ends , @xmath102 , at the nominal cooling radius .
figure 2 displays the ratio of cooling times as a function of @xmath103 .
note that for cooling time ratios less than 10 ( essentially all for which the parameter is interesting ) , it is equal to the density contrast @xmath103 within 20% . from figure 2
we see that cooling takes more than twice as long on the tenuous end for @xmath104 , or roughly speaking , for @xmath105 , yielding the unsurprising result that when the density differential between the two ends is large at the nominal onset of cooling , the evolution will contain a long period with a partial shell .
this result may contribute to the explanation of the fact that we usually can see only one side of the expanding hi@xmath106shells ( heiles 1979 , koo & heiles 1991 ) . for completeness
, we now provide an approximate evaluation of the time at which the first parcel on the equator cools , as a measure of the proximity of a remnant to having half a complete shell .
this is somewhat more difficult , because the location of the equator is shifting with @xmath3 .
any parcel cooling at the equator was not on the equator when shocked . nevertheless , we are able to provide a plausible approximation for the nearly spherical case .
we use equation ( [ e : a8 ] ) , which relates the post - shock pressure to the explosion energy and volume , but approximate the volume by @xmath107 , and then substitute for @xmath17 the solution of the sedov 1d problem @xmath108 .
the approximate post - shock pressure is then @xmath109 from equations ( [ e : deltat ] ) and ( [ e : a13 ] ) , the cooling time for a parcel on remnant s minor axis , which is hit by the shock at the time @xmath110 , when the center of the remnant was at a distance @xmath28 from the explosion site , is @xmath111 by assuming that @xmath112 , which is an exact equality for the uniform medium only , an approximate formula for @xmath110 can be derived .
the corresponding approximate expression for @xmath87 is @xmath113 and approaches the cooling time for sedov solution when @xmath114 .
the dependence on @xmath28 can be removed by expressing @xmath28 in terms of @xmath17 from equation ( [ e : a7 ] ) and substituting @xmath17 from the sedov 1d solution .
thus we find the iterative solution for the cooling time on the minor axis @xmath115
the assumptions of uniform post - shock pressure and an exponential atmosphere lead directly to an equation ( [ e : a7 ] ) for the displacement @xmath28 of the remnant center from the explosion site , as a function of the semimajor axis , @xmath17 . in the hydrodynamical models performed for the particular case of snr w44 by shelton ( 1998 , paper ii )
, we noticed that this equation appeared to overestimate the displacement of the center .
that implies that the post - shock pressure at the high density end is higher than at the low density end . in the following ,
we provide a rough justification for such a differential .
note that numerical studies of @xmath116 may yield a reasonably useful test case for intercomparison of 2d hydrocodes .
the upturn in pressure at the edge of the sedov remnant can be attributed to the effective gravity in the decelerating post - shock gas .
the post - shock mass velocity is @xmath117 , so the outwardly directed effective gravity is roughly @xmath118 . assuming that the post - shock gas is in hydrostatic equilibrium , we get @xmath119 , given @xmath120 ( @xmath121 is the coordinate locally perpendicular to the shock front and directed outwards ) . the jump condition ( [ e : a1 ] ) however specifies the post - shock pressure , so the fractional gradient is @xmath122 from the same jump condition we have @xmath123 . for the exponential atmosphere , @xmath124 at the dense and tenuous ends of the remnant respectively .
thus @xmath125 or finally , at the ends of the remnant , @xmath126 in this equation , the leading term on the right - hand side is the positive gradient term present in the sedov solution for a uniform medium .
the second term is the correction factor we are looking for , positive on the high density end , negative on the low density end .
it expresses the fact that the higher density end has a more rapid deceleration and therefore a steeper pressure gradient . for the sedov case , equation ( [ e : a8 ] ) gives @xmath127 . with the semimajor axis @xmath17 in place of @xmath128
, we use equation ( [ e : a19 ] ) to estimate the relative gradients between the two ends at @xmath129 upon integrating from the central pressure plateau up the slope along the major axis , the ratios of the post shock pressures should be similar to , but less extreme than , this ratio of gradients . as a rough parameterization , we can write it as @xmath130 where @xmath131 is somewhat greater than 3 . at the end of this section
, we provide a way of estimating the value of @xmath131 from numerical studies .
now , we can see that the kompaneets approximation of constant post - shock pressure holds only for @xmath132 , i.e. at times well before blowout .
our approximation of the shock by an ellipsoid is indistinguishable from the kompaneets solution at this stage ( the @xmath133 deviations smaller than 10@xmath134 ) .
the post - shock pressure can be expressed as a function of time only , when @xmath17 in equation ( [ e : a21 ] ) is approximated by the 1d sedov solution @xmath135 where the top sign refers to the dense end , the bottom to the tenuous one , and @xmath136 has replaced @xmath137 . keeping in mind that we are looking for a first order correction , we used the fact that at early times @xmath138 . for a strong , non - radiative shock , @xmath139 . taking into account that @xmath140 , but @xmath141 , we can write @xmath142 which after integrating gives @xmath143 where we first expanded the numerator and denominator to the two leading terms , and then substituted @xmath17 from the sedov solution . note that for uniform post - shock pressure , the ratio ( [ e : a24 ] ) is equal 1 , and after substituting @xmath30 , @xmath31 , one gets the formula ( [ e : a7 ] ) for the center shift @xmath28 .
similarly , in the case of nonuniform post - shock pressure , we express @xmath8 and @xmath7 in terms of @xmath17 and @xmath28 and after simple algebraic manipulations we get the final formula for the center shift correction @xmath144 the first order @xmath28 correction is about 4 times smaller than the @xmath12 correction in equation ( [ e : a21 ] ) .
the center shift correction derived above applies to models with no cooling ; for numerical models , one could plot the quantity @xmath145 to evaluate the free parameter @xmath131 in our method .
assuming that the shock wave propagating in an exponentially stratified medium takes the shape of an ellipsoid , we were able to find explicit expressions for its size and expansion velocity as functions of time , and for the cooling time at both ends and on the equator . as presented in table 1 ,
these expressions take forms similar to the sedov ( 1959 ) solution for a uniform ambient medium .
nevertheless , while the sedov solution is self - similar , there are no such solutions for nonuniform media .
the reason is that there are at least three independent dimensional parameters ( in our case @xmath46 , @xmath5 and @xmath6 ) .
a successful model of a snr should adopt parameters that reproduce the observed quantities : for example the size , expansion velocity and progress of shell formation .
if a pulsar is seen , the age can be derived as well .
our simple formulae provide quick ways to explore this 3-dimensional parameter space .
a useful approach is to generate topographic ( or contour ) plots on the @xmath5-@xmath6 plane for assumed values for @xmath46 and linear size of the remnant .
the contoured quantities include the time required to reach the assumed size , the post - shock pressure , the cooling and shell formation time - scales for the dense , equatorial , and tenuous directions , and the shock velocities in those directions .
although the kompaneets model , like the sedov model , can not be used very reliably to estimate quantities after shell formation has occurred , we make the usual approximation that shortly after shell formation the shell velocity is about three - quarters of the shock velocity predicted for the non - radiative evolution . by this
means of exploring the parameter space , a set of parameters fully reproducing the observables for snr w44 was found and is given in paper ii .
there are other asymmetric snrs which can be interpreted in terms of expansion into a nonuniform medium , for example a partially open hi@xmath106shell in ctb 80 ( koo 1990 ) , and an incomplete shell of cta 1 ( pineault 1993 ) . the semi - circular shape of ctb 109 is ascribed to interaction with a molecular cloud ( tatematsu 1990 ) .
g 84.2 -0.8 ( feldt & green 1993 ) shows a striking resemblance to w44 in radio continuum , though observations of the hi@xmath106shell are not convincing . in the catalogue of snrs with hi@xmath106shell emission (
koo & heiles 1991 , table 3 ) all sources show high - velocity hi@xmath106gas on one side only : either receding or approaching . heiles ( 1979 ) , who catalogued hi@xmath106shells in our galaxy , noticed that for most of them , he could see only the approaching or only the receding hemisphere a fact that he called `` disturbing '' .
putting aside possible observational biases , a partially formed shell appears to be more a rule than an exception .
we want to point out that explanation of such asymmetric snrs or shells does not require any abrupt density change , such as encountering the edge of a molecular cloud ( dohm - palmer & jones 1996 ) .
the method of estimation of the cooling time presented in 3 shows that the ratio of the cooling times on the dense and tenuous ends of the remnant is of the order of the ratio of ambient density extremes .
we can see one side of the hi@xmath106shell only , because at first there is only one side , and then by the time the other side is formed , either the part expanding into denser medium has slowed down and its emission blends with that of local gas or the tenuous end remains invisible due to the high column density contrast between low and high density ends ( silich 1992 ) .
our explicit analytical approximation provides an efficient tool for a quick exploration of the space of initial parameters for the snr models .
it can be used to select the initial parameters of hydro runs , which in turn can give us a detailed insight into the structure of the remnant , and can verify the above conclusions .
we would like to thank the referee , s.a .
silich , for valuable comments , and linda sparke for helpful suggestions on the paper .
dpc was supported in part by nasa grant nag5 - 3155 to the university of wisconsin .
bisnovatyi - kogan , g.s .
, & silich , s.a .
1995 , rev .
67 , 661 cox , d.p .
1986 , apj , 304 , 771 cox , d. p. & anderson , p. r. , 1982 , apj , 253 , 268 dohm - palmer , r.c .
, & jones , t.w .
1996 , apj , 471 , 279 feldt , c. , & green , d.a .
1993 , a&a , 274 , 421 garcia - segura , g. , & mac low , m .- m .
1995 , apj , 455 , 145 gnatyk , b.i .
1988 , sov .
, 14 , 309 heiles , c. 1979 , apj , 229 , 533 jogee s. , kenney , j.d.p . , & smith , b.j .
1998 , apj , 494 , l185 jun , b .-
i . , jones , t.w . , & norman , m.l .
1996 , apj , 468 , l59 kahn , f. 1976 , a&a , 50 , 145 kompaneets , a.s . 1960 , soviet phys .
dokl . , 5 , 46 koo , b .- ch . ,
reach , w.t . ,
heiles , c. , fesen , r.a . , & shull , j.m .
1990 , apj , 364 , 178 koo , b .- ch . , & heiles , c. 1991 , apj , 382 , 204 maciejewski , w. , murphy , e.m . ,
lockman , f.j .
, & savage , b.d .
1996 , apj , 469 , 238 pineault , s. , landecker , t.l . ,
madore , b. , & gaumont - guay , s. 1993 , aj , 105 , 1060 sedov , l. i. , 1959 , similarity and dimensional methods in mechanics , ( new york : academic press ) shelton , r.l . ,
cox , d.p . ,
maciejewski , w. , smith , r.k . ,
plewa , t. , pawl , a. , & ryczka , m.n . , ( paper ii ) 1998 , apj , submitted ( astro - ph/9806090 ) silich , s.a .
1992 , astrophys .
space sci .
, 195 , 317 smith , r.k . , krzewina , l.g . ,
cox , d.p . , edgar , r.j . & miller , w.w .
1996 , apj , 473 , 864 tatematsu , k. , fukui , y. , iwata , t. , seward , f.d . , & nakano , m. 1990 , apj , 351 , 157 @xmath146 } = 17.32 \ ; \frac{r_{\rm pc}^{5/2 } \ ; n_*^{1/2}}{e_{51}^{1/2 } } & & t \ { \rm [ yr ] } = 17.32 \ ; \frac{(2h_{\rm pc})^{5/2 } \ ; n_*^{1/2}}{e_{51}^{1/2 } } \ ; i\left(\frac{a}{2h}\right ) \\ & & { \rm where } \ ; i(x ) \equiv \frac{5}{2 } \int_{0}^{x } dy \frac{\sqrt{y}}{\cosh^2y } \arctan \sinh y \ \simeq \
\frac{3x^{5/2}}{3 + 2x^{5/2 } } \ { \rm for } \
x<3\\ \hline & & \\ v_s \ = \ \frac{2}{5 } \ ; \zeta^{5/2 } \ ; \left(\frac{e}{\rho_*}\right)^{1/2 } r^{-3/2 } & { \bf velocity - size } & v_s \ = \ \frac{2}{5 } \ ; \zeta^{5/2 } \ ; \left(\frac{e}{\rho_*}\right)^{1/2 } ( 2h)^{-3/2 } \ ; \ ; f_\pm\left(\frac{a}{2h}\right ) \\ & & \\ v_s
\ { \rm [ km / s ] } = 22600 \ , \left(\frac{e_{51}}{n_*}\right)^{1/2 } r_{\rm pc}^{-3/2 } & & v_s \ { \rm [ km / s ] } = 22600 \ , \left(\frac{e_{51}}{n_*}\right)^{1/2 } ( 2h_{\rm pc})^{-3/2 } \
; f_\pm\left(\frac{a}{2h}\right ) \\ & & { \rm where } \ ; f_\pm(x ) \equiv \frac{\exp(\mp x ) \ ; \cosh x}{\sqrt{x } \ ; \arctan \sinh x } \hspace{1 mm } ^{*}\\ \hline & & \\ t_{\rm
cool } \ { \rm [ yr ] } = 4.6\times10 ^ 4 \ ; \frac{e_{51}^{3/14}}{n_{*}^{4/7 } } & { \bf cooling \ ; time } & t_{\rm cool } \ { \rm [ yr ] } = 17.32 \ ; \frac{(2h_{\rm pc})^{5 /2 } n_*^{1/2}}{e_{51}^{1/2 } } \ ; \left [ i(x_c^\pm ) \ ; + \ ; a \ : g_\pm(x_c^\pm ) \right ] \\ & & { \rm where } \ ; \ ; g_\pm(x ) \equiv \frac{\cosh^5x \ ; \exp(\mp5x)}{x^{3/2 } \ , ( \arctan \sinh x)^3 } \hspace{1 mm } ^ { * } , \ ; a \ ; = \ ; 6.2\times10 ^ 8 \ ; \frac{e_{51}^2}{n_{*}^3 ( 2h_{\rm pc})^7}\\ & & { \rm and } \ ; x_c^\pm \ ; { \rm is \ ; set \ ; by } \ ; a = \frac{5}{2 } \frac{x^2 \ , \exp(\pm 5x ) \ , ( \arctan \sinh x)^4}{\cosh^6x \ , \left(\pm 5 { \rm e}^{\mp x } + \frac{3 \cosh x}{2 x } + \frac{3}{\arctan \sinh x}\right ) } \hspace{1 mm } ^{*}\\ \hline\end{aligned}\ ] ] @xmath149 the top signs refer to expansion into the dense medium , the bottom into the tenuous medium ; for the shock expansion velocity in the equatorial direction , the exponential factor should be dropped from @xmath150 | we propose simple , explicit , analytical approximations for the kinematics of an adiabatic blast wave propagating in an exponentially stratified ambient medium , and for the onset of radiative cooling , which ends the adiabatic era .
our method , based on the kompaneets implicit solution and the kahn approximation for the radiative cooling coefficient , gives straightforward estimates for the size , expansion velocity , and progression of cooling times over the surface , when applied to supernova remnants ( snrs ) .
the remnant shape is remarkably close to spherical for moderate density gradients , but even a small gradient in ambient density causes the cooling time to vary substantially over the remnant s surface , so that for a considerable period there will be a cold dense expanding shell covering only a part of the remnant .
our approximation provides an effective tool for identifying the approximate parameters when planning 2-dimensional numerical models of snrs , the example of w44 being given in a subsequent paper . |
early , confessional attempts aimed at establishing a consistent biblical chronology , such as the seder olam rabbah , strove to identify the timeline of events in the pentateuch purely from information contained in the text , such as genealogical information and narrative clues .
while internally consistent , the resulting chronologies were ultimately incompatible with a growing body of scientific and historical knowledge : for instance , the seventeenth - century irish bishop james ussher calculated that god created the universe on october 22 , 4004 bce [ 1 ] , an untenable thesis in the light of modern physical cosmology .
the pentateuch s own insistence on timekeeping suggests that a precise timeline for at least some of the events was available at the time the text was written .
yet , dauntingly , the genealogical and biographical information contained in it proved insufficient to identify a unique , reliable chronology , and as a consequence none of those events could be provided with a firm historical basis .
modern , non - confessional approaches have variously attempted to correlate the narratives with external information , such as historical records or archaeological evidence , in order to identify a reliable timeline for the events described in the pentateuch . yet
, the introduction of external elements has often been carried out at the expense of consistency with internal textual constraints : for instance , the birth of agriculture in the near east according to archaeological evidence seems irreconcilable with the dating implied by the genealogies of genesis . as we shall see
the pentateuch does contain a complete and reliable internal timekeeping system , but its reckoning of time is not limited to time spans : it also involves astronomical information , reflecting the high status of the study of astronomy in judaism .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ r. simeon b. pazzi said in the name of r. joshua b. levi on the authority of bar kappara : he who knows how to calculate the cycles and planetary courses , but does not , of him scripture saith , but they regard not the work of the lord , neither have they considered the operation of his hands [ isa 5:12 ] .
r. samuel b. nahmani said in r. johanan s name : how do we know that it is one s duty to calculate the cycles and planetary courses ? because it is written , for this is your wisdom and understanding in the sight of the peoples [ deu 4:6 ] : what wisdom and understanding is in the sight of the peoples ? say , that it is the science of cycles and planets .
babylonian talmud , shab .
75a _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
the above passage from the babylonian talmud of course represents a much later tradition : its interest in our context comes from the fact that it emphasizes the duty to calculate , as opposed to simply observe or record , astronomical events .
another element which also points to a calculational perspective is the observation that , in some manuscripts of the samaritan pentateuch , the word `` astrolabe '' appears in place of `` idols '' in the context of jacob s escape from laban s house . in our analysis of pentateuch chronology
a calculational - as opposed to observational - perspective , as could have been afforded at the time by the use of an astrolabe , will turn out to be essential .
as we shall strive to demonstrate , nearly all of the astronomical descriptions which are found in the text refer to solar eclipses at the northern ( also known as spring , vernal , or paschal ) equinox . in those descriptions ,
the sun typically represents the `` glory of god '' : directly , or symbolically through the current protagonist of the narrative or his firstborn .
the moon , which in a solar eclipse appears to partially or totally cover the sun , variously represents the `` hand of god '' , the hand of the protagonist , or that of a third party ; but also a rainment , the mouth of a well , or the egyptian bondage .
the ecliptic , along which the constellations of the zodiac are found , is a caravan circle , a bow , a sword , or a ladder .
eclipses which , from the point of view of an observer located in the near east , take place at sunrise or sunset are often used to identify a geographical feature at the visible horizon , for instance a town or mountain . in the northern hemisphere the vernal equinox
is closely associated with the renewal of life , and in several calendar systems of the near east marks the beginning of the new year . in particular , whenever it falls on a new moon , the vernal equinox also marks the beginning of the new year in the jewish religious calendar ( nisan 1 ) .
solar eclipses only occur with a new moon , and hence in the jewish calendar they may only take place at the beginning of each lunar month .
approximate alignments of the sun and the moon on the same calendar day follow the 19-year metonic cycle .
the cycle was presumably well known to the author of the pentateuch , as the nineteenth verse in genesis 1 is the conclusion of the fourth day , in which the sun and the moon are created .
also worth of mention , in the light of our proposed identification of the sun and the moon as the glory and the hand of god , respectively , is the opening of psalm 19 : `` the heavens declare the glory of god ; the skies proclaim the work of his hands '' .
the analysis will proceed as follows .
taking noah s flood as the starting point , and proceeding until the death of moses , we shall identify a sequence of passages in the text , and demonstrate that each passage can be associated to a specific solar eclipse , in such a way that : ( i ) all solar eclipses at the northern equinox in the reconstructed timelines of the narratives are accounted for ; ( ii ) all occurrences of the words `` appeared '' and `` covenant '' in the corresponding portions of the text are accounted for .
next , we shall turn to the genealogies of the pre - flood patriarchs , which unlike those of their post - flood counterparts widely differ across manuscripts . for each of the three main extant manuscript families ( masoretic , septuagint and samaritan ) we test the hypotesis that the corresponding chronologies are independent of eclipse events .
as we shall see , using a simple binomial model , while for the masoretic and septuagint chronologies independence can not be rejected at any interesting significance level , the chance that the dates in the samaritan chronology were set independently of eclipse events turns out to be lower than 1 in 10,000 .
the origins of western astronomy can be traced to mesopotamia [ 2 ] . in the earliest babylonian star catalogues , dating from about 1200 bce
, many star names appear in sumerian , suggesting a continuity reaching into the early bronze age . for convenience
, we shall sometimes express dates in astronomical notation .
in such notation the year -1000 corresponds to 1001 bce , reflecting the fact that there is no year 0 in the gregorian calendar .
we use stellarium 0.10.4 ( available as a free download at www.stellarium.org ) to reconstruct the position of the celestial bodies at specific times and places .
finally , we refer to all eclipses which admit a totality ( respectively , annularity ) path as total ( respectively , annular ) ; we refer to all other eclipses as partial . for each event
we also report whether , from the point of view of an observer located in the near east , it occurred during daytime or nighttime . in almost all cases ,
we conventionally take jerusalem as the reference point .
william ryan and walter pitman suggested in 1999 that several related near eastern flood stories , including noah s , could be associated with the sudden and perhaps violent flooding of the black sea region in the middle of the 6th millennium bce [ 3,4 ] .
the analysis of geologic and organic sediments on the black sea floor revealed ancient shorelines and deltas , and the abrupt disappearence around 5500 bce of freshwater mollusks , replaced by marine species . according to ryan and pitman , around that time the waters of the mediterranean
, whose level had been increasing by over 50 meters since the beginning of the holocene , spilled over the bosphorus and flooded the vast plains and shorelines surrounding what at the time was a large freshwater lake .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ gen 6:12 and god looked upon the earth , and , behold , it was corrupt [ _ lit .
_ : smeared ] ; for all flesh had corrupted [ _ lit .
_ : smeared ] his way upon the earth . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ gen 6:18 but with thee will i establish my covenant ; and thou shalt come into the ark , thou , and thy sons , and thy wife , and thy sons wives with thee . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ -5504/4/30 nighttime , partial eclipse at the northern equinox .
the sun in gemini ( the brotherhood of mankind , as the `` glory of god '' ) , following its path upon the surface of the earth , smears it with the shadow of the moon ( by its own hand ) .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the event takes place just before the flood , as noah begins to build the ark .
it announces , via the 19-year metonic cycle , another nighttime , total eclipse which takes place on -5485/5/1 , shortly after the end of the flood . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ gen 9:13 - 14 i do set my bow in the cloud , and it shall be for a token of a covenant between me and the earth . and it shall come to pass , when i bring a cloud over the earth , that the bow shall be seen in the cloud .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the bow in the cloud not only symbolizes the rainbow , but also the ecliptic : in which case , whenever the moon ( as the `` hand of god '' ) with its shadow brings a cloud over the earth , it also appears to touch the ecliptic ( the bow ) .
knowledge of the ecliptic allows for a lunisolar calendar , which is kept in tune by direct astronomical observation of the equinoxial points .
the additional knowledge of the metonic cycle also affords some ability to predict the occurrence of eclipses , which in turn , as the passage suggests , come to be regarded as a symbol of covenant . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ gen 9:23 and shem and japheth took a garment , and laid [ it ] upon both their shoulders , and went backward , and covered the nakedness of their father ; and their faces [ were ] backward , and they saw not their father s nakedness . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ -5420/5/3 total eclipse at the northern equinox .
the sun in gemini is covered by the moon .
the twins appear to walk backward , escorting the sun along the ecliptic .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the equinoxial points slowly rotate with respect to the constellations of the zodiac , returning to the same position approximately every 25,800 years . as a consequence , no other daytime ( in the near east ) , total eclipse at the paschal equinox in the last 30,000 years had the sun standing just in the middle of gemini .
note that , here and in other descriptions of astronomical events , the position of the relevant constellations was of course not directly observable as it was daytime , and could only be deduced through astronomical calculations .
the genealogies of the post - flood patriarchs , like those of their pre - flood predecessors , are comprised of two figures per patriarch .
[ cols= " < , < , < " , ]
josephus on manetho ( contra apionem , book 1 ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ manetho [ ... ] promised to interpret the egyptian history out of their sacred writings , and premised this : that `` our people had come into egypt , many ten thousands in number , and subdued its inhabitants ; '' and when he had further confessed that `` we went out of that country afterward , and settled in that country which is now called judea , and there built jerusalem and its temple . '' [ ... ] he mentions amenophis [ ... ] ; he then ascribes certain fabulous stories to this king [ ... ] .
`` this king was desirous to become a spectator of the gods , as had orus , one of his predecessors in that kingdom , desired the same before him ; he also communicated that his desire to his namesake amenophis , who was the son of papis , and one that seemed to partake of a divine nature , both as to wisdom and the knowledge of futurities . ''
manetho adds , `` how this namesake of his told him that he might see the gods , if he would clear the whole country of the lepers and of the other impure people ; that the king was pleased with this injunction , and got together all that had any defect in their bodies out of egypt ; and that their number was eighty thousand ; whom he sent to those quarries which are on the east side of the nile , that they might work in them , and might be separated from the rest of the egyptians .
'' he says further , that `` there were some of the learned priests that were polluted with the leprosy ; but that still this amenophis , the wise man and the prophet , was afraid that the gods would be angry at him and at the king , if there should appear to have been violence offered them ; who also added this further , [ out of his sagacity about futurities , ] that certain people would come to the assistance of these polluted wretches , and would conquer egypt , and keep it in their possession thirteen years ; that , however , he durst not tell the king of these things , but that he left a writing behind him about all those matters , and then slew himself , which made the king disconsolate . ''
after which he writes thus verbatim : `` after those that were sent to work in the quarries had continued in that miserable state for a long while , the king was desired that he would set apart the city avaris , which was then left desolate of the shepherds , for their habitation and protection ; which desire he granted them .
now this city , according to the ancient theology , was typho s city .
but when these men were gotten into it , and found the place fit for a revolt , they appointed themselves a ruler out of the priests of hellopolis , whose name was osarsiph , and they took their oaths that they would be obedient to him in all things .
he then , in the first place , made this law for them , that they should neither worship the egyptian gods , nor should abstain from any one of those sacred animals which they have in the highest esteem , but kill and destroy them all ; that they should join themselves to nobody but to those that were of this confederacy .
when he had made such laws as these , and many more such as were mainly opposite to the customs of the egyptians , he gave order that they should use the multitude of the hands they had in building walls about their city , and make themselves ready for a war with king amenophis , while he did himself take into his friendship the other priests , and those that were polluted with them , and sent ambassadors to those shepherds who had been driven out of the land by tefilmosis to the city called jerusalem ; whereby he informed them of his own affairs , and of the state of those others that had been treated after such an ignominious manner , and desired that they would come with one consent to his assistance in this war against egypt .
he also promised that he would , in the first place , bring them back to their ancient city and country avaris , and provide a plentiful maintenance for their multitude ; that he would protect them and fight for them as occasion should require , and would easily reduce the country under their dominion .
these shepherds were all very glad of this message , and came away with alacrity all together , being in number two hundred thousand men ; and in a little time they came to avaris . and
now amenophis the king of egypt , upon his being informed of their invasion , was in great confusion , as calling to mind what amenophis , the son of papis , had foretold him ; and , in the first place , he assembled the multitude of the egyptians , and took counsel with their leaders , and sent for their sacred animals to him , especially for those that were principally worshipped in their temples , and gave a particular charge to the priests distinctly , that they should hide the images of their gods with the utmost care .
he also sent his son sethos , who was also named ramesses , from his father rhampses , being but five years old , to a friend of his .
he then passed on with the rest of the egyptians , being three hundred thousand of the most warlike of them , against the enemy , who met them .
yet did he not join battle with them ; but thinking that would be to fight against the gods , he returned back and came to memphis , where he took apis and the other sacred animals which he had sent for to him , and presently marched into ethiopia , together with his whole army and multitude of egyptians ; for the king of ethiopia was under an obligation to him , on which account he received him , and took care of all the multitude that was with him , while the country supplied all that was necessary for the food of the men .
he also allotted cities and villages for this exile , that was to be from its beginning during those fatally determined thirteen years .
moreover , he pitched a camp for his ethiopian army , as a guard to king amenophis , upon the borders of egypt .
and this was the state of things in ethiopia . [ ... ] it was also reported that the priest , who ordained their polity and their laws , was by birth of heliopolis , and his name osarsiph , from osyris , who was the god of heliopolis ; but that when he was gone over to these people , his name was changed , and he was called moses . ''
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ i will inquire into what cheremon says . for he
also , when he pretended to write the egyptian history , sets down the same name for this king that manetho did , amenophis , as also of his son ramesses , and then goes on thus : `` the goddess isis appeared to amenophis in his sleep , and blamed him that her temple had been demolished in the war . but that phritiphantes , the sacred scribe , said to him , that in case he would purge egypt of the men that had pollutions upon them , he should be no longer troubled .
with such frightful apparitions . that amenophis accordingly chose out two hundred and fifty thousand of those that were thus diseased , and cast them out of the country : that moses and joseph were scribes , and joseph was a sacred scribe ; that their names were egyptian originally ; that of moses had been tisithen , and that of joseph , peteseph : that these two came to pelusium , and lighted upon three hundred and eighty thousand that had been left there by amenophis , he not being willing to carry them into egypt ; that these scribes made a league of friendship with them , and made with them an expedition against egypt : that amenophis could not sustain their attacks , but fled into ethiopia , and left his wife with child behind him , who lay concealed in certain caverns , and there brought forth a son , whose name was messene , and who , when he was grown up to man s estate , pursued the jews into syria , being about two hundred thousand , and then received his father amenophis out of ethiopia . ''
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the names osarseph ( osiris - seph ) and peteseph ( ptah - seph ) are related to joseph through the association of osiris and ptah with the moon , and the phonetic analogy of iah ( the egyptian word for moon ) with jah .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` i have heard of the ancient men of egypt , that moses was of heliopolis , and that he thought himself obliged to follow the customs of his forefathers , and offered his prayers in the open air , towards the city walls ; but that he reduced them all to be directed towards sun - rising , which was agreeable to the situation of heliopolis ; that he also set up pillars instead of gnomons , under which was represented a cavity like that of a boat , and the shadow that fell from their tops fell down upon that cavity , that it might go round about the like course as the sun itself goes round in the other . ''
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ | some of the narratives in the pentateuch can be associated with known astronomical events to provide absolute dates for biblical chronology . |
one of the most compelling puzzles in present day astronomy is the question how galaxies formed . in particular , we need to understand the wide variety of sizes , masses and morphologies of galaxies observed , as well as their dynamics , ages and metallicities . in addition , we need to understand the origin of the scaling relations , such as the fundamental plane relations for ellipticals and the tully - fisher relation for spirals , as well as the so - called density - morphology relation ( dressler 1980 ) .
the latter shows that the more compact galaxies , such as ellipticals and s0s , are preferentially found in overdense regions such as galaxy clusters .
the classical way of depicting the different morphologies of galaxies is by means of the hubble diagram , which reveals the gross distinction of three sorts of galaxies : spirals , s0s , and ellipticals .
several studies over the past years have shown that elliptical galaxies can be roughly divided in two subclasses : the rotation supported , low luminosity disky ellipticals and the pressure supported , bright , boxy ellipticals ( e.g. , bender 1988 ; bender et al .
1989 ; capaccioli , caon & rampazzo 1990 ) .
this dichotomy has recently been strengthened by properties observed in their central regions ( e.g. , nieto , bender & surma 1991 ; jaffe et al .
1994 ; ferrarese et al . 1994 ; gebhardt et al . 1996
; faber et al .
1997 ) . based on this dichotomy amongst elliptical galaxies
, it has been suggested that the classical hubble diagram should be revised ( kormendy & bender 1996 ) , to use a more physical subdivision of the class of the ellipticals , rather than the apparent flattening .
kormendy & bender proposed to use the velocity anisotropy which is well measured by the isophote shapes ( bender 1988 ; bender et al .
1989 ) .
the main morphological parameter that sets the classification of galaxies in the ( revised ) hubble diagram is the disk - to - bulge ratio @xmath2 .
understanding the origin of the hubble sequence is thus intimately related to understanding the parameters and processes that determine the ratio between the masses of disk and bulge .
the diskiness of the low luminosity ellipticals is generally interpreted as due to an embedded disk .
these disks , which we term ` embedded disks ' in the following , are smaller and brighter than disks in s0s and spirals ( scorza & bender 1995 ) , and it has been emphasized many times that the disky ellipticals build a continuous sequence with s0 galaxies ( capaccioli 1987 ; carter 1987 ; bender 1988 , 1990 ; van den bergh 1990 : capaccioli et al . 1990 ; rix & white 1990 ; capaccioli & caon 1992 ; bender et al .
1993 ; scorza & bender 1995 , 1996 ; scorza & van den bosch 1998 ) .
this continuity suggests similar formation histories , whereby one or several parameters of the proto - galaxy vary smoothly .
disk dominated systems such as spirals are believed to have formed by cooling of the baryonic matter inside a virialized dark halo . as the gas cools , its specific angular momentum is conserved , and the amount of angular momentum of the dark halo thus determines the size of the disk ( fall & efstathiou 1980 ; see section 2 for more details on this disk - formation scenario ) . the formation of bulge dominated disk systems is far less clear .
the aim of this paper is to investigate to what extent the disky ellipticals and s0s can be incorporated in the fall & efstathiou theory for the formation of galactic disks , by incorporating a simple picture for the formation of the bulge .
we envision an inside - out formation scenario for the bulge .
it is assumed that the bulge forms out of the low - angular momentum material in the halo , which cools and tries to settle into a small , compact disk .
such disks are however unstable , and we assume here that this instability , coupled with the continuous supply of new layers of baryonic matter that cool and collapse , forms the bulge .
this inside - out bulge formation is self - regulated in that the bulge grows until it is massive enough to allow the remaining gas to form a stable disk component .
we do not describe the bulge formation in any detail but merely use empirical relations of the characteristic structural parameters of bulges and ellipticals to describe the end result of the formation process as a realistic galaxy .
we use this simple formation scenario to investigate the predicted disk - to - bulge ratios and disk scale - lengths as a function of the halo angular momentum , and as a function of formation redshift and cosmology .
the main focus of this paper is to investigate whether this inside - out formation scenario can account for two orders of magnitude variation in disk - to - bulge ratio , required in order to incorporate a wide variety of disk - bulge systems : from late - type spirals with @xmath3 to s0s ( @xmath4 ) to disky ellipticals ( @xmath4 ) .
this paper is organized as follows . in section 2
we describe the formation scenario , the galaxy formation efficiency , and the stability criterion for our disk - bulge - halo models . in section 3
we use these models to investigate the position of different sorts of disks in the parameter space of disk central surface brightness versus disk scale - length . in section 4
we discuss to what extent s0s and disky ellipticals can be incorporated in this formation scenario , and what parameters may be responsible for the origin of ( the major part of ) the hubble sequence . in section 5
we briefly discuss alternative formation scenarios for disky ellipticals .
our results are summarized and discussed in section 6 .
a remarkably successful model for the formation of galaxies is the white & rees ( 1978 ) theory , wherein galaxies form through the hierarchical clustering of dark matter and subsequent dissipational settling of gaseous matter within the dark halo cores . coupled with the notion of angular momentum gain by tidal torques induced by nearby proto - galaxies ( hoyle 1953 ; peebles 1969 )
, the white & rees theory provides the background for a model for the formation of galactic disks ( fall & efstathiou 1980 ; faber 1982 ; gunn 1982 ; fall 1983 ; van der kruit 1987 ) . in this model ,
disks form through the collapse of a uniformly rotating proto - galaxy . owing to the non - dissipative character of the dark matter , its collapse halts when the system virializes . at regions where the baryonic density is high enough
, the baryons can cool and decouple from the dark matter .
the angular momentum of the baryons , acquired from tidal torques from nearby proto - galaxies , prevents the collapse from proceeding all the way to the center , and causes the baryons to settle in a rapidly rotating disk .
this simple picture of disk formation yields realistic disk sizes and surface brightness profiles provided that there is little angular momentum transfer from the baryonic matter to the dark halo , and that the dark halo has a density profile such that the resulting galaxy has a flat rotation curve . fall ( 1983 )
extended the original fall & efstathiou disk formation scenario by placing it in a cosmological context , and by deriving freeman s law ( freeman 1970 ) and the tully - fisher relation .
recently , two papers reexamined this disk formation model and included the effects of halo - contraction due to the accumulation of baryonic matter in the center of the galaxy : dalcanton , spergel & summers ( 1997 ) showed that the disk formation model outlined above can not only explain the properties of normal , high surface brightness ( hsb ) disks , but also those of the class of low surface brightness ( lsb ) disks ( see also jimenez et al .
1998 ) ; mo , mao & white ( 1998 ) compared the outcome of disks in different cosmologies , and used this to investigate the effects of disk evolution and the predicted population of ly@xmath5 absorbers in qso spectra .
the assumptions we make are similar to those made by dalcanton et al .
( 1997 ) and mo et al .
( 1998 ) , but we extend upon their work by including bulges .
we investigate how the disky ellipticals and s0s may be fitted into this particular formation scenario , by comparing the differences in properties of the proto - galaxies that result in spirals , s0s , and disky ellipticals .
the turn - around , virialization , and subsequent dissipational settling of the baryonic matter of a proto - galaxy is an inside - out process .
first the innermost shells virialize and heat its baryonic material to the virial temperature .
because of the relatively high density and low angular momentum of this gas , it will rapidly cool and ( try to ) settle in a very small , compact disk .
however , self - gravitating disks are violently unstable and form a bar ( e.g. , hohl 1971 ; ostriker & peebles 1973 ) . meanwhile the next shell of the proto - galaxy cools and tries to settle into a disk structure at a radius determined by its angular momentum .
the resulting structure in the center of the dark halo consequently becomes more and more unstable .
bars are efficient in transporting gas inwards ( e.g. , friedli & martinet 1993 ; wada & habe 1995 ) , and can cause vertical heating by means of a collective bending instability ( e.g. , combes et al .
1990 ; pfenniger 1984 ; pfenniger & friedli 1991 ; raha et al .
both these process lead ultimately to the dissolution of the bar ; first the bar takes a hotter and triaxial shape , but is later on transformed in a spheroidal bulge component ( e.g. , combes et al .
1990 ; pfenniger & norman 1990 ; pfenniger & friedli 1991 ; pfenniger 1993 ; friedli 1994 ; martinet 1995 ; norman , sellwood & hasan 1996 ) . here
we assume that this bar instability and subsequent dissolution , coupled with the continuous supply of new layers of baryonic matter that cool and collapse , forms the bulge .
the numerical simulations performed by the numerous studies listed above , have been able to show that bars can build small and close to exponential bulges ( see also courteau , de jong & broeils 1996 for observational evidence for bulge formation out of disks ) .
however , no simulations have been able to use bar instabilities to produce very large bulges or even ellipticals . on the other hand
, all these simulations started from a normal , marginally unstable disk , whereas the process we are describing starts from a highly unstable , compact disk , includes the continuous supply of baryonic matter from prolonged cooling flows , and is likely to have additional heating sources due to the star formation and feedback processes which are expected to take place in this phase of the bulge formation .
detailed numerical simulations that take all these processes into account are required in order to investigate whether this can indeed lead to massive bulges as assumed here .
finally , we emphasize that the high densities of the inner shells yield cooling time - scales that may be significantly shorter than the dynamical time - scale .
this can add to the inside - out formation of a bulge component , as suggested by kepner ( 1997 ) .
we assume that the bulge formation process discussed above continuous until the bulge has become massive enough such that the subsequent layers of baryonic material can cool and form a disk which is stable against bar - formation .
the process of disk - bulge formation is thus a self - regulating one in that the bulge grows until it is massive enough to sustain the remaining gas in the form of a stable disk . since the stability is directly related to the amount of angular momentum of the gas ( see section 2.4 ) , we expect a clear correlation between the disk - to - bulge ratio and the angular momentum of the dark halo out of which the galaxy is assembled
. we will show that this correlation is present , and that it nicely follows the marginal stability curve as expected from the self - regulating mechanism proposed here .
the galaxies under investigation in this paper , varying from spirals to disky ellipticals , are assumed to consist of a disk plus bulge embedded in a dark halo .
virialized dark halos can be quantified by a mass @xmath6 , a radius @xmath7 , and an angular momentum @xmath8 ( originating from cosmological torques ) .
spherical collapse models ( e.g. , gunn & gott 1972 ) show that virialized halos have an over - density of approximately 200 . in a recent series of papers navarro , frenk & white ( 1995 ; 1996 ; 1997 , hereafter nfw ) showed that virialized dark halos have universal equilibrium density profiles , independent of mass or cosmology ( but see moore et al .
1997 ) , which can be well fit by @xmath9 where @xmath10 with @xmath11 the concentration parameter , @xmath12 a scale radius , and @xmath13 the critical density for closure .
the radius @xmath7 is defined as the limiting radius of a virialized halo , and corresponds to the radius within which the mean density is @xmath14 .
throughout we only consider spherical halos .
rather than specifying the halo by its mass @xmath6 , it is customary to specify it by its circular velocity @xmath15 .
given the mass @xmath6 , the formation redshift @xmath16 , and the specific cosmology , the concentration parameter @xmath17 can be calculated using the procedure outlined in appendix a of navarro , frenk & white ( 1997 ) . in this paper
we compare two different cosmologies : a standard cold dark matter cosmology ( scdm ) with a total matter density of @xmath18 , a hubble constant of @xmath19 with @xmath20 , and with a rms linear over - density at @xmath21 in spheres of radius @xmath22 mpc of @xmath23 .
this latter parameter normalizes the cdm power spectrum .
the second cosmology discussed here is an open cold dark matter model ( ocdm ) with @xmath24 , @xmath25 , and @xmath26 .
both models have a zero cosmological constant and the parameters are typical of those favored by large - scale structure constraints .
the baryon density is set to be @xmath27 , in agreement with the nucleosynthesis constraints ( e.g. , walker et al .
1991 ) . the baryonic mass fraction is then given by @xmath28 .
we assume the final system to have a disk - to - bulge ratio @xmath29 , such that the masses of disk , @xmath30 , and bulge , @xmath31 , can be written as @xmath32 and @xmath33 here @xmath34 is the galaxy formation efficiency , expressed as the fraction of the available baryons that ultimately end up in the galaxy ( see section 2.3 ) . as mentioned in section 2.1 , we assume that the bulge is formed out of the low - angular momentum material .
furthermore , we assume that the material that builds the disk does not loose its specific angular momentum . under these conditions we have that @xmath35 , where @xmath36.\ ] ] the function @xmath37 depends on the disk - to - bulge ratio and on the detailed physics that describe how the fraction @xmath34 of the baryonic material ends up in the disk - bulge system ( see section 2.3 ) .
this function is derived in the appendix . in the following we parameterize @xmath8 by the dimensionless spin parameter @xmath38 , defined by @xmath39 here @xmath40 is the halo s total energy .
several studies , both analytical and numerical , have shown that the distribution of spin angular momentum of collapsed dark matter halos is well approximated by a log normal distribution : @xmath41 ( e.g. , barnes & efstathiou 1987 ; ryden 1988 ; cole & lacey 1996 ; warren et al
. 1992 ) .
we consider a distribution that peaks around @xmath42 with @xmath43 .
these values are in good agreement with the @xmath44-body results of warren et al .
( 1992 ) .
we assume the disk to be an infinitesimally thin exponential disk , with surface brightness @xmath45 and total luminosity @xmath46 in order to convert the disk parameters to an observers frame we need to assume a mass - to - light ratio @xmath47 .
bottema ( 1997 ) found @xmath48 for @xmath49 for hsb disks . using @xmath50 ( e.g. , de jong 1996b ) and
@xmath51 this yields @xmath52 in the @xmath53-band .
although bulges and elliptical galaxies are reasonably well fit by a @xmath54 ` de vaucouleurs ' law , it has recently been shown that a better fit is obtained by using a generalized @xmath55 law .
this fitting function , introduced by sersic ( 1968 ) , has the form @xmath56,\ ] ] where @xmath57 is the central surface brightness , @xmath58 is the effective radius , and @xmath59 is the exponential variable . for @xmath60 one has a pure exponential , while for @xmath61 one recovers the de vaucouleurs profile .
the parameter @xmath62 is determined by the requirement that @xmath58 is the radius encircling half of the total luminosity and is very well approximated by @xmath63 ( ciotti 1991 ) .
the total luminosity of a system with a @xmath55 luminosity profile ( assuming sphericity ) is given by @xmath64 where @xmath65 is the gamma function .
it was found that the best fitting exponent @xmath59 is correlated with the total luminosity of the spheroid : andredakis , peletier & balcells ( 1995 ) used the @xmath55 law to fit the luminosity profiles of bulges of spiral galaxies , and found the effective radius @xmath58 to be related to the total luminosity of the bulge by the empirical relation @xmath66 in addition , caon , capaccioli & donofrio ( 1993 ) found that the exponent @xmath59 is correlated with the effective radius through the empirical relation @xmath67 although this empirical relation has only been shown to fit early - type galaxies , we assume here that it is also valid for bulges .
we take @xmath59 to be limited between @xmath60 ( exponential ) and @xmath61 ( de vaucouleurs ) .
none of our results are significantly influenced by this assumption however .
the main parameter for the bulge is its total mass ; changes in its actual density distribution are only a second order effect .
the relations ( [ sigbulge ] ) ( [ nbulge ] ) allow us to describe the bulges by a single parameter ; once the total luminosity of the bulge is known , we can use the empirical relations to determine all other bulge parameters !
van der marel ( 1991 ) found that ellipticals have @xmath68 ( johnson @xmath69 ) . using @xmath70 ( e.g. , van der marel 1991 ) and
@xmath71 this yields @xmath72 in the @xmath53-band . for our combined disk - bulge model
we assume that the mass - to - light ratio is constant over the entire galaxy with a value equal to the luminosity weighted average of the mass - to - light ratios of the disk and bulge .
alternatively , we could have assigned the bulge and disk separate mass - to - light ratios of @xmath47 and @xmath73 respectively , but the mass - to - light ratio of the disks in s0s and disky ellipticals is clearly larger than for spiral discs . assuming a constant , luminosity weighted ratio is thus more realistic .
once the baryons inside the virialized dark halo start to cool , this induces a contraction of the inner region of the dark halo .
we follow blumenthal et al .
( 1986 ) and flores et al .
( 1993 ) and assume that the halo responds adiabatically to the slow assembly of the baryonic matter in disk and bulge , and that it remains spherical ( see also dalcanton et al .
1997 and mo et al .
1998 ) . under these assumptions
the angular action of the dark halo material is an adiabatic invariant , such that a particle initially at a radius @xmath74 ends up at mean radius @xmath75 according to @xmath76 here @xmath77 is the initial mass distribution ( the nfw profile ) given by @xmath78,\ ] ] with @xmath79 , and @xmath80 is the final mass distribution given by @xmath81 here @xmath82,\ ] ] and @xmath83 { { \rm d}x \over \sqrt{x^2 - y^2}},\ ] ] with @xmath84 .
mo et al . (
1998 ) have shown that the scale - length of the disk can be written as @xmath85 where @xmath86 a parameter that indicates the amount of self - gravity of the disk , and @xmath87 ^ 2}.\ ] ] the circular velocity @xmath88 is given by @xmath89 with @xmath90 @xmath91 as given by equation [ 2.196 ] in binney & tremaine ( 1987 ) , and @xmath92 \over r}.\ ] ] we can solve this set of equations in an iterative way , following the procedure used by mo et al .
( 1998 ) . for a given @xmath6 , @xmath38 , @xmath93 and @xmath94
we first calculate the concentration parameter @xmath17 of the dark halo before collapse of the baryons .
we then calculate the parameters of the bulge using equations ( [ abulge ] ) ( [ nbulge ] ) .
next we start with a guess for @xmath95 , by setting @xmath96 , and we calculate the disk parameters @xmath97 and @xmath98 using equations ( [ rdlambda ] ) and ( [ ldisk ] ) respectively
. subsequently we solve for @xmath74 as a function of @xmath75 using equations ( [ initialrad ] ) ( [ bulmassfunc ] ) , and thus obtain @xmath80 .
we then calculate @xmath99 , and substitute this in equation ( [ frint ] ) to obtain a new value for @xmath95 .
convergence is rapidly achieved : 3 to 5 iterations yield an accuracy of @xmath97 already better than one percent .
since we have defined the mass of the dark halo such that the average over - density of the halo is 200 times the critical density for closure at the virialization redshift @xmath100 , we have that @xmath101 this halo mass can be related to the disk luminosity ( assuming no bulge , i.e. , @xmath102 ) , by @xmath103 ( cf .
equation [ [ dismass ] ] ) . combining the above equations yields @xmath104 this is similar to the tully - fisher relation : @xmath105 ,
whereby the zero - point , @xmath106 , is inversely proportional to @xmath107 .
however , current data suggest no zero - point evolution ( vogt et al . 1996 , 1997 ;
mao , mo & white 1997 ) .
in other words , if we assume a significant scatter in the formation redshifts of spiral galaxies ( some massive spirals are already in place at @xmath108 , see vogt et al .
1996 ) , the dependence of equation ( [ modtf ] ) would yield a scatter in the tully - fisher relation much larger than observed : if spirals form between @xmath21 and @xmath109 ( and they are not significantly modified since their formation ) a scatter of @xmath110 mag is predicted for an einstein - de sitter universe , much larger than the observed scatter of @xmath111 mag .
this suggests that the redshift dependence of the tully - fisher zero - point has to be corrected for by either imposing @xmath112 or @xmath113 .
we consider the former option unlikely , since this predicts spirals at @xmath114 to be much bluer than observed ( see vogt et al .
we therefore consider a scenario in which the galaxy formation efficiency is redshift dependent .
combining the observed tully - fisher relation with equation ( [ modtf ] ) yields @xmath115 where @xmath116 . using the tully - fisher relation of giovanelli et al .
( 1997 ) , converted to the @xmath53-band using @xmath117 as an average for spirals ( de jong 1996b ) , yields @xmath118 ( i.e. , giovanelli et al .
find @xmath119 ) .
here we assumed that the hi - linewidth equals twice the circular velocity @xmath120 .
the important factor here is @xmath121 which shows that the galaxy formation efficiency depends on the specific cosmology and the formation redshift .
here @xmath122 , with @xmath123 the cosmological constant . in the scdm model with @xmath20 and @xmath18 ,
equation ( [ diskefb ] ) yields @xmath124 .
galaxies that were assembled at higher redshift thus formed more efficiently .
more efficiently here means that a larger fraction of the baryonic matter available in the dark halo actually made it into the disk - bulge system . at @xmath21
this fraction is approximately forty percent , and this increases to hundred percent at @xmath125 .
halos of a given mass are denser at higher redshift , and higher density implies higher escape velocity and higher cooling efficiency , both of which could give a plausible explanation for an increase of the galaxy formation efficiency with increasing redshift . in what follows
we consider two extreme cases of this galaxy formation ( in)efficiency . in the first scenario , which we call the ` cooling'-scenario , we assume that only the inner fraction @xmath34 of the baryonic mass is able to cool and form the disk - bulge system : the outer parts of the halo , where the density is lowest , but which contain the largest fraction of the total angular momentum , never gets to cool . in the second scenario , referred to hereafter as the ` feedback'-scenario , we assume that the feedback from star formation and evolution is such that each baryon in the dark halo , independent of its specific angular momentum , has a probability of @xmath34 to ultimately end up as a constituent of the disk or bulge .
the choice of either of these scenarios enters our equations in the function @xmath37 ( see the appendix ) .
as is well known , disks that are self - gravitating tend to be unstable against bar formation . halos with smaller @xmath38 yield disks with smaller scale - lengths ( see equation [ [ rdlambda ] ] ) . for a given disk - to - bulge ratio ,
there is thus a critical @xmath38 such that for @xmath126 the disk will be bar unstable .
christodoulou , shlosman & tohline ( 1995 ) derived a stability criterion for bar formation based on the angular momentum content of the disk , rather than the energy content ( such as the well known ostriker & peebles ( 1973 ) criterion ) . for a disk - bulge - halo system
this stability criterion can be written such that disks are stable against bar formation if @xmath127 with @xmath128 and @xmath129 the characteristic circular velocities of the disk and the composite disk - bulge - halo system ( equation [ [ vctot ] ] ) , respectively .
this criterion is similar to the one proposed by efstathiou , lake & negroponte ( 1982 ) . as characteristic velocities we consider the circular velocities at @xmath130 , since mo et al .
( 1998 ) have shown that this is the radius where typically the rotation curve of the composite system reaches a maximum . for a thin exponential disk , approximately 80 percent of the disk mass
is located inside @xmath131 , and @xmath132 substituting equation ( [ vdapprox ] ) in ( [ stabalpha ] ) , and using equation ( [ rdlambda ] ) we find that disks are stable if the spin parameter of their halos obeys @xmath133 with @xmath134 this reduces to @xmath135 for stellar disks and @xmath136 for gaseous disks , using @xmath137 and @xmath138 , respectively ( see christodoulou et al .
this is in excellent agreement with the empirical value of @xmath139 found by efstathiou et al .
( 1982 ) based on @xmath44-body experiments . throughout
we take @xmath140 to take into account that disks are often made up of a mix of gas and stars .
the critical value of @xmath38 given by equation ( [ stablambda ] ) depends on the galaxy formation efficiency @xmath34 , and thus on the specific cosmology and formation redshift according to equation ( [ hubconst ] ) .
furthermore , since @xmath141 , and @xmath142 is different for the cooling and feedback scenarios , @xmath143 depends on the particular model we choose for the galaxy formation ( in)efficiency . in figure 1 this dependence is shown for three different cosmologies : scdm , ocdm , and @xmath123cdm .
the latter is a model with a non - zero cosmological constant with @xmath24 , @xmath144 , @xmath25 and @xmath145 .
results are shown for both the feedback scenario ( left panel ) and the cooling scenario ( right panel ) .
figure 1 plots the logarithm of the critical spin parameter versus redshift .
the dotted lines outline the probability distribution of the spin parameter ( equation [ [ spindistr ] ] ) ; the indices ( indicated in the right panel only ) denote the probability of finding a halo below that line . at low formation redshifts @xmath143 increases with redshift , due to the increasing galaxy formation efficiency with redshift . the turnovers to a flat line at higher redshift are caused by the requirement that @xmath146 .
note that in a scdm cosmology with @xmath147 ( for which @xmath148 ) about 50 percent of all halos yield unstable disks , and thus require the formation of a bulge component in order to stabilize the disk .
this increases to @xmath149 percent in models with @xmath150 .
furthermore , it is apparent that @xmath143 is much less redshift dependent in the case of the cooling scenario , than in the case of the feedback scenario .
the implications of this are discussed in section 4 . as can be seen , the results are very similar for the two low @xmath151 models . since @xmath143 is the important parameter for what follows , we only focus on a comparison of scdm with ocdm cosmologies , and note that results for the @xmath123cdm model are similar as for the ocdm model .
it has recently been argued that the results of navarro , frenk & white concerning the universal density profiles of dark halos may be incorrect .
very high resolution simulations by moore et al .
( 1997 ) indicate that the central density cusp of dark halos may be even steeper than @xmath152 .
we therefore investigate the sensitivity of our results to the assumption of a nfw halo profile by comparing our results with those obtained considering isothermal halo profiles , i.e. , halos with a density distribution given by @xmath153 for such halo profiles the circular velocity , @xmath129 , is independent of radius and equal to the circular velocity at the virial radius , @xmath120 .
the results for the isothermal density profiles can be obtained from the equations in this paper by taking @xmath154 .
we have rerun all models with an isothermal halo , and compared the results with those of the nfw halo profile .
none of the results presented in this paper are significantly influenced by the assumption of a nfw halo profile .
one of the strongest suggestions for a smooth transition from spirals and s0s towards disky ellipticals along a sequence of decreasing disk - to - bulge ratio is provided by the @xmath155@xmath97 diagram , in which the central surface brightness of disks , @xmath155 , is plotted against the disk s scale - length @xmath97 ( see scorza & bender 1995 ) .
scorza & van den bosch ( 1998 ) investigated the continuity of disk properties of a variety of disk galaxies and extended the diagram towards much higher @xmath155 and smaller @xmath97 by including the nuclear disks discovered from hubble space telescope photometry in a number of early - type galaxies ( van den bosch et al . 1994 ; kormendy et al . 1996 ; van den bosch , jaffe & van der marel 1997 ) .
this extended @xmath155@xmath97 diagram , already published in scorza & van den bosch ( 1998 ) , is shown in figure 2 .
it plots the central surface brightness ( in @xmath53-band ) as a function of the logarithm of the scale - length of a vast variety of disks , ranging from the extraordinary large disk of malin i ( data taken from bothun et al .
1987 ) , to the very small nuclear disks ( data taken from scorza & van den bosch 1998 ) .
in addition , figure 2 plots the parameters of embedded disks ( solid circles , data taken from scorza & bender 1995 and scorza et al .
1998 ) , and of a combined sample of s0s ( stars ) , hsb disks ( open circles ) , and lsb disks ( triangles ) .
data are taken from kent ( 1985 ) , de jong ( 1996a ) , sprayberry et al .
( 1995 ) , de blok , van der hulst & bothun ( 1995 ) , and mcgaugh & bothun ( 1994 ) .
figure 2 combines the parameters of disks that span 4 orders of magnitude in both scale - length and central surface brightness !
although there is considerable ` scatter ' , the @xmath156@xmath97 diagram of figure 2 seems to reveal an almost linear relation .
it is tempting to believe that this may teach us something about the formation of galactic disks .
however , it should be noted that selection effects play an important role at the lower left part of the diagram : most of the spirals and s0s in figure 2 are taken from large diameter limited surveys , and are thus strongly biased towards the largest scale - lengths .
furthermore , one is always strongly biased against low surface brightness disks , especially if they are embedded in a much brighter spheroid .
even more , the fact that selection effects are different for the various data sets combined in figure 2 prohibits a statistical conclusion on this lower left part of the diagram ( e.g. , see detailed discussions in de jong 1996b and dalcanton et al .
however , the absence of disks in the upper right part of the diagram is real , and should in principle be related to the formation and evolution of disk systems .
we now use our disk - bulge - halo models to investigate the locations of the different sorts of disks in the @xmath156@xmath97 diagram .
first we note that spirals and s0s with the highest central surface brightness for given scale - length lie approximately along a line of constant disk luminosity .
part of the upper boundary in the @xmath156@xmath97 diagram thus seems related to a cut - off in disk luminosity .
this is to be expected from the cut - off in the galaxy luminosity function , which is well described by the schechter ( 1976 ) function .
this tells us that there are only very few galaxies with luminosities exceeding @xmath157 .
we thus assume that the upper boundary is a line along which the sum of bulge and disk luminosities is equal to @xmath157 .
the dotted line in figure 2 ( which is partly drawn as a solid thick line ) corresponds to @xmath158 .
the position of a disk along a line of constant @xmath159 is determined by the spin parameter @xmath38 : disks originating from halos with smaller @xmath38 have smaller scale - lengths , and higher central surface brightness . as shown in section 2.4
, there is a critical spin parameter such that for @xmath126 disks are unstable .
the disk for which @xmath158 and @xmath160 is indicated by a thick solid dot in figure 2 .
if we want to build disks with scale - lengths smaller than this , we need a bulge to help stabilize the disk . the curved thick solid line , that originates at the thick solid dot and extends towards smaller scale - lengths , corresponds to a line with @xmath161 , @xmath162 , and with a disk - to - bulge ratio @xmath94 running from @xmath102 ( big solid dot ) towards @xmath163 .
the dashed line is the line with @xmath162 , @xmath102 , and @xmath159 running from @xmath157 ( big solid dot ) to @xmath164 . using these limits on the expected scale - lengths and central surface brightnesses of disks in our disk - bulge - halo models
, we can distinguish 4 different regions in the @xmath156@xmath97 diagram , labeled i to iv in figure 2 .
region i is the region where the disk luminosity exceeds @xmath157 .
region ii is the region where @xmath165 and @xmath166 ; disks in this region are stable against bar formation even without a bulge .
disks in region iii , however , require a bulge to be stable .
region iv , finally , is the region where systems with @xmath167 have an unstable disk despite the presence of the bulge .
the position of the big solid dot , and therewith of the boundary between regions ii and iii , depends strongly on the formation redshift and cosmology ( cf .
figure 1 ) .
figure 2 is based on a scdm cosmology with zero formation redshift .
although quite a number of assumptions have gone into our disk - bulge - halo models , the upper boundary ( indicated by the thick solid line ) nicely separates the region without disks , from the region most densely occupied by disks .
of course , the fact that the number density is highest just below the boundary is an observational bias effect as discussed above .
one has to keep in mind that the @xmath155@xmath97 diagram can be rather confusing since there are _ three _ parameters that determine the position of a disk in this diagram , namely the disk luminosity , the spin parameter , and the disk - to - bulge ratio . in the following we therefore focus on another diagram in which the luminosity dependence has been removed .
we now ask ourselves whether we can learn something about the formation of disks and bulges from a comparison of our disk - bulge - halo models with real galaxies .
we do this by calculating , for each galaxy in our combined sample , the spin parameter @xmath38 of the dark halo which , under a given set of assumptions , yields the observed disk properties ( scale - length and central surface brightness ) .
we thus use our formation scenario to link the _ disk _ parameters to those of the dark halo , and use the statistical properties of dark halos to discriminate between different cosmogonies .
we start by choosing a cosmology and a formation redshift and we read in the disk - to - bulge ratio , @xmath94 , and the scale - length and central surface brightness of the disk . from this
we determine the luminosities of the disk and bulge , which , upon using equations ( [ dismass ] ) , ( [ bulmass ] ) , and ( [ diskefb ] ) , yield the total mass @xmath6 and the particular density distribution of the disk - bulge - halo system .
we then calculate @xmath38 from equation ( [ rdlambda ] ) and plot this against the disk - to - bulge ratio @xmath94 .
this @xmath38 is the value of the spin parameter that the halo would have had if the observed disk - bulge system formed out of that halo in the way envisioned here .
one of the advantages of this diagram is that the dependence on luminosity is eliminated : the position of a disk in this diagram is virtually independent of the disk s total luminosity . in figures 3
( scdm ) and 4 ( ocdm ) we plot the location of disks of lsb spirals ( triangles ) , hsb spirals ( open circles ) , s0s ( stars ) , and disky ellipticals ( solid circles ) in a @xmath168 $ ] vs. @xmath169 $ ] diagram .
the five small - dashed lines in figures 3 and 4 correspond to the value of @xmath38 below which we expect @xmath170 , @xmath171 , @xmath172 , @xmath173 , and @xmath174 percent of the halos for a distribution function of spin parameters given by equation ( [ spindistr ] ) with @xmath42 and @xmath43 ( see also figure 1 ) .
these lines are drawn to guide the eye .
if our assumptions for the formation of disk - bulge - halo systems are correct , the distribution of inferred halo spin parameters should be similar to this log normal distribution .
the thick solid lines in figures 3 and 4 correspond to @xmath160 ( equation [ [ stablambda ] ] ) .
disks that lie below this line are unstable against bar formation .
this critical value of the spin parameter has been calculated for a halo with a total mass of @xmath175 , but is virtually identical for other halo masses .
the crowding of disks at @xmath176 is due to the fact that we have assigned a disk - to - bulge ratio of ten to systems for which no value of @xmath94 was determined ( these are mainly a number of the lsb spirals ) .
the panels on the left correspond to @xmath177 , the panels in the middle to @xmath178 , and the panels on the right to @xmath179 .
the upper three panels correspond to models in which @xmath180 , i.e. , in which _ all _ the baryons in the dark halo make it into the galaxy
. the panels in the middle correspond to the cooling scenario , and the lower panels to the feedback scenario .
the main difference between the upper and the middle and lower panels is that the former have @xmath181 independent of formation redshift ( there is a non - zero but negligible dependence on @xmath16 left due to the fact that the concentration parameter @xmath17 of the nfw halo profiles depends on formation redshift ) , whereas @xmath143 increases with @xmath16 in the middle and lower panels ( according to the curves in figure 1 ) . since @xmath182 ( equation [ [ rdlambda ] ] ) and @xmath183
, disks forming in a halo with mass @xmath6 and spin parameter @xmath38 are smaller for higher @xmath16 .
a higher formation redshift thus implies a larger spin parameter in order to yield the observed value of @xmath97 , something which is evident from figures 3 and 4 .
furthermore , disks seem to follow the @xmath160 curve , i.e. , disk - bulge systems do not have bulges that are significantly more massive than required by disk - stability .
this is at least consistent with the inside - out bulge formation scenario proposed in section 2.1 , in which the formation of disk and bulge is self - regulating and governed by disk stability . the ocdm model with constant and near to unity galaxy formation efficiency ( upper panels in figure 4 ) can be ruled out . in this scenario systems with a large disk - to - bulge ratio , such as spirals ,
are rare ; only @xmath184 percent of the dark halos is expected to yield such systems . the stability line @xmath162 crosses the average @xmath185 of the spin parameter distribution at @xmath186 , and thus half of the halos
is expected to form systems with a disk - to - bulge ratio less than unity .
this is in conflict with the fact that the major fraction of field galaxies are spirals .
furthermore , stable spirals can only form at high redshift ( @xmath187 ) , since for lower formation redshifts the spirals fall below the stability line . in the ocdm cooling scenario , spirals that formed recently tend to be slightly more stable , but still , the probability that a certain halo yields a system with a large disk - to - bulge ratio is rather small , rendering this scenario improbable . however , for the feedback ocdm cosmogony , a promising scenario unfolds : proto - galaxies that collapse at high redshifts are preferentially found in overdense regions such as clusters .
the densities of these virialized halos are high , and this ( somehow ) causes the galaxy formation to be very efficient ; nearly all the baryons in the dark halos make it into the galaxy . as a consequence , the systems that form require relatively small disk - to - bulge ratios ( e.g. , similar to s0s ) in order to be stable . as the universe evolves , and proto - galaxies with smaller over - densities ( which are predominantly located in the field ) start to turn around and collapse , the galaxy formation efficiency decreases , and systems with larger disk - to - bulge ratios are allowed to form .
this scenario thus automatically yields a morphology - density relation , in which systems with smaller disk - to - bulge ratios ( e.g. , s0s ) are preferentially to be found in overdense regions .
unfortunately , the disky ellipticals seem to not fit in this scenario . even at formation redshifts of three , the average disky
elliptical lies near the line with @xmath188 .
the scenario outlined above for our ocdm cosmology also applies to the scdm model with @xmath189 , albeit somewhat less restricting .
after all , @xmath143 only gets as large as @xmath190 .
consequently , in a scdm universe our model predicts less of a morphology - density relation than in a low density universe ; i.e. , the magnitude of the morphology - density relation depends on the baryon fraction @xmath28 .
we define the typical formation redshift of a subset of galaxies as that redshift for which the average disk yields @xmath191 . in figure 5
we plot histograms of inferred formation redshifts for spirals ( hsb and lsb disks combined ) , s0s , and disky ellipticals .
the three dotted lines are drawn to guide the eye and correspond to redshifts of 1 , 3 , and 10 .
the inferred formation redshift is determined by calculating the redshift at which a halo with @xmath192 yields the observed disk parameters .
if a disk lies already above the @xmath193 line at @xmath194 , a formation redshift of zero is assigned .
the arrows in figure 5 indicate the mean of the inferred formation redshifts for each sample . as can be seen ,
spiral disks are consistent with having formed fairly recently : @xmath195 with a mean formation redshift of zero .
this conclusion was already reached by mo et al .
( 1998 ) based on the same disk formation scenario as used here , but on somewhat different arguments .
the distribution of @xmath16 is much more spread out for s0s , with a mean formation redshift of @xmath196 .
note that misclassifications may be partially responsible for this large spread .
finally , the inferred formation redshifts for the disky ellipticals are unrealistically high , with a mean of @xmath197 . we thus conclude that unless there is a significant loss in specific angular momentum of the material that forms these systems , disky ellipticals seem to _ not _ form a continuous sequence with s0s and spirals , at least not in the sense of the formation scenario investigated here .
the histograms in figure 5 correspond to the feedback scdm cosmogony .
however , the results are very similar for the other cosmogonies discussed in this paper : although the details of the histograms are somewhat different , the averages are the same within the errorbars .
as discussed above it seems unlikely that disky ellipticals can be fit in the simple bulge - disk formation scenario envisioned here .
the small disk - to - bulge ratios in these systems renders an inside - out formation scenario for the bulges in disky ellipticals unlikely .
two alternative formation scenarios for disky ellipticals have been proposed : ( i ) disky ellipticals are the outcome of gas - rich mergers ( e.g. , bender , burstein & faber 1992 ; scorza & bender 1996 ) , and ( ii ) embedded disks are formed out of internal mass loss from the old stars in the spheroid ( scorza 1993 ; brighenti & mathews 1997 ) .
hernquist & barnes ( 1991 ) have shown how in gas - rich mergers the gas becomes segregated in two components .
the first component looses most of its specific angular momentum and collapses to form a dense central cloud .
when converted into stars , these compact blobs of gas may well account for the high central densities in disky ellipticals ( barnes 1998 ) .
the second component gets propelled to larger radii in the tidal tails originating from the merger .
this gas may either escape or reaccrete ( through cooling ) onto the newly formed ` elliptical ' and form a new disk component , resembling the embedded disks .
the numerous physical processes at play and the large parameter space available make it difficult to come up with detailed predictions for the resulting merger remnant .
however , at present there seem to be no clear observations that rule out gas - rich mergers as the origin for disky ellipticals ( but see kissler - patig , forbes & minniti 1998 )
. in the internal mass loss scenario one expects a clear correlation between the mass of the disk and the mass of the spheroid .
this correlation is indeed present and scorza & bender ( 1998 ) showed that typical mass loss rates for the old spheroid population are in excellent agreement with the typical ratio between the masses of the embedded disk and the spheroid .
another prediction from this scenario is that the spheroid and disk rotate in the same direction and that their specific angular momenta are similar ( at least if we assume that the gas blown into the interstellar medium by the spheroid does not loose its specific angular momentum when cooling to form the embedded disk ) .
scorza & bender ( 1995 ) found this to be the case from kinematic decompositions of line - of - sight velocity profiles of a number of disky ellipticals from which they determined estimates of the specific angular momenta of the embedded disk and the spheroid .
their data , which we plot in figure 6 , yields an average ratio of @xmath198 a third prediction associated with this scenario is that the disks are younger than the spheroids .
recently , de jong & davies ( 1997 ) reported that disky ellipticals have higher h@xmath199 line indices than the more luminous boxy ellipticals , and suggested this to be due to a young population , most likely the embedded disk ( see also brighenti & mathews 1997 ) . a final prediction for the internal mass loss scenario
is that the disks have surface brightness profiles that are steeper than exponential , since the gas does not cool from a uniform sphere in solid body rotation , but from a differentially rotating , centrally concentrated gas distribution ( scorza 1993 ; brighenti & mathews 1997 ) . indeed , scorza & bender ( 1995 ) and scorza et al .
( 1998 ) have found that the embedded disks often have surface brightness profiles that are significantly steeper than exponential .
the masses , specific angular momenta , surface brightness profiles , and ages of embedded disks are thus all consistent with them having formed out of mass loss from the old stellar population of the spheroid
. more accurate determinations of the specific angular momenta of embedded disks and their spheroids in a larger sample of disky ellipticals will proof helpful in constraining this further .
understanding galaxy formation is intimately linked with understanding the origin of the hubble sequence .
an important clue is provided by comprehending the formation of disky ellipticals and s0s , simply because these systems form the transitional class from the classical ellipticals to the spirals .
disky ellipticals seem to form a continuous sequence with spirals and s0s towards smaller disk - to - bulge ratio , and it is thus tempting to believe that disky ellipticals , s0s and spirals all formed in a similar fashion . the aim of this paper has been to investigate a simple formation scenario for disk - bulge - halo systems , and to search for the main parameters and/or processes that determine the disk - to - bulge ratio and thus explain to a large extent the origin of ( the major part of ) the hubble sequence .
we considered the disk formation scenario originally proposed by fall & efstathiou ( 1980 ) , in which the size of the disk , which is formed by cooling of the gas in a dark halo , is determined by the amount of angular momentum of the halo .
we have extended upon this formation scenario by including a simple picture for the inside - out formation of an additional bulge component out of the inner , low angular momentum material .
stability arguments are used to suggest that the formation of the bulge is a self - regulating process in which the bulge grows until it is massive enough to allow the remaining gas to form a stable disk component .
we do not describe the bulge formation in any detail but merely use empirical relations which allow us to describe the bulge by a single parameter , namely its mass .
each dark halo contains a fraction @xmath28 of baryons .
we introduced a galaxy formation efficiency @xmath34 which describes the fraction of those baryons that actually build up the disk - bulge system .
the theory of spherical collapse coupled with the definition of a virialized halo predicts a tully - fisher relation of the form @xmath200 as observed , with a zero - point that depends on the hubble constant @xmath107 .
recent observations , however , suggest that the tully - fisher zero - point is independent of redshift , implying that the galaxy formation efficiency is proportional to @xmath107 . a physical explanation for this redshift dependence may be the higher escape velocities and cooling efficiencies at higher redshifts .
for a combined sample of @xmath201 galaxies , varying from spirals to disky ellipticals , we calculated the value of the halo s spin parameter which yields the observed disk properties under the assumption that disk - bulge systems form in the way envisioned here .
we compared two cosmologies ( scdm vs. ocdm ) and investigated the differences between assuming a galaxy formation efficiency of unity and two ( extreme ) scenarios in which @xmath113 : a cooling scenario , in which we assume that only the inner fraction @xmath34 of the available baryons cools to form the disk and bulge , and a feedback scenario , in which each baryon , independent of its specific angular momentum , has a probability @xmath34 of ultimately ending up in the disk or bulge .
our main conclusions are the following : * disk - bulge systems do not have bulges that are significantly more massive than required by stability of the disk component .
this suggests a coupling between the formation of disk and bulge , and is consistent with the self - regulating , inside - out bulge formation scenario proposed here . *
if we live in a low - density universe ( @xmath202 ) , the only efficient way to make spiral galaxies is by assuring that only a relatively small fraction of the available baryons makes it into the galaxy , and furthermore that the angular momentum distribution of those baryons is similar to that of the entire system ; i.e. , the probability that a certain baryon becomes a constituent of the final galaxy has to be independent of its specific angular momentum . in the cooling scenario ,
most of the angular momentum of the system remains in the outer layers , and most halos form disk - systems with massive bulges , such as s0s .
if , however , the galaxy formation efficiency is regulated as described by our ` feedback ' model , a promising scenario unfolds : at formation redshifts @xmath203 the galaxy formation efficiency is unity , and systems that form build up a large bulge to support the disk that assembles around them .
these galaxies resemble s0s .
galaxies that form later , at @xmath204 , no longer require a massive bulge , and spirals are preferentially formed .
coupled with the notion that density perturbations that collapse early are preferentially found in high density environments such as clusters , this scenario automatically predicts a morphology - density relation in which s0s are most likely to be found in clusters . in a scdm universe a similar , albeit less restrictive , mechanism is at work , which predicts a morphology - density relation of smaller amplitude .
* a reasonable variation in formation redshift and halo angular momentum can yield approximately one order of magnitude variation in disk - to - bulge ratio , and our simple formation scenario can account for both spirals and s0s .
however , disky ellipticals have too large bulges and too small disks to be incorporated in this scenario .
apparently , their formation and/or evolution has seen some processes that caused the baryons to loose a significant amount of their angular momentum .
merging and galaxy harassment ( moore et al . 1996 ) are likely to play a major role for these systems .
finally we wish to emphasize of few of the major shortcomings of the oversimplified formation scenario discussed here .
first of all , we have neglected the fact that the merging of halos is an ongoing process , and that this is very effective in destroying disks ( e.g. , toth & ostriker 1992 ) .
gas - dynamical simulations that do not involve the energy and momentum feedback to the gas from supernova explosions , stellar winds , uv radiation etc .
produce galactic disks that are some two orders of magnitude smaller than the observed spiral disks ( e.g. , navarro & steinmetz 1997 ) : merging ( and also harassment ) are very effective in transporting angular momentum out into the halo , thus yielding more and more compact galaxies .
this problem with forming galactic disks of proper dimensions is often referred to as the angular momentum problem . despite our ignorance regarding the effects of merging , harassment , feedback , and ( re)-ionization of the universe , the observed sizes of ( spiral ) disks clearly suggest however , that the combine effect of all these processes is apparently such that the material that forms galactic disks has not lost much of its original angular momentum acquired from cosmological torques .
the use of the fall & efstathiou disk formation scenario thus seems justified , despite the aforementioned shortcomings . in the past semi - analytical simulations of galaxy formation
have been mainly based on the assumption that all bulges result from the merging of disk galaxies ( e.g. , kauffmann , white & guiderdoni 1993 ; cole et al .
1994 ; baugh , cole & frenk 1996 ; somerville & primack 1998 ) . in this paper
we have examined an inside - out bulge formation scenario , which should be regarded as complimentary to this merging scenario .
our formation scenario can account for both spirals and s0s , but fails to build systems that are even more bulge dominated .
although it has been shown that bar - instabilities can lead to the formation of small ( and close to exponential ) bulges , a more detailed study is required to investigate whether the inside - out bulge formation scenario discussed here can yield more massive bulges as well .
it is at least intriguing that the critical spin parameter is of the same order of magnitude as the typical spin parameter for halos , suggesting that a significant fraction of halos will yield unstable disks , unless part of the baryonic material is transformed into a bulge component as suggested here .
so despite the clearly oversimplified nature of the formation scenario envisioned here , it may provide a useful framework for future investigations of galaxy formation .
for discussions and advice , i m indebted to george lake , vincent icke , stephan courteau , julio navarro , and to the referee .
this work was supported by a hubble fellowship , # hf-01102.11 - 97a , awarded by stsci .
mestel ( 1963 ) noticed that disks of spirals have specific angular momentum distributions similar to that of a uniform sphere in solid body rotation .
exponential disks are thus a natural consequence of the fall & efstathiou disk formation scenario , if we assume that proto - galaxies have a specific angular momentum distribution similar to that of a uniformly rotating , uniform sphere , and that the material that forms the disk does not loose its specific angular momentum ( e.g. , gunn 1982 ; dalcanton et al .
1997 ) .
the total angular momentum of a uniformly rotating , uniform sphere with radius @xmath62 and constant angular velocity @xmath205 is @xmath206 and the mass with specific angular momentum less than @xmath207 is given by @xmath208.\ ] ] since the baryonic and dark matter are well mixed , the angular momentum distribution for the baryonic matter component is identical to that of the total mass of the system , and @xmath209 .
the bulge forms out of the low angular momentum material with a _ cylindrical _ radius @xmath210 . solving @xmath211 yields @xmath212 and
the total angular momentum of the bulge is @xmath213,\ ] ] which is the total angular momentum of the intersection of a solid sphere with radius @xmath62 and a solid cylinder with radius @xmath214 ( where the sphere and cylinder have a common center ) .
the disk forms out of the material with @xmath215 , and has a total angular momentum @xmath216 .
we assume that the specific angular momentum of this material is conserved during disk formation . in the feedback scenario ,
in which each baryon has a probability @xmath34 of ending up in the final disk - bulge system , we find upon substituting @xmath217 ( equation [ [ bulgerad ] ] ) in equation ( [ angmombulge ] ) , and with the use of equations ( [ dismass ] ) and ( [ bulmass ] ) @xmath218 \ ; { t\over 1 + t } \ ; \biggl [ { 5\over 2 } - { 3 \over 2 } \biggl({t \over 1 + t}\biggr)^{2/3}\biggr ] \equiv f(t,\epsilon_{gf } ) \ ; [ m_d + m_b]\ ] ] in the cooling scenario , only the baryonic material with @xmath219 cools to form the disk and bulge , where @xmath220 and one finds @xmath221 \equiv f(t,\epsilon_{gf } ) [ m_d + m_b].\ ] ] | we investigate the formation of disk - bulge - halo systems by including bulges in the fall & efstathiou theory of disk formation .
this allows an investigation of bulge dominated disk galaxies , such as s0s and disky ellipticals .
these latter systems , which consist of an elliptical spheroid with an embedded disk with a scale - length of typically a few hundred parsecs , seem to form a smooth sequence with spirals and s0s towards lower disk - to - bulge ratio . the aim of this paper is to examine whether spirals , s0s , and disky elllipticals all can be incorporated in one simple galaxy formation scenario .
we investigate an inside - out formation scenario in which subsequent layers of gas cool and form stars inside a virialized dark halo .
the inner , low angular momentum material is assumed to form the bulge .
stability arguments are used to suggest that this bulge formation is a self - regulating process in which the bulge grows until it is massive enough to allow the remaining gas to form a stable disk component .
we assume that the baryons that build the disk do not loose their specific angular momentum , and we search for the parameters and physical processes that determine the disk - to - bulge ratio , and therewith explain to a large extent the origin of the hubble sequence .
the spread in halo angular momenta coupled with a spread in the formation redshifts can explain the observed spread in disk properties and disk - to - bulge ratios from spirals to s0s .
if galaxy formation is efficient , and all available baryons are transformed into the disk - bulge system , cosmologies with @xmath0 can be excluded , since stable spiral disks would not be allowed to form . however ,
if we assume that the efficiency with which galaxies form depends on the formation redshift , as suggested by the small amount of scatter in the observed tully - fisher relation , and we assume that the probability for a certain baryon to ultimately end up in the disk or bulge is independent of its specific angular momentum , spirals are allowed to form , but only at small formation redshifts ( @xmath1 ) . at higher formation redshifts , stability arguments suggest the formation of systems with smaller disk - to - bulge ratios , such as s0s .
since density perturbations in clusters will generally collapse earlier than those in the field , this scenario naturally predicts a density - morphology relation , the amplitude of which depends on the baryon fraction of the universe .
disky ellipticals are too compact to be incorporated in this scenario , and they thus do not form a continuous sequence with spirals and s0s , at least not in the sense of the galaxy formation scenario envisioned in this paper .
alternative formation scenarios for the disky ellipticals , such as gas - rich mergers or an internal mass loss origin for the embedded disks , are much more viable . |
the decline in average star - formation rate provides a fundamental constraint to hierarchical models of galaxy formation , over half the age of the universe . using the taurus tunable filter ( ttf ) on the anglo - australian telescope ( aat ) we have obtained an emission - line sample of faint star - forming galaxies over half the age of the universe . this
_ ttf field galaxy survey _
( @xcite 1999 ) is primarily sensitive to star - forming galaxies seen through at three discrete wavelength intervals , corresponding to redshifts @xmath1 , 0.24 and 0.39 . in selecting through emission - lines
, we are using a quantity that scales directly with star - formation rate , effectively selecting galaxies by precisely the parameter we aim to measure . by scanning with wavelength
, we automatically define a strict volume - limited sample , less affected by the magnitude - volume sampling effects of conventional broadband surveys .
thus , the tunable filter technique provides a unique means of quantifying the star - formation activity across redshifts relevant to interpretation of the most recent star - formation history of the universe .
the red taurus tunable filter is a tunable fabry - perot interferometer covering 6500@xmath89600 ( @xcite 1998a , b ) .
the spacing between the ttf plates is much narrower and adjustable over a wider range than conventional etalons , thereby giving its function as a tunable narrow passband . in effect
, the ttf affords monochromatic imaging with an adjustable passband of between 6 and 60 .
ttf has an important advantage in the detection of faint line emission over conventional redshift surveys .
deep pencil - beam surveys typically pre - select objects down to @xmath9 ( @xcite 1995 ) or @xmath10 ( @xcite 1995 ) , since these represent the practical spectroscopic limits on the 4 meter telescopes used .
however , these limits are set assuming the objects are dominated by continuum . objects with continuum flux beyond the spectroscopic limit but with emission - lines above the limit could be detected by these same instruments , albeit through the lines alone . however , because the initial selection is made from the broadband flux , such objects are excluded _ at the outset _ on the basis of having too faint a continuum level . in order to match ttf , a broadband - selected redshift survey needs to select objects to a fainter limit , in the knowledge that for the faintest objects , only those with emission - lines will register a detection spectroscopically .
however , this represents a vastly inefficient way of finding emission - line galaxies . between @xmath9 and @xmath11 , the number of objects on the sky ( per magnitude interval ) increases by an order of magnitude ( @xcite 1995 ) .
such object densities would be impractical to search given the limitations of current multi - slit spectrographs in terms of object multiplex . in this paper
we compare the very different emission - line galaxy samples obtained by the _ ttf field galaxy survey _ and the autofib galaxy redshift survey ( @xcite 1996 ) in regions where the two overlap . in this paper
we describe the _ ttf field galaxy survey_. section [ s : strategy ] discusses the adopted survey strategy and the reasons for this approach .
section [ s : selection ] describes the selection of the main emission - line sample and its attributes .
known selection effects and potential sources of contamination are reviewed and their effect on the sample discussed .
the main characteristics of the sample measurable from these ttf data are given in sect .
[ s : character ] . included in this section
are narrowband number - counts and a comparison of our narrowband selected emission - line galaxies with those of the broadband - selected autofib galaxy redshift survey .
we also derive preliminary luminosity functions and compare them to those of existing surveys .
concluding remarks are made in sect .
[ s : conclremarks ] . unless stated otherwise , we assume a friedmann model cosmology ( cosmological constant @xmath12 ) with a hubble constant @xmath13 and deceleration parameter @xmath14 .
considerable freedom exists in a scanning narrowband survey conducted with a tunable filter , by virtue of the flexibility of the instrument .
the elements that remain fixed are field coverage on the sky and the airy - profile shape of the passband .
the adjustable parameters are the passband width , the passband coverage and sampling rate with wavelength , and the exposure time per slice . individually , each affects the number of objects obtained and a judicious selection is necessary if sample size is to be maximized .
the _ ttf field galaxy survey _ is a survey for redshifted emission - line galaxies in the field .
all fields were taken with ttf in 2 at the @xmath15 cassegrain focus of the anglo - australian telescope ( aat ) .
the survey comprises 15 scans of 10 slices , at random high - galactic latitude fields scattered around the sky . by _ slice _ ,
we mean an exposure at a particular wavelength , or the image obtained by co - adding many such exposures .
the total sky coverage is 0.27 with most fields selected on the celestial equator for maximum accessibility from both northern and southern observatories .
the scans are distributed between the 707/26 ( @xmath16 ) , 814/33 ( @xmath17 ) and 909/40 ( @xmath18 ) ttf blocking filters .
each scan was repeated 3 times with the telescope offset by @xmath19 between each so that cosmic - rays and ghost images could be removed through median filtering .
two ccds were used : a @xmath20 pix mit - ll and a @xmath21 pix tektronix with pixel scales of 0.371 and 0.594 respectively at aat @xmath15 . in both cases ,
individual fields were limited by a circular field stop of 9diameter .
table [ t : stepping ] summarizes the passband settings as used for scans in each spectral region .
mean passbands with fwhm ( @xmath22 ) of 12.9 , 16.4 and 22.3 were used to cover 707/26 , 814/33 and 909/40 respectively . in velocity space
the passbands had widths of 548 , 604 and 725 respectively .
we would have preferred to maintain a fixed velocity width between wavelength windows although this proved difficult in practice since many of the ttf spectral regions were being wavelength - calibrated for the very first time during these observations .
passbands were stepped in increments of @xmath23 as a compromise between velocity coverage and sampling continuity .
table [ t : scans ] gives the observational characteristics for scans in each of the three spectral regions , in terms of and . for ,
the spectral regions sample redshift windows at @xmath1 , @xmath24 and @xmath25 .
note that the passbands used include the contribution of nearby emission with .
corresponding redshifts in the line are @xmath26 , @xmath27 and @xmath28 , although we expect fewer galaxies at our flux limits .
we attempted to compensate somewhat for the increased volume ( per scan ) represented at higher redshifts and longer wavelengths .
as such , 7 fields were scanned in the 707/26 filter , 5 in the 814/33 and 3 in the 909/40 .
this arrangement provided approximately equal volumes in ( @xmath29 3 ) but a large range in ( @xmath30 @xmath31 3 ) , because of the low redshift for the 707/26 window .
the only way to match the low and high redshift volumes in an efficient manner would be through a wide - field instrument such as an objective prism on a schmidt telescope . however , these are typically 6 magnitudes less - sensitive than ttf in the case of photographic surveys ( e.g. @xcite 1994 ) and 4 magnitudes with the use of a ccd ( @xcite 2000 ) . for wide - field tunable filter searches ,
the main limitation is the degradation of bandpass transmission that an interference device such as a tunable filter experiences in the fast beam of a wide - field instrument ( @xcite 1959 ) .
@xcite ( 2001 ) describe the design of a wide - field tunable lyot filter that avoids these difficulties .
exposure times were typically 600 s per slice for the 707/26 and 814/33 scans and 1080 s per slice for the 909/40 scans .
of the 13 nights obtained on the anglo - australian telescope ( aat ) during 1997 april 1998 april , 7 were photometric with reliable flux measurements .
only scans taken under photometric conditions have been included .
these exposure limits permit us to reach @xmath32 2band as a @xmath33-detection in a 2 aperture , in our deepest exposures for the 909/40 interval .
for a 20 bandpass , this flux is equivalent to @xmath34 or @xmath35 as broadband continuum .
alternatively , if the @xmath32 2band is confined solely to a _ line _ , ( with no other continuum ) , then this is equivalent to a broadband detection limit of @xmath36 . the instantaneous star - formation rate ( sfr ) scales directly with the and line luminosity . using the calibrations for and from the literature review of @xcite ( 1998 ) we can determine the limiting star - formation rates of our sample .
a limiting flux detection of @xmath37 2bandis sufficient to detect and measure star - formation rates of a few tenths , comparable with smc - type levels at the nearest redshifts and the lmc at @xmath38 .
the narrowband scanning of ttf selects a galaxy sample that differs in two important ways from conventional redshift surveys .
first , the narrowband technique produces a _ volume - limited _ sample of emission - line galaxies .
the volume of redshift space covered is chosen by the observer through the range and placement of narrowbands .
this differs from the broadband limit of conventional redshift surveys which allows galaxies from a wide range in redshift , making total volume a complex function of redshift .
the second difference is that ttf selects objects through narrowband flux which is sensitive to the line flux and therefore star - formation rate .
most conventional redshift surveys define emission - line sub - samples from the initial broadband - selected sample .
the inclusion of a galaxy in such samples is dictated by the continuum of the combined stellar population and not by the instantaneous star - formation rate .
to define a complete star - forming sample over a given volume , a narrow bandpass well - matched to flux detection in an emission - line is essential .
since the ttf technique is without precedent we directed 9 of our fields to include emission - line galaxies found as part of the autofib galaxy redshift catalogue ( @xcite 1996 ) .
the autofib catalogue contains over 1700 galaxies in the apparent magnitude range @xmath39 .
since it is the combination of many optical surveys , emission - line galaxies are detected primarily through .
of the ttf fields observed , there are 18 emission - line galaxies in the autofib catalogue with redshifts placing in one of either the 707/26 , 814/33 or 909/40 blocking filters .
we defined the autofib emission - line sample as those galaxies with rest - frame equivalent widths ( rews ) greater than 10 . through the galaxies common to both catalogues
, we can examine how representative a broadband - selected survey is of the underlying star - forming population .
furthermore , we can gain insight into how the ttf detection rate goes as a function of broadband magnitude .
table [ t : observations ] shows the observations undertaken for the _ ttf field galaxy survey_. the number of autofib galaxies expected in is also included .
figure [ figaitoff ] shows the distribution of fields on the sky .
the fields overlapping with the autofib galaxies are indicated .
we made a search of the nasa extragalactic database ( ned ) to check whether any known clusters lie near our fields .
no clusters with measured redshifts overlap with our fields although 4 of the optically - identified cluster candidates of @xcite ( 1996 ) lie within 6 radius of our 10_3f , 13_3e , 10_3c and 10_3h field centres . within the @xcite sample , these cluster candidates are all rated with low significance .
however , we later examine the number density and wavelength distribution of emission - line candidates within these fields as a check .
all narrowband imaging emission - line searches are open to different lines at varying redshift .
any line for which both the rest wavelength and source redshift act to place it within a ttf survey interval will be detected if sufficiently luminous .
the fixed flux detection limit translates to different limiting luminosities in each emission - line , depending on its redshift .
furthermore , the fixed wavelength coverage of the scan defines different co - moving volumes at each of the redshifts .
this needs to be taken into account when optimising survey coverage , not just in terms of object density , but also in the _ type _ of emission - lines sought , since some are preferred to others for measuring star - formation rates ( @xcite 1992a , 1998 ) . in a blind search at these wavelengths
the emission - lines we would most expect to see in galaxies are , , and ( @xcite 1999 ; @xcite 1999 ; @xcite 1992b ) .
the ttf passband is too wide to separate and at least one of .
the detection of ly@xmath40 would be rare at the levels we are probing .
recent estimates for the density of ly@xmath40 emission - line galaxies with line fluxes of 3 to @xmath41 number @xmath42 per unit @xmath43 ( @xcite 2000 ) .
this implies negligible numbers of ly@xmath40-emitters at our brighter flux limit ( @xmath44 ) .
is another line we do not expect to find since it is are too far into the red to be found in an appreciable volume . in any narrowband survey
, we expect a competing luminosity - volume effect between being probed to fainter luminosities over smaller volumes , and background emission - lines at brighter luminosities but larger volumes .
is the most preferred line for measuring the instantaneous star - formation rate , being the strongest balmer recombination line and in the part of the optical spectrum affected least by extinction ( @xcite 1992a ; @xcite 1997 ; @xcite 1998 ; @xcite 1999 ) .
it is directly related to the number of short - lived ob stars , indicative of the stellar birthrate in a galaxy .
these hot , massive stars produce ultraviolet ( uv ) radiation that ionises the nearby gas to create an region .
recombination produces the observed emission - lines and of these , most closely traces the amount of ionising uv flux . of the other balmer lines ,
is weaker and prone to filling - in by the underlying stellar absorption . can be also be related to the instantaneous star - formation rate , although the underlying assumptions are less robust , since this ( and other lines such as , and ) , depend heavily on the metal fraction of the gas ( @xcite 1992a ) . in the case of lines with a higher ionisation potential ,
the energy of the incident uv flux is also an important factor . as a consequence , they are unreliable star - formation indicators .
of all the optical lines , is the preferred choice .
given the reasons outlined above , is the obvious target for our survey of star - formation .
however , it is prudent for us to examine the numbers of background emission - line galaxies likely from a scanning narrowband survey .
accordingly we made estimates of the galaxy numbers expected from each of the other major optical emission lines relative to .
for a friedmann model cosmology with @xmath12 , the co - moving volume of space sampled in an emission - line at redshift @xmath43 , is @xmath45 here , @xmath46 is the area of the field on the @xmath47 of the sky and @xmath48 is the redshift coverage of the scan and @xmath49 is the proper distance .
galaxy emission - line luminosity functions have been determined by several authors for h@xmath40 at a range of redshift ( @xcite 1995 ; @xcite 1998 ; @xcite 1998 ; @xcite 1999 ) . in most of these cases ,
the luminosity function is found to be well - matched by a @xcite ( 1976 ) function , @xmath50 where @xmath51 is the number of galaxies per unit volume at luminosity @xmath52 .
the parameters @xmath53 , @xmath54 and @xmath40 are constants describing the shape of the distribution .
more specifically , the density of h@xmath40 galaxies per unit volume , @xmath55 $ ] , is binned per 0.4 interval in @xmath56 , and can be related to the schechter description through @xmath57 \ ,
\frac{d \ , \log l_{{\rm h}\alpha}}{0.4 } = \phi(l_{{\rm h}\alpha } ) \ , d l_{{\rm h}\alpha } , \label{transfm}\ ] ] for @xmath58 , luminosity in h@xmath40 ( @xcite 1995 ) .
the total number @xmath59 of emission - line galaxies ( per unit volume ) detected to limiting luminosity @xmath60 can be obtained by integrating eqn .
( [ schecht ] ) to that luminosity , @xmath61 .
\\ \end{array } \label{totnumber}\ ] ] here , @xmath62 is the incomplete gamma function , where @xmath63 is the gamma function and @xmath64 . ] evaluated in terms of the schechter parameters .
we applied the function in eqn .
( [ totnumber ] ) to estimate the total numbers of emission - line galaxies seen in each of the ttf scan intervals .
as the luminosity function evolves rapidly with redshift , we assumed an evolving @xmath53 , @xmath54 and @xmath40 of the forms @xmath65 following the forms adopted by @xcite ( 1997 ) for evolution in the broadband luminosity function .
the free parameters @xmath66 , @xmath67 , @xmath68 , @xmath69 , @xmath70 and @xmath71 were constrained by the luminosity functions of @xcite ( 1995 ) at @xmath72 ( @xmath73 3 , @xmath74 , @xmath75 ) and @xcite ( 1998 ) at @xmath76 ( @xmath77 3 , @xmath78 , @xmath79 ) .
table [ t : calculations ] lists the parameters describing describing our adopted luminosity function evolution , as constrained by these two samples . with no equivalent emission - line luminosity functions for lines of , and ,
our best approach is to scale the luminosities by the mean line flux ratios of @xcite ( 1992a ) .
kennicutt deliberately selected his sample of e - irr galaxies to match the mix of morphological types seen in the revised shapley - ames ( rsa ) catalog ( @xcite 1981 ) .
as such , it can not be considered to be representative of the true space density these types .
accordingly , we computed mean flux ratios for , , , and weighting @xcite s galaxy types by their occurrence . measurements of @xmath53 by @xcite ( 1997 ) were used to estimate the relative space densities of different types and their evolution with redshift .
figure [ figkenevol ] shows that under such a weighting scheme , all line ratios show little change out to @xmath80 .
this is consistent with @xcite ( 1999 ) who find broad agreement between the / ratios of their @xmath81 sample and the local @xcite ( 1992a ) sample .
the greatest change is the sharp decrease in / ratios over @xmath82 .
we suspect this is due to the steep decline in the numbers of early - type galaxies ( fig .
[ figkenevol ] , _ inset _ ) , in which strong emission is most prevalent ( @xcite 1986 ) .
this constancy in line flux ratios is indeed what we might expect , given a field population composed almost exclusively of star - forming galaxies at such lookback times ( e.g. @xcite 1996 ; @xcite 1999 ) .
the effects of extinction aside , line ratios will only exhibit significant variation in cases of active galactic nuclei ( @xcite 1992a ) , for which @xcite ( 1999 ) find no density evolution over @xmath83 and mild evolution between @xmath84 and @xmath85 .
figure [ figcontamevol ] shows the relative predicted numbers that our calculations yield for each of the 707/26 , 814/33 and 909/40 ttf spectral windows .
numbers are shown as a function of flux for both individual flux bins ( fig .
[ figcontamevol ] , _ left _ ) and cumulatively with decreasing flux ( fig .
[ figcontamevol ] , _ right _ ) .
we have assumed the galaxies to have a mean internal extinction of 1 mag at and a mean /(@xmath86 ) flux ratio of 0.69 ( @xcite 1998 ) .
note however that both @xcite and @xcite ( 1999 ) find a trend in flux against .
we have not included this in the correction , since both surveys use slit spectroscopy ( 1.75 and 8 slits respectively ) which misses light from the outer regions of large galaxies in a way that narrowband imaging does not .
the flux limits of our raw sample are @xmath87 to @xmath88 .
as expected , galaxies dominate the cumulative numbers at the brightest fluxes because background galaxies in the other lines need to be intrinsically more luminous , and hence , much rarer objects . at the fainter fluxes ,
we find a broader mix of emission - lines as the larger volumes sampled for the background lines begin to balance the fainter luminosity limits probed in
. around the flux limits of our survey , dominates over thereby representing more than 90% of the full emission - line sample in the redder 814/33 and 909/40 bands . in the 707/26 band ,
the and galaxies comprise approximately equal numbers ( @xmath89% ) with comparatively fewer numbers of galaxies seen in and .
the dominance of at stronger fluxes in all bands suggests that samples restricted to these values will essentially constitute , with little contamination from background . in the case of 707/26 , this means truncating the emission - line sample artificially higher ( @xmath90 to @xmath91 ) than the natural flux limits of the survey .
the trade - off is that there are far fewer galaxies with higher line fluxes . in short ,
we expect to galaxies to dominate samples at fluxes exceeding @xmath92 for the 707 interval , and @xmath37 for the 814 and 909 intervals .
the observations were reduced as follows .
individual frames were bias - subtracted and divided by flatfields .
night - sky rings were removed and images aligned and co - added in three separate scans . object detection and measurement was undertaken using ( @xcite 1996 ) .
objects were detected from the original frames and subsequently photometered from a matching set of frames degraded to match the worst seeing frame .
cosmic - ray and ghost image catalogues were constructed for each scan and used to remove these detections from the raw emission - line candidate catalogues .
emission - line candidates in which line flux was detected and continuum was not were selected by the line detections appearing in one or two adjacent frames and no others . these we term our _ line - only _ candidates .
candidates showing line flux on an otherwise detectable continuum we call our _ line - on - continuum _ candidates .
these were found through a more complex procedure .
initially , all objects in the field with continuum detected in 5 or more frames had a straight - line of best - fit iteratively fit to the continuum flux with wavelength .
fluxes deviant from this fit were rejected on successive iterations refining the fit .
any objects showing flux deviations in one or two adjacent frames above the flux uncertainties of the continuum were included in the emission - line catalogue .
all objects in this group ( in excess of @xmath93 ) were inspected on a mosaic of postage - stamp images ( through the scan ) to confirm the rejection of cosmic - rays and ghosts .
the quality of the deviation above the fit was also checked to confirm the fit as true to the continuum and unaffected by bad data points . the continuum and emission - line fluxes in both sets
were calibrated through standard star observations from the same nights .
measured fluxes were corrected for extinction by our own galaxy ( albeit small since the fields were all at high galactic latitude ) and for flux truncation by the apertures where necessary .
the specific procedures are described by @xcite ( 1999 ) .
figure [ d13e7best ] shows example scans and ttf spectra for a subset of candidates from one of the _ ttf field galaxy survey _ fields .
all of the objects shown are line - on - continuum candidates , for which the continuum flux has been fit in order to find the emission .
preliminary and final continuum fits are shown in the spectra ( _ dotted _ and _ dashed _ lines respectively ) .
one object ( 214.14 ) is one of the 18 autofib galaxies within our fields . in detection terms , the relative strength of an emission - line is quantified in terms of its _
@xmath94-deviation_. this is the line flux in units of background @xmath94 . for line - only candidates , @xmath94 is the background rms of the local sky noise against which the detection was made .
for example , a line - only detection with a @xmath94-deviation of 3 is a 3@xmath94-detection in the usual sense . for line - on - continuum candidates
, the situation is complicated by the presence of continuum . in this case , the `` background @xmath94 '' is that determined by either the rms scatter of fluxes about the continuum fit , or the mean uncertainty in the continuum fluxes , whichever is greatest .
hence , objects with strong emission flux might be rated with a low @xmath94-deviation if that emission is superposed on a continuum of comparable or greater strength .
the @xmath94-deviation is an expression of the _ ease _ with which an emission - line was detected , not its strength in flux .
the raw catalogue of emission - line candidates was selected from all objects with @xmath94-deviations of 3 or greater . at @xmath33 , those candidates seen as a peak in a single frame ideally require confirmation through a follow - up observation such as spectroscopy . figure [ figlimits_sig ] shows the distribution of @xmath94-deviations for each field . the line - only detections ( _ horizontal tickmarks _ ) and line - on - continuum detections ( _ crosses _ ) are shown separately .
observe in figs .
[ figlimits_sig](@xmath95 ) and ( @xmath96 ) that from the point of view of detection , the majority of candidates ( @xmath97% ) are weak detections of between 3 and 4@xmath94 , as might be expected .
most fields show a similar distribution of @xmath94-deviations with respect to each other .
the exception is the 00_3a field which shows substantial numbers of strong @xmath94-deviation objects in both cases of object .
we suspect that this may be due to either the field being observed under above - average conditions and/or a possible chance alignment with a galaxy cluster .
we also note that this field has comparatively higher numbers of faint line - only detections , consistent with either hypothesis .
finally , note that that there is a systematic offset between the limiting line flux of the line - only sample and its corresponding line - on - continuum sample for the same field .
this is simply a reflection of the fact that a @xmath33-deviation detection seen against background sky can probe to fainter line levels than an equivalent detection measured against object continuum .
these differences in sensitivity under different circumstances must be taken into account when deriving a homogeneous sub - sample of objects .
the selection of a homogeneous sub - sample of emission - line candidates across many fields must necessarily take into account several observables simultaneously .
there are many factors ( both external and intrinsic to an object ) that influence its detection probability . by _ homogeneous sub - sample _ ,
we mean a sample of emission - line objects that conforms to a set of well - defined selection criteria , within which we know our sample is complete , or near complete . by _
complete _ , we mean a 100% detection probability for all objects within a given set of selection criteria .
the detection probability can be assessed by generating artificial datasets spanning similar object characteristics as the real sample .
since the observational properties of the artificial objects are known from the outset , a success rate for detection can be measured when these objects are passed through the same selection methods as were used for the real objects .
this can be used to assess the amount of incompleteness as a function of selection level and make for its correction . in practice ,
selection cuts are placed at some threshold level , below which , the incompleteness corrections become intolerably large .
corrections for incompleteness are made by weighting object numbers by the known missing fraction , provided the observed galaxy numbers are sufficiently large and the correction factors small . in the _ ttf field galaxy survey _
, all objects are first selected on the basis of a detection by , which becomes increasingly incomplete at the faintest limits .
the line - only candidates are subject to this alone whereas the line - on - continuum objects undergo further selection on the basis of line flux relative to continuum .
the detection probability of both procedures must be measured before we can set bounds to define a homogeneous sub - sample .
the bounds should be such that we can combine both types of object into a single uniform catalogue , within which the differences in selection method are no longer important .
we examined the detection probabilities for the fields observed in cases of worst seeing and/or highest sky background .
images were constructed containing artificial objects under identical conditions of seeing and background rms as the real observations .
the artificial objects populated a @xmath98 grid of object sizes and line fluxes across a range of values encompassing those from the real data .
objects of a given size and flux were duplicated 20 times .
was executed on the artificial images using an identical detection configuration as the real field concerned .
its success at finding some fraction of objects at a given size and flux is a measure of the detection probability as a function of both parameters , under the particular observing conditions . in all fields
, the detection probability fell away quite steeply from 100% detectability , over a relatively narrow strip along the grid .
two fields ( 22_3a for 707 ; 10_3c for 909 ) showed detection probabilities declining at higher flux levels than the other fields in that group .
these two fields also contribute significantly less to the final object numbers .
given this , and the high levels of incompleteness they would impose on the other fields , we excluded them from the homogeneous sub - samples .
figure [ figlimits5 ] shows just the raw line - only detection candidates in terms of both mean detected flux and object size .
the mean detected flux for these objects is simply the line flux detected on a single narrowband frame .
the samples in fig .
[ figlimits5 ] are labeled by blocking filter and the 22_3a ( 707 ) and 10_3c ( 909 ) candidates have been removed .
the solid lines are contours of detection probability for the _ single worst field _ of each panel , with the 60%-contour emboldened .
all other fields have contours at fainter flux levels , not shown on this figure .
the significant number of points in the @xmath99% region come from these more sensitive fields , which exhibit completeness to fainter flux levels .
the detection contours show that the natural detection limit of a narrowband survey such as this is emission - line surface brightness .
accordingly , we apply two cuts : a minimum line flux cut and a minimum surface brightness cut . taken together
, they select candidates from _ all _ fields from within the 60% detection contour of the _ worst _ field . individually ,
all fields except the worst are complete to the selection limits imposed by the latter .
note that because this contour only applies to the worst field , table [ t : selectlimits ] lists the flux and surface brightness cuts .
a value for surface brightness at the object center is inferred from its fwhm and total flux , assuming an exponential disk profile .
in addition to the cuts above , the line - on - continuum objects are subject to further selection based on the strength of the line flux _ relative _ to the continuum .
this is necessary because of the influence that the continuum level can have in determining whether an emission - line is detected , irrespective of line flux .
figure [ figlimits6 ] demonstrates how we apply these cuts to the line - on - continuum objects .
initially , we apply identical flux and surface brightness selection for the line - only candidates , since both catalogues depend upon object detection in the first instance ( fig .
[ figlimits6 ] , _ left _ ) .
note that in this case , the mean detected flux is the object _ continuum _
, not the line as in the line - only case .
second , we restrict the line fluxes to the _ same _ level as the mean detected continuum flux ( fig .
[ figlimits6 ] _ right _ , vertical line ) .
this is because the same detected flux cut applies to both the line - only and line - on - continuum objects , regardless of whether this detected flux is line or continuum .
in doing so , we have a uniform cut in line flux across both samples .
we must ensure that these flux cuts define a region of near-100% detection probability .
this time , the detectability is measured for a sample of artificial objects with both line _ and continuum_. a grid of objects with line and continuum values spanning the ranges in fig .
[ figlimits6 ] ( _ right _ ) was constructed using the same worst - case noise values for fig .
[ figlimits5 ] .
the detection probability as a function of these two parameters was obtained by measuring the success rate of the software in finding such objects .
the contours in fig .
[ figlimits6 ] ( _ right _ ) show levels of detection probability in this space .
the artificial objects were assumed to have the sky noise associated with the largest available aperture for the field . as with the real object catalogue , they were only included if the resulting line had a @xmath94-deviation of 3 or greater .
the boundary between 2 and @xmath33-deviation detections is represented in fig .
[ figlimits6 ] ( _ right _ ) by the dashed line .
this is the locus of points with 3@xmath94 line fluxes in the largest aperture .
had we defined emission - line objects at a lower level than 3@xmath94 , we would have populated the region above this line with candidates as well . however , in such cases it is difficult to be satisfied that deviations less than @xmath33 are significant compared to the natural scatter of points about the mean continuum level .
our third and final cut is that representing the faintest limit to which we can detect line flux relative to continuum .
this is an observed equivalent width ( ew ) cut , and the levels we have chosen are shown by the labeled solid diagonal lines in fig .
[ figlimits6 ] ( _ right _ ) .
the definition for equivalent width from narrowband measurements , is @xmath100 here , @xmath101 and @xmath102 are the respective line and continuum fluxes and @xmath103 is the effective width of the passband .
the effective width , @xmath104 , is that width of the rectangular bandpass with equal area to the true passband , @xmath22 , and with 100% peak transmission .
they are related through @xmath105 .
the _ rest - frame equivalent width _ ( rew ) is that measured in the reference frame of the galaxy .
it is related to the observed equivalent width by @xmath106 , where @xmath43 is the galaxy redshift .
we make equivalent width cuts at a level ensuring that we are complete both in terms of detection probability and 3@xmath94 line detections . we know that we could have detected an object anywhere within the boundaries of line and continuum flux , and observed equivalent width .
table [ t : selectlimits ] lists the equivalent width cuts applied in each case .
also listed are the rest - frame equivalent widths for at its redshift in these passbands .
we determine our overall sample completeness by comparing the numbers of objects actually observed with those expected over the same selection space . for each object
we determined a _ completeness weighting factor _ , being the reciprocal of the detection probability in the region of that object .
most objects have weighting factors of 1.0 since they lie in the region of 100% detection probability for their respective field . in summing the completeness weights , each object
is represented by the numbers we _ should _ have detected , thereby giving the total number of objects we could have expected for the homogeneous sub - sample .
figure [ figcomplete2 ] shows the cumulative completeness of the homogeneous sub - samples as a function of flux . note
that since completeness is a function of surface brightness , its decline is not due to flux alone .
however , fig .
[ figcomplete2 ] demonstrates that all samples are @xmath107% complete or greater , at the limits we have set .
table [ t : homogeneous ] gives the breakdown of types for our homogeneous sub - sample , and the total numbers of observed and expected objects . defining homogeneous emission - line samples is insufficient on its own if we wish to characterize the star - formation history of the universe .
we expect our line - flux - selected samples to contain a combination of galaxies seen through , and at a variety of redshifts matching the observed wavelength . however , the conclusions of sect .
[ s : vollimited ] ( in particular , fig . [ figcontamevol ] ) suggest that at the brightest fluxes , our samples will be dominated by emission - line galaxies .
the reason for this is as follows .
such fluxes translate into luminosities brighter than @xmath54 in the emission - line luminosity functions for the background galaxies . as a consequence ,
galaxies dominate , being the only set from which the members are significantly sampled beyond @xmath54 . as can be seen in all cases of fig .
[ figcontamevol ] , the transition from an -dominated sample to that inclusive of other lines is quite sharp .
this is due to the exponential decline in galaxy numbers for a schechter luminosity function for luminosities @xmath108 .
the homogeneous sub - samples in the 814/33 and 909/40 filters have fluxes greater than 0.71 and @xmath109 respectively ( table [ t : selectlimits ] ) . by fig .
[ figcontamevol ] , we would expect these samples to consist almost entirely of emission ( @xmath110% ) .
we define two samples for each of the 707/26 , 814/33 and 909/40 sets : one derived from the full sample of emission - line objects and the other from the homogeneous sub - sample .
they are selected by taking only those galaxies above the flux limit ensuring complete selection , as determined by fig .
[ figcontamevol ] .
table [ t : halphasamp ] summarizes the flux cuts made and the sizes of the resulting samples .
the original homogeneous sub - samples in the 814/33 and 909/40 filters remain intact since the original flux limits ( 0.71 and @xmath109 respectively ) restrict these samples to galaxies almost entirely ( @xmath110% ) . however ,
the homogeneous sub - sample in the 707/26 filter has a wider mix of , and galaxies and so we take our sample as those with line fluxes exceeding @xmath111 . at this level
the galaxies constitute more than 80% of the total sample .
table [ t : halphasamp ] also lists the limiting line luminosities for flux at the selection limit , at the mean redshift in each interval .
although this dominance of at higher flux is fortuitous for the present analysis , future work is necessary to solve redshift ambiguity , in the form of follow - up spectroscopy .
not only does this see the ttf emission - line identified , but the nature of the galaxy emission and its reddening can be revealed through ratios with other emission - lines .
although such follow - up is beyond the scope of this paper , it is the obvious direction for subsequent work on the survey .
an alternative solution to redshift ambiguity was explored in the form of broadband imaging .
as the major optical emission - lines are well - separated in wavelength , the corresponding galaxy redshifts separate the broadband colors unambiguously in a color - color plane .
this differs from conventional photometric redshift techniques ( see discussion by @xcite 1999 ) in that a _ precise _ starting point for the redshifts is given through the limited possibilities offered by , or .
the disadvantage of this method is that it requires _ all _ emission - line galaxies ( that make it into the final sample ) to have sufficient continuum for a broadband detection .
as the galaxies from our survey with sufficient continuum are few ( @xmath112% ) , they do not constitute a sample that both meets this criterion and offers sufficient numbers to be useful .
observations representing many hours on a 4 m - class telescope or greater are required , in which case , spectroscopic follow - up becomes an increasingly feasible alternative .
we obtained broadband colors for 12 of the 15 _ ttf field galaxy survey _ fields .
the regions were imaged in @xmath113 on the 1.0 m telescope at siding spring observatory , australia .
this yielded usable colors for the subset of our galaxies ( @xmath114% ) with sufficient continuum ( @xmath115 ) .
such colors can then be used to solve the galaxy redshift ambiguity .
more details on the application and results of this approach are presented by @xcite ( 1999 ) . however , because the brightest galaxies were too few to be useful , this method could not be applied as a redshift discriminant to the _ ttf field galaxy survey _ as a whole .
we tested the ttf detection of emission - line galaxies against a conventional broadband - selected redshift sample .
the autofib redshift survey ( @xcite 1996 ) is a composite sample of 1700 galaxies obtained through the combination of the dars ( @xcite 1985 ) , autofib ( @xcite 1996 ) , bes ( @xcite 1988 ) , ldss-1 ( @xcite 1990 , 1993 ) and ldss-2 ( @xcite 1995 ) samples .
galaxies have been selected from the magnitude range @xmath116 and the presence of emission is judged through the line .
passband is defined by the combined wavelength sensitivities of the kodak iiia - j emulsion and a gg385 glass filter . ]
the different samples probe different magnitude ranges and areas on the sky .
there are 827 members of the autofib redshift survey that can be classified as emission - line galaxies , defined as such by rest - frame equivalent widths greater than 10 . of these
, there are 18 within one of our 9 overlap fields at redshifts placing within the relevant scan window ( 707/26 , 814/33 or 909/40 ) .
we examined the detection success of ttf on these galaxies as a function of both @xmath117 magnitude and rest - frame equivalent width .
table [ t : figautofib ] summarizes the results .
in addition to the ttf detection rate , we show how many of the ttf object detections occurred as line - on - continuum and how many were line alone .
table [ t : figautofib ] shows the trend of fainter galaxies ( @xmath118 ) becoming line - only detections in increasing numbers .
this we expect since the narrow ttf passband is optimized for the detection of faint line flux and not faint continuum .
there is also the suggestion of a declining detection rate with fainter limiting apparent magnitude , although it is difficult to tell with such small numbers of galaxies .
_ all _ of the undetected galaxies ( except one ) lie at @xmath119 .
the one fainter @xmath120 galaxy which was detected , was detected as a line - only detection .
we conclude from this that @xmath121 is around the limit at which we can detect galaxies in both their line and continuum . beyond this ,
line detections dominate . when ranked according to rest - frame equivalent width , little
is seen in the way of ttf detection trends .
again , the small size of the overlap sample is a limitation , particularly because the majority of objects have emission with low to moderate rest - frame equivalent widths ( @xmath122 ) .
we have assumed here , however , that all galaxies exhibiting emission will have equally strong .
although this is commonly true it is not always the case , and some of the ttf non - detections in table [ t : figautofib ] may have been simply because there was little or no .
@xcite ( 1999 ) find a small but significant fraction of the stromlo - apm survey ( 11% ) that show but no .
they attribute this to strong stellar absorption and do not find such galaxies to be more prevalent at a particular redshift .
so an alternative possibility is that table [ t : figautofib ] underestimates the detection success of ttf in some cases .
however , this 11% is less than the @xmath123% of autofib galaxies that ttf does not detect .
this suggests that an additional factor is responsible , although it is difficult to be certain when such small sample sizes are involved .
@xcite note that the fraction of their broadband - selected sample seen in emission ( @xmath124% ) is the same , irrespective of whether emission is defined through the presence of or .
remarkably , they find that the number of galaxies with and no negates the galaxy numbers with and no . figure [ figautofib ] shows the relationship between spectral quantities measured around by ttf and by the autofib survey for the galaxies in common . in panels ( @xmath125 ) and ( @xmath95 ) , we find little correlation between either the and emission - line fluxes or between the corresponding equivalent width measures .
we do not see the ew(@xmath86 ) versus ew ( ) correlation of @xcite ( 1992a ) or @xcite ( 1999 ) , probably because this correlation is weak with large scatter and our sample size is small .
the large scatter in fig .
[ figautofib](@xmath95 ) may also be due to a tendency of spectroscopic equivalent widths to be overestimates at the limits of a broadband - selected sample , as discussed in sect .
[ s : sample ] . either way
, our results lend further weight to @xcite ( 1992a ) who finds flux and equivalent width to be loosely connected to the star - formation rate in a galaxy . in fig . [
figautofib](@xmath96 ) we show the relationship between the autofib @xmath117 magnitudes and the ttf narrowband continuum measurements .
there is reasonable agreement between the two with the line of best - fit being @xmath126 where @xmath117 is the photographic magnitude and @xmath127 is the narrowband continuum in the ttf bandpass ( @xmath128 2band ) . in making the fit we fixed the slope to @xmath129 and weighted the points by the @xmath130 continuum photometry uncertainties
the agreement in panel ( @xmath96 ) indicates that the scatter in the equivalent width measurements of panel ( @xmath95 ) is due primarily to the dispersion in line flux seen in panel ( @xmath125 ) .
equation ( [ contfit ] ) implies that the _ continuum _ detection limit of ttf is about the same as the @xmath131 limit of the broadband - selected survey . there is little overlap between the emission - line selected galaxies found in the the broadband - selected autofib redshift survey and the _ ttf field galaxy survey_. the vast majority of ttf - selected galaxies have continua too faint for inclusion in the broadband catalogue while the redshifts of the autofib line - emitters are too wide - ranging for the relatively narrow scan windows of ttf .
for the galaxies in common , ttf detects the brighter autofib galaxies ( @xmath132 ) in both line and continuum while fainter line - emitters are almost exclusively detected in the line alone .
while there is good agreement between the redshifts and continuum fluxes measured by ttf and autofib around and , there is considerable scatter between the measurements of line flux and equivalent width .
however , much of this dispersion can be attributed to intrinsic differences between the and lines themselves , as seen in the loose correlations between these lines by @xcite ( 1992a ) and @xcite ( 1999 ) .
improved insight into the relative merits of ttf over broadband - selected emission - line samples would be obtained through a redshift survey with near - infrared spectra overlapping with the ttf wavelength intervals directly , so that the same emission - line could be observed by each technique . previously ( sect .
[ s : strategy ] ) we discussed the need to examine the emission - line distributions with wavelength for signs of clustering . in the case of our sub - samples we can interpret the wavelength distributions directly as distributions in redshift space .
figure [ figpeaks2 ] shows such distributions by field for galaxies in the entire sub - sample .
no clear clustering is evident , although there are suggestions of peaks in the 14_3a , 22_3a and 13_3b fields .
the 22_3a clustering is irrelevant because this field has been excluded from the homogeneous samples used in our final analysis .
certainly we find no strong evidence for clusters in the 10_3c , 10_3f , 10_3h or 13_3e fields which coincide ( directionally at least ) with the @xcite ( 1996 ) candidates . given the limited impact that clustering appears to have on our field samples ,
we leave them intact as representative samples of the field population . of greater interest
are the fluctuations in the total numbers of galaxies between individual fields , particularly in the 707 set , and to a lesser extent in 814 .
this is due to the high flux cut imposed upon the 707 sample to ensure a sample of galaxies . at this level ,
the numbers of galaxies field - by - field are sufficiently few that fluctuations of this order are inevitable .
such variations are also seen between fields of broadband - selected redshift surveys ( cf .
redshift spikes in fig . 3 of @xcite 1988 ) .
for fig .
[ figpeaks2 ] , the peak wavelength of an emission - line was determined through two different methods depending on the nature of the emission . in objects with no continuum , the peak wavelength was that of the slice showing emission , or the mean wavelength if two adjacent slices showed emission . in objects with continuum , the peak wavelength of the most deviant slice , or the mean wavelength of two adjacent deviant slices was used .
mean wavelengths were computed by weighting the wavelengths of individual slices by the line flux in each . the mean flux ratio /(@xmath86 ) @xmath133 ( @xcite 1998 ) was used to adjust fluxes for the contribution of . in all cases ,
wavelength corrections were applied to account for the phase effect with object position relative to the optical centre of the fabry - perot beam .
table [ t : emisstype ] summarizes the occurrence of emission - lines in one and two slices for the homogeneous sub - sample of candidates .
the cases of detection and non - detection of continuum are shown separately .
single - slice emission - lines represent @xmath89% of objects with continuum , and a much larger fraction ( @xmath134% ) of those objects without .
we would expect a reasonable fraction of the emission - lines to appear in two , rather than one slice , given the spacing and width of the slices .
this is true for the line - on - continuum detections .
the extreme ratio of one to two - slice candidates for line - only case suggests that many of the one - slice objects could be noise peaks .
however , not all are , as table [ t : figautofib ] shows with the comparison to the autofib data .
only follow - up multi - object spectroscopy can resolve this difficult but crucial issue . in this section
we examine the evolution in the number count and volume - averaged star - formation rates of the _ ttf field galaxy survey_. from an observational perspective we would like some measure of the density of emission - line sources on the sky .
this should comprise all sources with redshifts placing the emission - line within the observed wavelength interval , whatever that emission - line may be . knowing this quantity is fundamental to optimizing source coverage and maximizing sample size for a volume - limited emission - line survey ( sect .
[ s : strategy ] ) .
multi - object spectroscopic follow - up to narrowband samples also requires a knowledge of object densities when deciding between strategies of spectral coverage and slit placement .
from the perspective of galaxy evolution , projected number counts potentially provide a strong constraint on evolutionary models , since they comprise galaxies from more than one redshift bin simultaneously . thus a chosen model must reproduce both the combined numbers observed at each wavelength and the relative mix of , , and other galaxies , all subject to the same evolution .
as such , narrowband number counts should provide a stringent test for competing evolutionary models of the star - forming population .
figure [ fignumbersevol2 ] shows the cumulative narrowband number counts from the complete homogeneous sub - sample of the _ full _ emission - line set ( including ) .
sample cuts at 4 and 5@xmath94 demonstrate how the weakest detections influence the counts obtained .
the sigma values in this case are the line strengths expressed as a statistical deviation above the background , be it sky or continuum noise .
corrections for incompleteness have been applied to all of the counts shown in fig .
[ fignumbersevol2 ] .
counts are shown down to the selection limits of the homogeneous sub - sample ( table [ t : selectlimits ] ) and the flux cuts for the sub - samples are indicated by the dashed lines .
counts are expressed in terms of the projected density of emission - line objects on the sky per bin per 100 wavelength scan coverage . in this way
, we have a useful measure of the object counts obtained in a flux - limited tunable filter survey , in terms of quantities directly observed .
note that the 100 is _ not _ the scan interval ( as given by column 2 of table [ t : scans ] ) , but rather the product of the effective bandpass and the number of slices per scan . also observe that 100 represents a different redshift coverage @xmath48 at the rest wavelength of each major optical emission - line .
however , it is fixed for that line , irrespective of the redshift at which it is observed .
the _ solid curves _ in fig .
[ fignumbersevol2 ] show the estimated numbers we should have expected based on evolving luminosity function calculation , of the type detailed in sect .
[ s : vollimited ] .
thus , the number evolution assumes our earlier model constrained by the @xcite ( 1995 ) and @xcite ( 1998 ) luminosity functions , namely @xmath135 this is identical to the general form presented in eqn .
( [ philwithz ] ) .
these surveys were chosen due to both their spread in redshift and use of galaxy spectra . the contribution from other major emission - lines
is calculated by scaling the luminosity function by the appropriate line - ratio ( _ c.f . _ fig .
[ figkenevol ] ) at the required redshift .
note that this is _ not _ a model of our observed counts , but a comparison of those we could have expected based on the luminosity functions of the earlier surveys .
figure [ fignumbersevol2 ] shows that the model expected counts ( _ solid line _ ) are a fair match to the @xmath136 distributions in the 707/26 and 814/33 intervals . in the 909/40 interval
they are a much closer match to the @xmath137 sample .
there are two potential contributions to this difference .
the most likely is that we are simply less - sensitive to the same type of galaxies at higher redshift , rendering what we detect as @xmath138 in 707/26 and 814/33 , as @xmath139 in 909/40 .
a second effect could be that the model constraints at @xmath140 ( @xcite 1995 ) and @xmath141 ( @xcite 1998 ) are insufficient for the extrapolation to @xmath142 where is seen in the 909/40 interval
. however , there are presently no -selected galaxy surveys that encompass @xmath142 .
observe also in fig .
[ fignumbersevol2 ] the significant excess ( @xmath143 to 3 times ) in the faintest narrowband number counts over the comparison with previous surveys ( _ solid line _ ) .
while we have not proven the validity of all detections less than @xmath144 , our visual confirmation of all line - on - continuum candidates ( 13% ) represents a greater fraction of the total counts than does the _ solid line _
, at the faintest limits where the difference is largest .
this suggests we have an excess in emission - line galaxies over the numbers we would otherwise estimate from these two previous surveys .
the corollary of this is that the population of faint star - forming galaxies is more numerous than these surveys have found , consistent with the preliminary findings of gronwall ( priv .
comm . ) for a new wide - field -survey at @xmath145 .
such a scenario is also consistent with our conclusions from the comparisons between the ttf and autofib samples in sect .
[ s : autofib ] . before interpreting such higher star - formation densities as indicative of the field
, we should consider possible sources of contamination .
both active galactic nuclei ( agn ) and quasi - stellar objects ( qsos ) are potential emission - line sources at non - zero redshift . like @xcite ( 1998 )
, we have not removed agn as their influence on the general emission - line population has previously been shown to be small ( _ i.e. _ @xmath146% ; @xcite 1996 ; @xcite 1996 ) . likewise , the occurrence of qsos to be rare against the general star - forming population : @xmath147 from within a typical high-@xmath43 ttf interval of @xmath148 , using the number - redshift distribution of @xcite ( 1988 ) .
this is consistent with the wide - field objective prism survey of @xcite ( 1976 ) for emission - line objects down to @xmath149 , with 96% as galaxies and 2% as possible qsos .
the most likely sources of galactic contamination are late - type stars .
@xcite ( 1993 ) in their 5-filter intermediate - band imaging survey using @xmath150 bandpasses , attributed @xmath19 to 20% of their _ rejected _ emission - line candidature to late - type stars .
such stars have strong molecular tio absorption bands with widths of typically @xmath151 .
presumably these spectral variations were difficult to distinguish from genuine line - emission in the broader @xmath150 bandpasses used by these authors .
however , our narrower bandpasses ( @xmath152 to 30 ) , the greater number of them ( 10 ) , and smaller scan interval ( @xmath153 to 300 ) , argues that we should not be susceptible to this problem in the same way .
given the unlikelihood of celestial contamination , our data suggest greater numbers of star - forming galaxies than has previously been attributed to the field population .
a consequence of this would be higher mean star - formation densities , although without confirmation of our faintest detections through additional observations , we can not quantify the increase .
specifically , follow - up spectroscopy is required to provide independent confirmation of the faintest emission - line detections , in addition to confirmation of the line as .
@xcite ( 2001 ) have obtained follow - up spectroscopy to a sample of ttf - selected emission - line galaxies in the field of qso mrc b0450@xmath8221 at @xmath154 .
of the 5 ttf candidates targetted , 3 were positively identified as emission - line sources , and 1 as a possible emission - line detection .
this important result demonstrates the capacity of ttf for detecting faint emission - line sources at high redshift . a more immediate approach to evolutionary trends is to compare luminosity functions directly .
figure [ figlfdn ] shows the luminosity functions of our homogeneous sub - samples in each interval for the @xmath137 ( _ filled circles _ ) sample .
the luminosity functions of @xcite ( 1995 ; _ open triangles _ ) and @xcite ( 1998 ; _ open circles _ ) are also shown in the relevant redshift interval .
all luminosity functions are shown uncorrected for extinction .
poissonian errors derived from the square - root of the galaxy numbers in each bin are shown .
for the ttf data , fluxes were corrected for the effect of using the mean flux ratio /(@xmath86 ) @xmath133 , obtained by @xcite ( 1998 ) for their sample at @xmath141 .
star - formation rates were derived using the same calibration as adopted by @xcite ( 1999 ) , namely @xmath155 from the @xcite ( 1996 ) models with solar metallicity and a salpeter imf .
no extinction corrections have been applied to the luminosities , although corrections have been made for the effect of the airy profile on sampling an ensemble of galaxy line fluxes ( @xcite 1999 ) .
we save the derivation of a luminosity function fit for when we have spectroscopic confirmation of the sample in hand .
figure [ figlfdn](@xmath125 ) demonstrates the excess in our @xmath1 counts over the faint - end of the @xcite ( 1995 ) distribution by almost an order of magnitude .
much closer agreement exists between our values and those of a preliminary luminosity function from the kpno international spectroscopic survey ( kiss ; gronwall 2000 , priv .
comm ) . taken together ,
these results suggest that the @xcite survey underestimates the faint - end of the local luminosity function , and hence , the star - formation content of the local universe . for the @xmath156 interval ( fig . [
figlfdn]@xmath95 ) we find broad agreement with the luminosity function derived by @xcite ( 1998 ) . due to the volume - limited nature of the survey
we have insufficient numbers to define the bright - end `` knee '' of the distribution . our measurements at @xmath157 ( fig .
[ figlfdn]@xmath96 ) are the first to use in the @xmath112 gyr of lookback time separating the samples of @xcite ( 1998 ) at @xmath141 and @xcite ( 1999 ) at @xmath158 .
our preliminary luminosity functions argue for an upwards revision of the star - formation content for the local universe .
this in turn implies a less - dramatic decline in the cosmic star - formation history than has been shown to date ( @xcite 1998 ; @xcite 1999 ; @xcite 1999 ) .
however , follow - up spectroscopy is needed to confirm the nature of the detections and to correct individual fluxes for the effects of extinction and . indeed ,
if many of our sub-@xmath139-detections are verified , then the measured excess would be greater .
we now summarize our preliminary findings from a census of the _ ttf field galaxy survey _ for star - forming galaxies . 1 .
using the taurus tunable filter ( ttf ) , we have obtained a narrowband - selected sample of emission - line galaxies in the field .
the survey is composed of 15 scans of 10 narrowband slices covering 0.27 on the sky , with 7 of the scans in a 707/26 nm wavelength interval , 5 in 814/33 nm and 3 in 909/40 nm .
there are 696 candidates ( at the @xmath33 level or greater ) in a homogeneous sub - sample for which we can correct for incompleteness ( 96.4% ) .
although some detections ( 13 ) are confirmed by the autofib galaxy redshift survey , the majority exist without separate confirmation .
this is required for both checking the validity of detections and correctly identifying the emission - line concerned .
the homogeneous sub - sample excludes two of the worst fields , one in each of the 707 and 909 intervals . through a synthesis of luminosity functions and median galaxy emission - line ratios
, we estimate the relative occurrence of different emission - lines in our sample . from this
we expect our samples to be dominated by galaxies in the sub - samples with line fluxes exceeding 1.50 , 0.71 and @xmath109 for the 707 , 814 and 909 intervals respectively . in these flux regimes , we have 660 candidates from the homogeneous sub - sample that we expect to consist mostly of galaxies , correctable for the small amount of incompleteness ( 96.5% ) inherent .
we find little evidence for strong clustering in either the full emission - line sample or the subset .
those of our fields overlapping with the @xcite ( 1996 ) cluster candidates did not yield any clear peaks in redshift space .
this allows us to rule out clusters in our field sample as seen in emission - lines .
it does not , of course , tell us anything about possible clusters outside of our scan intervals but in the same direction on the sky .
we compared the detection ability of ttf in finding emission - line galaxies to the broadband - selected autofib galaxy redshift survey ( @xcite 1996 ) .
our comparison was made between the galaxies found in autofib and the galaxies of ttf .
there is little overlap between the emission - line samples of the two surveys . for the sky regions that overlap ,
most of the ttf - selected galaxies are not seen in the autofib catalogue because they have continua beyond the selection limit of this catalogue .
conversely , ttf misses many of the autofib emission - line galaxies because they are beyond the comparatively narrow volumes sampled by ttf . for the galaxies in common
, ttf detects @xmath132 galaxies in both line and continuum while fainter galaxies are detected in the line alone .
ttf recovered 13 of an expected 18 autofib emission - line galaxies , although those that were missed may have been due to weaker emission than .
the two surveys agree in their continuum and redshift measurements although show scatter in their line flux and equivalent width measures
. however , much of this scatter may be due to intrinsic scatter in the correlation of and fluxes .
better insight into the detection rates of the two methods would be gained through a direct comparison concentrating on the same emission - line at wavelengths common to both .
the projected density of emission - line galaxies on the sky was compared to what we would have expected from existing measurements of the luminosity function . at the faintest limits we find a significant excess in these narrowband number counts , up to 3 times the expected counts at the faintest flux limits of our @xmath137 sample .
this excess is also seen when our preliminary luminosity functions are compared directly to those of other surveys , in each redshift interval individually . although the influence of false detections for the line - only candidates is yet to be quantified , our visual confirmation of the line - on - continuum candidates ( 13% ) is consistent with an excess of emission - line candidates .
such an excess is most likely due to our narrowband selection of the sample , which is without precedent for field emission - line galaxies at these redshifts .
the greater numbers imply higher star - formation densities than have previously been attributed to these epochs .
however , follow - up measurement of the detections is required to quantify this increase .
multi - object spectroscopic follow - up is necessary for future work on this sample .
such spectra would cover a wider wavelength range than ttf , and as such , would allow us to resolve many of the issues for which we have presently made assumptions .
spectroscopic information would first allow independent confirmation of the line detections and resolve the redshift ambiguity .
furthermore , with spectra we can determine the amount of reddening in individual galaxies through the /ratio and correct each for such .
if the spectra are of sufficiently high spectral resolution , the and lines can be separated , thereby avoiding contamination of the former by the latter . furthermore , the line fluxes could be measured directly from the line which would remove the need for the bandpass correction .
finally , through an assessment of ratios such as / and / we could identify and reject agn from our sample of star - forming galaxies ( e.g. @xcite 1987 ) .
many existing surveys ( e.g. @xcite 1998 ; @xcite 1999 ) include these emission - line sources , although they occur in small numbers relative to the general star - forming population . in short , the _ ttf field galaxy survey _
is a survey for star - forming galaxies is largely unprecedented in its systematic narrowband seach for field galaxies .
our data suggest galaxy counts and mean star - formation rates higher than those measured by previous and uv surveys .
an intriguing possibility for the higher values we observe is a population of faint star - forming galaxies , with continua sufficiently faint to place them beyond the range of conventional redshift surveys .
however , follow - up measurements utilizing multi - object spectroscopy are required for this to be confirmed .
it is a pleasure to thank matthew colless for many helpful discussions regarding this work .
we are also grateful to the referee caryl gronwall for her reading of the original manuscript and the many improvements that resulted from her suggestions .
the comments of the referees who examined this work during its presentation for a doctoral thesis are also gratefully appreciated .
we acknowledge the australian time allocation committee for their support of the ttf observations at the aat upon which this work is based . this research has made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , caltech , under contract with nasa .
dhj acknowledges the support of an australian postgraduate research award during the completion of this work .
folkes , s. , ronen , s. , price , i. , lahav , o. , colless , m. , maddox , s. , deeley , k. , glazebrook , k. , bland - hawthorn , j. , cannon , r. , cole , s. , collins , c. , couch , w. , driver , s. p. , dalton , g. , efstathiou , g. , ellis , r. s. , frenk , c. s. , kaiser , n. , lewis , i. , lumsden , s. , peacock , j. , peterson , b. a. , sutherland , w. and taylor , k. 1999 , 308 , 459 line : & & & & & & & & 707/26 & @xmath159 & 0.077 & 438 & 8610 & 7 & 1170 & 39.43 & 0.02 814/33 & @xmath160 & 0.24 & 1230 & 9510 & 5 & 6660 & 40.44 & 0.22 909/40 & @xmath161 & 0.39 & 1800 & 11200 & 3 & 9850 & 40.87 & 0.59 & & & & & & & & line : & & & & & & & & 707/26 & @xmath162 & 0.90 & 3290 & 8610 & 7 & 49500 & 41.67 & 6.57 814/33 & @xmath163 & 1.18 & 3880 & 9510 & 5 & 50300 & 41.93 & 12.14 909/40 & @xmath164 & 1.44 & 4320 & 11200 & 3 & 42500 & 42.12 & 18.73 & & & & & & & & 707/26 scans : & & & & & & 00_3a & 00 54 48.0 & @xmath830 25 00 & 9 & 600 & 1997 oct 08 & 2 10_3f & 10 45 18.0 & @xmath800 05 00 & 10 & 600 & 1997 may 01 & 10_3d & 10 45 42.0 & @xmath800 15 00 & 10 & 620 & 1998 apr 04 & 2 12_3a & 12 44 18.0 & @xmath800 07 00 & 10 & 600 & 1997 may 01 & 13_3e & 13 44 12.0 & @xmath800 01 00 & 10 & 600 & 1998 apr 04 & 2 14_3a & 14 45 30.0 & @xmath8600 05 00 & 10 & 400 & 1997 apr 30 & & & & 10 & @xmath86200 & 1997 may 01 & 22_3a & 22 05 10.9 & @xmath818 35 00 & 10 & 400 & 1997 apr 30 & & & & 10 & @xmath86200 & 1997 may 01 & & & & & & & 814/33 scans : & & & & & & 01_3a & 01 05 48.0 & @xmath830 00 00 & 9 & 600 & 1997 oct 08 & 1 13_3c & 13 43 27.0 & @xmath800 20 00 & 10 & 620 & 1997 apr 30 & 2 13_3b & 13 44 04.8 & @xmath800 22 00 & 10 & 600 & 1998 apr 05 & 2 13_3e & 13 44 12.0 & @xmath800 01 00 & 10 & 600 & 1998 apr 04 & 2 14_3a & 14 45 30.0 & @xmath8600 05 00 & 10 & 600 & 1997 may 01 & & & & & & & 909/40 scans : & & & & & & 10_3c & 10 44 30.0 & @xmath800 07 00 & 9 & 1080 & 1998 apr 05 & 10_3h & 10 46 24.0 & @xmath800 24 00 & 10 & 1080 & 1998 apr 03 & 2 13_3d & 13 44 42.0 & @xmath800 11 30 & 10 & 1080 & 1998
apr 03 & 3 & & & & by @xmath117 magnitude : & & & & & & & & @xmath177 & @xmath178 & 0 & 0 & 1 @xmath179 & @xmath180 & 0 & 1 & 3 @xmath181 & @xmath182 & 1 & 1 & 3 @xmath183 & @xmath184 & 4 & 2 & 1 @xmath185 & @xmath178 & 0 & 1 & 0 & & & & & & & & by ew : & & & & & & & & @xmath186 & @xmath187 & 1 & 0 & 1 @xmath188 & @xmath189 & 1 & 0 & 0 @xmath190 & @xmath178 & 0 & 0 & 1 @xmath191 & @xmath192 & 1 & 1 & 4 @xmath193 & @xmath187 & 1 & 0 & 1 @xmath194 & @xmath192 & 1 & 4 & 1 | recent evidence suggests a falling volume - averaged star - formation rate ( sfr ) over @xmath0 .
it is not clear , however , the extent to which the selection of such samples influences the measurement of this quantity . using the taurus tunable filter ( ttf ) we have obtained an emission - line sample of faint star - forming galaxies over comparable lookback times : the _ ttf field galaxy survey_. by selecting through emission - lines , we are screening galaxies through a quantity that scales directly with star - formation activity for a given choice of initial mass function .
the scanning narrowband technique furnishes a galaxy sample that differs from traditional broadband - selected surveys in both its volume - limited nature and selection of galaxies through emission - line flux .
three discrete wavelength intervals are covered , centered at redshifts @xmath1 , 0.24 and 0.39 .
galaxy characteristics are presented and comparisons made with existing surveys of both broadband and emission - line selection .
little overlap is found in a direct comparison between the _ ttf field galaxy survey _ and a traditional galaxy redshift survey , due to the respective volume and flux limitations of each .
when the number - counts of emission - line objects are compared with those expected on the basis of existing surveys , we find an excess of @xmath2 times at the faintest limits .
while these detections are yet to be independently confirmed , inspection of the stronger subsample of galaxies detected in both the line and continuum ( line - on - continuum subsample ; 13% ) is sufficient to support an excess population .
the faintest objects are galaxies with little or no continuum , rendering them undetectable by conventional redshift surveys .
this increase in the emission - line field population implies higher star - formation densities over @xmath3 . however , further study in the form of multi - object spectroscopic follow - up is necessary to quantify this and confirm the faintest detections in the sample .
2taurus-2 2bandergs@xmath4@xmath5band@xmath4 3mpc@xmath6 3mpc@xmath7 |
during the years 19942007 the hera @xmath0 collider in hamburg collided electrons or positrons with an energy of @xmath1 and protons with an energy of @xmath2 , at a centre - of - mass energy of @xmath3 . ] .
the last months of data taking were dedicated to special low energy runs , with proton energies of @xmath4 and @xmath5 , corresponding to centre - of - mass energies of @xmath6 and @xmath7 , respectively . during the whole data taking period , the h1 detector recorded almost @xmath8 in @xmath9 collisions and almost @xmath10 in @xmath11 collisions at high centre - of - mass energy .
these data are now fully exploited to achieve milestones of the hera physics program .
a few recent results are presented here @xcite .
at hera , @xmath12 bosons may be singly produced in the reaction @xmath13 .
leptonic decays of the @xmath12 bosons lead to the spectacular signatures of a high momentum isolated lepton and missing transverse momentum due to the decay neutrino .
such events are measured by h1 @xcite as a function of the transverse momentum of the hadronic system @xmath14 . at large @xmath15
an excess of @xmath16 observed over @xmath17 expected events is seen in @xmath11 scattering . in @xmath9 scattering only @xmath18 event is seen , with @xmath19 expected from the standard model ( sm ) .
the transverse mass distribution of the isolated lepton and the neutrino is exhibits a shape with a falling edge above @xmath20 , as expected from @xmath12 decays .
the production cross - section of single @xmath12 at hera is measured as a function of @xmath14 and is shown in figure [ fig : isolep ] , together with the observed transverse mass distribution .
a total cross - section of @xmath21 is found .
production at hera , title="fig : " ] production at hera , title="fig : " ] events with several high @xmath22 leptons ( @xmath23 or @xmath24 ) are also investigated . within the sm
such events are predominantly produced from @xmath25 collisions .
the h1 and zeus collaboration both measure individually these topologies @xcite . in order to enhance the sensitivity to new phenomena , a combined h1 and zeus analysis
is performed @xcite in a common phase - space .
multi - lepton events are observed at high scalar sum of transverse momentum @xmath26 . in @xmath11 collisions ,
@xmath27 events are observed for @xmath28 expected , whereas in @xmath9 collisions no event is seen , for @xmath29 expected .
the cross - section for multi - lepton production is measured as a function of the scalar sum of the lepton @xmath22 and as a function of the invariant mass , as depicted in figure [ fig : multilep ] .
( 1,0.27 ) ( 0.0,0 ) ( 0.333,0 ) ( 0.667,0 ) the h1 data also have been searched for excited fermions .
mass - dependent limits on the coupling @xmath30 for the production of excited neutrinos @xcite , excited electrons @xcite and excited quarks @xcite are shown in figure [ fig : excited ] .
the excited quark limits are derived with the additional assumption @xmath31 , complementary to tevatron searches , where @xmath32 is assumed . using the conventional assumption @xmath33 , masses @xmath34 , @xmath35 and @xmath36 are excluded at @xmath37 cl .
at low @xmath38 the measurement of structure functions is one of the key elements of hera physics .
recent analysis from h1 reach unprecedented accuracy in this area .
the data at low momentum transfer @xmath39 , have a precision of @xmath40 percent @xcite . in figure
[ fig : lowq2 ] they are compared to phenomenological models based on the dipole approach and good agreement is found . at medium
@xmath41 up to @xmath42 the precision reaches @xmath43 @xcite .
a nlo qcd fit ( h1pdf2009 ) of the complete h1 data collected up to the year 2000 is performed and parton densities are extracted .
the new data are described very well by the qcd fit .
the extracted parton densities have much reduced errors , as compared to earlier fits . the new data on the structure function @xmath44 and the results of the h1pdf2009 fit are shown in figure [ fig : inclusive ] .
three error sources are indicated on the fit results : the innermost error band are experimental errors , followed by theory uncertainties .
the outermost error band includes the uncertainty due the pdf parametrisation , which is defined here for the first time .
( left ) and the parton distribution functions obtained from the h1pdf2009 fit ( right),title="fig : " ] ( left ) and the parton distribution functions obtained from the h1pdf2009 fit ( right),title="fig : " ] , as a function of @xmath45 and @xmath41 ( left ) and as a function of @xmath41 , averaged over the corresponding @xmath45 values ( right),title="fig : " ] , as a function of @xmath45 and @xmath41 ( left ) and as a function of @xmath41 , averaged over the corresponding @xmath45 values ( right),title="fig : " ] using the data collected in dedicated low energy runs , a direct measurement of the longitudinal structure function @xmath46 is performed .
data collected at medium @xmath41 @xcite are combined with measurements at lower and higher @xmath41 and together cover the kinematic range @xmath47 .
the results are shown in figure [ fig : fl ] .
the new data at low @xmath41 are shown as a function of @xmath45 and @xmath41 .
all h1 data for a given @xmath41 are then averaged over @xmath45 and the result is shown as a function of @xmath41 .
the data are compared to predictions of the h1pdf2009 qcd fit and to a dipole model . at small @xmath41
the @xmath46 predicted from the qcd fit is lower than observed , whereas the dipole model is found to be in agreement with the data .
the data has a good potential to constrain the models of the proton structure at low @xmath45 .
charm and beauty quarks are produced at hera predominantly in the boson gluon fusion process , where the virtual photon and a gluon from the proton couple to a heavy quark - antiquark pair .
heavy meson production thus depends on the gluon content of the proton .
as the heavy quark mass is non - negligible compared to the scale @xmath48 , set by the exchanged virtual photon , theoretical prediction need to include the quark masses .
measurements of the contributions from heavy quarks to the inclusive structure function , @xmath49 and @xmath50 hence provide both interesting tests of qcd and sensitivity to the gluon density . ,
measured with a lifetime analysis , title="fig : " ] , measured with a lifetime analysis , title="fig : " ] figure [ fig : f2ccbb ] shows these contributions measured using a lifetime analysis : in addition to the scattered electron , tracks or vertexes significantly displaced from the primary vertex are required .
the analysis covers the kinematic range @xmath51 .
nlo qcd predictions are in agreement with these measurements .
another way to experimentally access the charm contribution is by reconstructing @xmath52 mesons .
the production of @xmath52 mesons is measured in the kinematic range @xmath53 .
good agreement with calculations is found , and the measurement is extrapolated to the full phase - space .
figure [ fig : dstar ] shows the cross - section for @xmath52 as a function of @xmath41 and the structure function @xmath49 after combining the @xmath52 data with the lifetime analysis .
meson production ( left ) and the structure function @xmath49 , after averaging @xmath52 and lifetime data ( right),title="fig : " ] meson production ( left ) and the structure function @xmath49 , after averaging @xmath52 and lifetime data ( right),title="fig : " ] the structure function @xmath49 is compared to qcd predictions based on different parameterisations of the parton densities .
the qcd predictions are found to be in agreement with the combined @xmath49 data . at low @xmath41
the data start to discriminate between various schemes of including heavy quark masses in the theory .
the precision of the data is comparable with the spread in the theoretical predictions and prefigures further constraints on the proton structure .
the production of jets at hera provides a clean environment to study fundamental properties of qcd , like the strong coupling constant @xmath54 .
the h1 collaboration measures inclusive jet production normalised to inclusive dis , as well as the rate of 2-jet and 3-jet events @xcite .
the measurements are performed as a function of @xmath41 and @xmath55 in the kinematic range @xmath56 , where @xmath55 is the jet transverse momentum .
a value @xmath57 is found in an nlo qcd analysis .
the small experimental error on @xmath54 compared to the dominant theoretical error emphasises the need of higher order calculations for jet production at hera . from jets at hera , title="fig : " ] from jets at hera , title="fig : " ] the running of @xmath54 with @xmath41 , as seen in jet production at hera and a comparison to other @xmath54 measurements is shown in figure [ fig : alphas ] . as a further investigation of qcd effects in the hadronic final state structure ,
the production of charged particles is studied .
the charged particle spectra are sensitive to the fragmentation process , in which hadrons are produced from the initial state partons . as opposed to the hard process , which can be calculated in perturbation theory , fragmentation is described by phenomenological models .
h1 measured the production of charged particles @xcite as a function of @xmath58 , the momentum fraction in the current hemisphere of the breit frame , in the kinematic range @xmath59 .
more insights in the fragmentation process are obtained by studying particle production separately for negatively and positively charged particles . these rates and
the associated asymmetry are shown in figure [ fig : cpa ] .
the asymmetry is small at low @xmath58 and rises up to @xmath60 for high @xmath58 , in agreement with various fragmentation models .
in contrast , the charge asymmetry observed at parton level , does not depend strongly on @xmath58 .
] diffractive processes contribute at a level of about @xmath61 to the dis cross - sections at hera .
they are characterised by the fact that the proton stays intact or fragments into a system with small mass @xmath62 , even though a second hadronic system is produced .
this is interpreted as the exchange of a colour - neutral hadronic object . in leading order qcd ,
this is explained by the exchange of two gluons .
experimentally , diffractive events are detected by identifying the outgoing proton or by requiring a large rapidity gap free of hadronic activity and separating the two hadronic systems .
new results on inclusive diffraction are derived from data taken with the h1 forward proton spectrometer ( fps ) .
leading protons are detected in the fps , which is located about @xmath63 downstream the main h1 detector in the hera tunnel , using roman pot detectors inside the evacuated part of the beam transport system .
compared to earlier measurements with a leading proton @xcite , the analysed data sample corresponds to an integrated luminosity increased by one order of magnitude .
the resulting reduced cross - section is shown in figure [ fig : fps ] as a function of three variables : @xmath64 , @xmath65 and @xmath41 .
the variable @xmath65 is related to bjorken @xmath45 and the momentum fraction @xmath64 of the colourless object entering the hard interaction , @xmath66 .
the data are compared to predictions from a qcd fit to inclusive h1 rapidity gap data @xcite .
agreement is found , taking into account the fact that the fit includes data up to a mass @xmath67 , whereas the new data are collected with a leading proton , i.e @xmath68 .
compared to @xcite , the new measurements extend to higher @xmath41 .
the precision reached is of order @xmath69 at low @xmath41 , limited by systematic effects .
measured for three centre - of - mass energies and the extracted structure function @xmath70 ( lower left),title="fig : " ] measured for three centre - of - mass energies and the extracted structure function @xmath70 ( lower left),title="fig : " ] measured for three centre - of - mass energies and the extracted structure function @xmath70 ( lower left),title="fig : " ] measured for three centre - of - mass energies and the extracted structure function @xmath70 ( lower left),title="fig : " ] a first measurement of the diffractive longitudinal structure function @xmath70 is performed , using data collected in the low energy runs .
the diffractive reduced cross section @xmath71 is decomposed into a linear combination of the structure functions @xmath72 and @xmath70 .
the contribution from @xmath72 is independent of the centre - of - mass energy , in contrast to the contribution proportional to @xmath70 .
thus , by analysing @xmath71 at various beam energies , @xmath70 is extracted .
the quantity @xmath71 for various beam energies and the resulting structure function @xmath70 are shown as a function of @xmath65 in figure [ fig : fld ] .
the @xmath70 data agree with predictions derived from the diffractive pdf fit @xcite to inclusive h1 data .
it is also interesting to look for the production of leading neutrons , which are produced from an electrically charged , colourless exchange with the proton .
such neutrons are detected in the forward neutron calorimeter ( fnc ) located @xmath73 from the main detector in the hera tunnel .
compared to earlier measurements @xcite , the new data profit from much increased integrated luminosity and an improved fnc . figure
[ fig : f2ln ] shows the measured structure function @xmath74 as a function of the three variables bjorken @xmath45 , @xmath41 and @xmath75 , where @xmath75 is the neutron energy , normalised to the proton beam energy . with a leading neutron ( left ) and comparisons to parameterisations of the pion structure function ( right),title="fig : " ] with a leading neutron ( left ) and comparisons to parameterisations of the pion structure function ( right),title="fig : " ] the data are compared to predictions from two monte carlo ( mc ) models : django , where leading neutrons are produced in the fragmentation process , and rapgap @xmath76 where leading neutrons are produced by the exchange of a charged pion . after applying global normalisation constants ,
the sum of these mc models is able to describe the data .
also shown in figure [ fig : f2ln ] are the @xmath74 data at @xmath77 divided by a corresponding pion flux factor @xmath78 . at high
@xmath75 the data are dominated by the pion exchange , and hence the pion structure function may be extracted .
the @xmath74 data divided by the pion flux factor compare well in shape to various parameterisations of the pion structure function .
the h1 collaboration has presented a wealth of new results based on the datasets collected until mid 2007 . a selection of these results in the areas searches for new physics , structure functions , heavy flavour production ,
hadronic final states and diffraction are discussed .
most searches have now been completed , using the full hera statistics .
the combination of h1 and zeus data has started , in order to reach the best possible sensitivity to physics beyond the standard model .
the precision of inclusive cross - sections data measured by h1 is now as good as @xmath43 in the bulk of the phase - space , opening new possibilities for qcd fits .
the structure function @xmath46 is measured over the full @xmath41 range available at hera .
of particular interest is the region of lowest @xmath41 , where the models diverge most and the data start to discriminate amongst them . using the h1 vertex - detector ,
contributions from charm and beauty to the inclusive structure function @xmath44 are extracted . in combination with data from @xmath52
decays , the data show sensitivity to the heavy flavour treatment in qcd fits and may contribute to further constrain the proton structure .
new measurements of jet production at high @xmath41 lead to very small experimental errors on the strong coupling @xmath54 .
higher order calculations are needed to pin down the theory errors .
the charged particle asymmetry of the scattered hadronic final state is observed for the first time in @xmath0 collisions and found in agreement with predictions based on phenomenological models .
a measurement of the diffractive longitudinal structure function @xmath70 is presented here for the first time .
diffractive pdf fits to inclusive h1 data are in agreement with this new data . using dedicated detectors fps and fnc the production of leading protons and leading neutrons , respectively ,
is measured .
the new leading proton data extend the kinematic range and have superior precision as compared to earlier measurements .
they are in agreement with predictions of the diffractive pdf fit .
the leading neutron data can be used to test fragmentation models and to extract the pion structure function .
f. d. aaron _ et al . _
[ h1 collaboration ] , submitted to epjc , arxiv:0901.0488 .
f. d. aaron _ et al .
_ [ h1 collaboration ] , phys .
b * 668 * ( 2008 ) 268 [ arxiv:0806.3987 [ hep - ex ] ] .
z. collaboration , arxiv:0906.1504 [ hep - ex ] .
t. h. collaboration and t. z. collaboration , arxiv:0907.3627 [ hep - ex ] .
f. d. aaron _ et al .
_ [ h1 collaboration ] , phys .
b * 663 * ( 2008 ) 382 [ arxiv:0802.1858 [ hep - ex ] ] .
f. d. aaron _ et al . _
[ h1 collaboration ] , phys .
b * 666 * ( 2008 ) 131 [ arxiv:0805.4530 [ hep - ex ] ] .
h. collaboration , arxiv:0904.3392 [ hep - ex ] .
h. collaboration , arxiv:0904.0929 [ hep - ex ] .
h. collaboration , arxiv:0904.3513 [ hep - ex ] .
f. d. aaron _ et al .
_ [ h1 collaboration ] , phys .
b * 665 * ( 2008 ) 139 [ arxiv:0805.2809 [ hep - ex ] ] .
h. collaboration , arxiv:0904.3870 [ hep - ex ] .
f. d. aaron _ et al .
_ [ h1 collaboration ] , phys .
b * 654 * ( 2007 ) 148 [ arxiv:0706.2456 [ hep - ex ] ] .
a. aktas _ et al .
_ [ h1 collaboration ] , eur . phys .
j. c * 48 * ( 2006 ) 715 [ arxiv : hep - ex/0606004 ] .
a. aktas _ et al .
_ [ h1 collaboration ] , eur . phys .
j. c * 48 * ( 2006 ) 749 [ arxiv : hep - ex/0606003 ] . c. adloff
_ et al . _
[ h1 collaboration ] , eur .
j. c * 6 * ( 1999 ) 587 [ arxiv : hep - ex/9811013 ] . | an overview of recent physics results from the h1 collaboration is given .
the covered areas are : rare processes and searches for new physics , structure functions and inclusive measurements , heavy flavour production , qcd and hadronic final states , diffractive scattering . |
recently , two experimental groups @xcite , reported the observation of a novel phenomenon : the existence of zero - resistance states in an ultraclean @xmath0 sample subjected to microwave radiation and moderate magnetic fields .
the magnetoresistance exhibits strong oscillations with regions of zero resistance governed by the ratio @xmath1 , where @xmath2 is the cyclotron frequency . according to zudov @xmath3 ,
the oscillation amplitudes reach maxima at @xmath4 and minima at @xmath5 , for @xmath6 an integer . on the other hand mani .
@xmath3 reported also a periodic oscillatory behavior , but with maxima at @xmath7 and minima at @xmath8 . in spite of a large number of theoretical works ,
a complete understanding has not yet been achieved .
a pioneer work put forward by ryzhii @xcite predicted the existence of negative - resistance states .
durst and collaborators @xcite also found negative - resistance states in a a diagrammatic calculation of the photoexcited electron scattered by a disorder potential .
a possible connection between the calculated negative - resistance states and the observed vanishing resistance was put forward in reference @xcite , noting that a general analysis of maxwell equations shows that negative resistance induces an instability that drives the system into a zero - resistance state . in our model ,
the landau - floquet states act coherently with respect to the oscillating field of the impurities , that in turn induces transitions from levels below fermi to levels above it .
this formalism is complemented with a generalization of the kubo formalism in order to correctly include the floquet structure of the problem .
we consider the motion of an electron in two dimensions , subject to a uniform magnetic field @xmath9 perpendicular to the plane and driven by microwave radiation . in the long - wave limit
the dynamics is governed by the schrdinger equation @xmath10 \psi \ , , \ ] ] here @xmath11 is the landau hamiltonian coupled to the radiation @xmath12 @xmath13 is the effective electron mass , @xmath14 , and the vector potential @xmath15 includes the external magnetic field and radiation field ( in the @xmath16 limit ) contributions : @xmath17 $ ] .
the impurity scattering potential is decomposed in a fourier expansion @xmath18 where @xmath19 is the position of the @xmath20th impurity , and the explicit form of the potential coefficient @xmath21 for neutral impurity scattering is @xmath22 , whereas for charged impurities localized within the doped layer of thickness @xmath23 : @xmath24 with @xmath25 .
a three step procedure is enforced in order to solve the problem posed by eq .
( [ ecs1 ] ) : ( 1 ) the hamiltonian @xmath11 can be exactly diagonalized by a transformation of the form @xmath26 , with the @xmath27 operator given by @xmath28 where the functions @xmath29 and @xmath30 represent the solutions to the classical equations of motion and the @xmath31 operators are the generators of the electric magnetic translation symmetries @xcite . ( 2 ) the transformation induced by @xmath32 is applied to the schrdinger eq .
( [ ecs1 ] ) , transformed into @xmath33 where @xmath34 and @xmath35 .
note that the impurity potential acquires a time dependence brought by the @xmath27 transformation .
( 3 ) the problem is now solved in terms of an evolution operator @xmath36 , using the interaction representation and first order time dependence perturbation theory .
the solution to the original schrdinger equation in eq .
( [ ecs1 ] ) has been achieved by means of three successive transformations @xmath37 the explicit expressions for the matrix element of these operators in the landau - floquet base appear in detail in reference @xcite .
the usual kubo formula for the conductivity must be modified in order to include the floquet dynamics . in the presence of an additional @xmath38 electric field
the complete hamiltonian is @xmath39 where @xmath40 is the hamiltonian in eq .
( [ ecs1 ] ) and @xmath41 with @xmath42 the static limit is obtained with @xmath43 , and @xmath44 represents the rate at which the perturbation is turned on and off . in order to calculate the expectation value of the current density , we need the density matrix @xmath45 which obeys the von neumann equation @xmath46 } = { \left[h + v_{ext } , \rho\right]}.\ ] ] we write to first order @xmath47 , where the density matrix @xmath48 satisfies the equation @xmath49 } $ ] . in agreement with eq .
( [ 3trans ] ) the density matrix is transformed as @xmath50 , and obeys the equation @xmath51 } , $ ] where @xmath52 and @xmath53 are the external potential and quasi - equilibrium density matrix transformed in the same manner as @xmath54 .
the transformed quasi - equilibrium density matrix is assumed to have the form @xmath55 where @xmath56 is the usual fermi function and @xmath57 the landau - floquet levels .
the argument behind this selection is an adiabatic assumption that the original hamiltonian @xmath40 produces a quasi equilibrium state characterized by the landau - floquet eigenvalues .
we then obtain the expectation value of @xmath58 in the landau - floquet base , from which the current density is evaluated according to @xmath59 $ ] .
the macroscopic conductivity tensor that relates the average current density to the averaged electric field , is obtained performing a space - time average , as well as an impurity average , assuming non correlated impurities .
we take into account the lifetime @xmath60 of the quasiparticles induced by the weak scattering , in the usual born approximation .
we work out explicit expressions for the longitudinal and hall conductivities ( both for the dark and microwave induced contributions ) @xcite , and quote the result for the longitudinal photoconductivity @xmath61 where @xmath62 is the two dimensional impurity density , and the broadened spectral function is given as + @xmath63,$ ] where @xmath64 are the advanced and retarded green s functions with finite lifetime @xmath65 . the auxiliary functions @xmath66 and
@xmath67 are defined according to @xmath68 + q_y^2 \omega_c^2 - 2 \omega_c q_x q_y \eta } { \vert \left(\epsilon_{\mu\nu } + \omega l - \imath \eta \right)^2 - \omega_c^2 \vert^2},\ ] ] and @xmath69 a ( \epsilon - \epsilon_\nu + l \omega + \epsilon_0 ) \right\ } \right]_{\epsilon_0= 0}.\ ] ] in eq .
( [ condlw ] ) @xmath70 is the bessel function with @xmath71 $ ] , @xmath72 being the magnetic length , @xmath73 the electron radiative decay width and @xmath74 the polarization vector . finally @xmath75 with @xmath76 being the generalized laguerre polynomial and @xmath77 . for three selections of the mobility.,width=453,height=302 ]
we have selected parameter values corresponding to reported experiments @xcite : effective electron mass @xmath78 , relative permittivity @xmath79 , fermi energy @xmath80 , electron mobility @xmath81 , electron density @xmath82 , microwave frequency @xmath83 , magnetic fields in the range @xmath84 and temperatures @xmath85 .
typical microwave power is @xmath86 from which the electric field intensity is estimated as @xmath87 .
the broadening @xmath88 is known to increase with the square root of the magnetic field , hence we take @xmath89 , which is connected through the relaxation time @xmath90 with the zero field mobility @xmath91 . in the case of charged impurity scattering
, the distance @xmath23 between the impurity and the 2des is taken as @xmath92 , and @xmath93 is estimated as @xmath94 .
explicit calculations demonstrate that results for charged neutral impurity scattering are very similar @xcite .
finally , the electron radiative decay @xmath73 , is interpreted as coherent dipole re - radiation of the oscillating 2d electrons excited by microwaves ; hence , it is given by @xmath95 , using the values of @xmath96 and @xmath13 given above , it yields @xmath97 .
one of the puzzling properties of the observed giant magnetoresistance oscillations is related to the fact that they appear only in samples with an electron mobility exceeding @xmath98 .
this phenomenon is absent in samples in which @xmath99 is reduced by one - order of magnitude .
this behavior is well reproduced by the present formalism .
figure ( [ figure1 ] ) displays the @xmath100 plot for three selected values of @xmath99 . for @xmath101 the expected almost linear behavior @xmath102 is clearly depicted . as the electron mobility increases to @xmath103 the magnetoresistance oscillations
are clearly observed , but negative resistance states only appear when @xmath104 .
we notice that @xmath105 vanishes at @xmath106 for @xmath6 integer .
the oscillations follow a pattern with minima at @xmath107 , and maxima at @xmath108 , adjusted with @xmath109 .
( [ condlw]-[derspec ] ) contain the main ingredients that explain the huge increase observed in the longitudinal conductance when the material is irradiated by microwaves . in the standard expression for the kubo formula
there are no floquet replica contribution , hence @xmath110 can be set to zero in ( [ derspec ] ) , in which case @xmath111 becomes proportional to the energy derivative of the fermi distribution , that in the @xmath112 limit becomes of the form @xmath113 , and the conductivity depends only on the states at the fermi energy . on the other hand as a result of the periodic structure induced by the microwave radiation ,
@xmath111 is dominated by terms proportional to @xmath114 , hence several contributions arise from the transitions between different landau levels . the model can also be used in order to test quirality effects induced by the magnetic field . in reference
@xcite we carried out calculations for different @xmath115field polarizations with respect to the current .
the amplitudes of the resistivity oscillation are bigger for transverse polarization as compared to longitudinal polarization . selecting negative circular polarization ,
the oscillation amplitudes get the maximum possible value . instead , the negative resistance states disappear for positive circular polarization .
these results are easily understood , because for negative circular polarization and @xmath116 the electric field rotates in phase with respect to the electron cyclotron rotation .
finally , we mention that the above discussed results are well described by single photon processes , however the present formalism is well suited to explore the non - linear regime in which multiple photon exchange play an essential role .
we acknowledge the financial support endowed by conacyt through grant no . `
42026-f ` , the departamento de ciencias bsicas uam - a through project no . ` 2230403 ` and rectora general de la uam - a . | in this work we present a model for the photoconductivity of a two dimensional electron system ( 2des ) subjected to a magnetic field .
the model includes the microwave and landau contributions in a non - perturbative exact way , impurity scattering effects are treated perturbatively . based on this formalism , we provide a kubo - like formula that takes into account the oscillatory floquet structure of the problem .
we discuss results related with the recently discovered zero - resistance states . |
ly@xmath0 photons have been widely applied to study the physics of the universe at redshifts from 2 to 8 .
the redshifted ly@xmath0 photons carry the information of the photon source , the halo surrounding the source , and the igm at early universe .
the optical afterglow of gamma - ray burst ( grb ) has been modeled as the red wing of ly@xmath0 photon absorption , and used to estimate the column number density of neutral hydrogen of igm at high redshift @xcite .
the transmitted flux of qso absorption spectrum at redshift @xmath9 consists of complete absorption troughs separated by spikes ( e.g. , @xcite ) .
the spikes have been explained as ly@xmath0 photons leaking at low density areas ( e.g. @xcite ) , and used to probe the turbulent behavior of igm at high redshifts @xcite .
the last but not the least , searching for redshifted ly@xmath0 photons from star forming galaxies at high redshift is believed to be a basic tool to explore the epoch of reionization ( e.g. @xcite ) .
therefore , it is crucial to have a complete understanding of the radiative transfer of ly@xmath0 photons caused by their resonant scattering with neutral hydrogen atoms . the radiative transfer of ly@xmath0 photons in a medium consisting of neutral hydrogen atoms has been extensively studied either analytically or numerically . yet
, there are very few solutions on the time - dependent behavior of ly@xmath0 photon transfer @xcite .
all other analytical solutions are time - independent based on the fokker - planck equation @xcite
. time - independent solutions are important but can only be used to describe the limiting asymptotic behaviors " of the radiative transfer @xcite .
they tell us nothing about the time scales of the radiative transfer of ly@xmath0 photons .
these time - independent solutions can not describe the wouthuysen - field ( w - f ) effect @xcite , which is essential for the 21 cm emission and absorption of neutral hydrogen at high redshift ( e.g. @xcite ) .
it is because the fokker - planck approximation would miss the detailed balance relation of resonant scattering , which is necessary to keep the w - f local thermalization @xcite .
numerical method based on the monte carlo ( mc ) simulation is also popular in solving the transfer of resonant photons ( e.g. @xcite ) .
however , there are very few works dealing with time - dependent problems . for instance , the time - scale of the w - f local thermalization is still absent in these studies
. time - dependent behavior of the ly@xmath0 photon transfer is especially important to understand observations of ly@xmath0 photons at high redshifts .
first , either the life time or the age of photon sources at high redshift is generally short .
second , the optical depth of the igm or the halo cloud around the sources is generally large . if the time scale of the transfer of ly@xmath0 photons is comparable to the life - time or the age of the photon source , the limiting asymptotic state " will never be approached .
one more problem is caused by the cosmic expansion , of which the time scale is short at high redshifts .
when the time scale of the cosmic expansion is comparable to that of the resonant photon transfer in optically thick medium , time - independent solutions probably do not exist .
recently , a state - of - the - art numerical method based on the weno scheme has been introduced to solve the integro - differential equation of the radiative transfer of resonant photons @xcite .
it reveals many interesting features of the transfer of ly@xmath0 photons in an optically thick medium , which can not be seen with the time - independent solutions of the fokker - planck approximation .
for instance , it shows that the time scale of the formation of the w - f local thermal equilibrium actually is short , only about a few hundred times of the resonant scattering .
the double peaked frequency profile of ly@xmath0 photon can not be described by time - independent analytical solutions unless the optical depth of @xmath10 photons is as large as about 10@xmath3 .
this result directly indicates the need of time - dependent solutions .
the weno algorithm of the integro - differential equation of the radiative transfer is fine , but like other high order scheme with fixed grid without artificial - viscosity , the computation time is much more than the monte carlo method .
therefore , the weno method would be not easy to deal with cases of medium with very high optical depth .
the goal of this paper is two - fold .
the first is to show that the monte carlo simulation method can properly match the results of weno method on the time - dependent features of moderate optical depth .
secondly , we study the time - dependent solutions of ly@xmath0 photons escaped from optically thick medium .
we will not work on specific objects , but focus on the general time dependent features which will affect the observability of the ly@xmath0 sources embedded in , or behind optically thick medium .
the paper is organized as follows .
section 2 presents the basic models we will study .
the monte carlo simulation method and its tests for time - dependent problems of ly@xmath0 photon transfer in an optically thick medium will be given in section 3 .
the major results of the time - dependent solutions of ly@xmath0 photon emission in the dla and igm models are given in 4 and 5 , respectively .
problems of absorption by optically thick medium are presented in 6 .
discussions and conclusions are given in 7 .
all the relevant formulae of the radiative transfer of ly@xmath0 photons and the details of mc simulation are presented in appendix .
we study the time dependent transfer of ly@xmath0 photons in two typical models of neutral hydrogen @xmath11 medium .
the first one is the so - called dla ( damped ly@xmath0 system ) halo model , in which a source is surrounded by a static spherical halo of physical radius @xmath12 , consisting of homogeneously distributed neutral hydrogen with number density @xmath13 and temperature @xmath14 .
the second one is a source at redshift @xmath15 located in homogeneously hubble expanding igm , of which the density and temperature are given by the parameters of the concordance @xmath1cdm universe .
we call it igm model . the radiative transfer equation of the two models
are given in appendix a. in order to compare with the weno solutions , for the dla halo model we use dimensionless time and radial coordinate , defined , respectively , as @xmath16 and @xmath17 , where @xmath18 and @xmath12 are the physical variables of time and radial coordinate , and @xmath19 is the cross section of scattering at the resonant frequency @xmath20 s@xmath6 .
therefore , @xmath21 and @xmath22 are the time and length in the units of mean free flight - time and mean free path of photon @xmath10 , respectively .
the value of @xmath22 actually is equal to the optical depth of the spherical halo from @xmath23 to @xmath22 at frequency @xmath10 . for a signal propagating in the radial direction with the speed of light
, we have @xmath24 . as usual , in frequency space
, we use variable @xmath25 , where @xmath26 hz is the doppler broadening at frequency @xmath10 by thermal motion @xmath27 of gas with temperature @xmath14 .
the variable @xmath28 is then the deviation of frequency @xmath29 from @xmath10 in units of the doppler broadening . with the dimensionless variables ,
the specific number intensity of photons is @xmath30 , where @xmath31 being the direction relative to the radial vector @xmath22 .
thus , all the solutions of @xmath32 do not refer to a specific density @xmath33 and size @xmath12 ( see appendix
this helps to see the common features of the dla halo model .
the optical depth of a halo or cloud with column density @xmath34 at frequency @xmath28 is @xmath35 where @xmath36 is the scattering cross section , @xmath37 , and @xmath38 is the normalized voigt profile given by is different from the one used in some literatures by a factor @xmath39 .
consequently , the expressions for mean flight time , mean free path and optical depth may be different by a factor of @xmath39 . ]
@xmath40 where the parameter @xmath41 is the ratio of the natural to the doppler broadening . for the ly@xmath0 line , @xmath42 .
the profile eq.(2 ) describes the joint effect of the gaussian distribution of the velocity of neutral hydrogen atom and the lorentz profile of cross section in the rest frame of the atom . for an expanding ( or collapsing ) halo or turbulent gas cloud
, the bulk velocity of the gas might be larger than the thermal velocity @xmath27 .
even in this case , the doppler or thermal broadening is still important , as it is the key factor leading to a local thermal equilibrium of ly@xmath0 photons ( the w - f effect ) . in an optically thick spherical cloud ,
most time - dependent behaviors can be described by two time - dependent distribution functions : 1 . )
angularly averaged , @xmath43-rescaled , specific number intensity @xmath44 , and 2 . )
@xmath43-rescaled number flux @xmath45 , which describes the photons escaped from a spherical halo of radius @xmath22 .
the equations of @xmath46 and @xmath47 are given in appendix eqs.(a1 ) and ( a2 ) .
for the igm model , we use the physical variable @xmath18 and @xmath12 because of the specific cosmological model we adopted .
the temperature of igm is taken as @xmath48 k and the parameter @xmath49 [ see eq.(a3 ) ] , which describes the hubble expanding , is equal to @xmath50 in the concordance @xmath1cdm universe .
we use monte carlo ( mc ) method to simulate @xmath51 and @xmath52 of the previous section
. most monte carlo codes of simulating time - independent solutions of radiative equation can easily be modified to deal with time - dependent problems .
if the feedback of photon transfer on the parameters of neutral hydrogen is negligible , the radiative transfer equations are linear with respect to @xmath53 , and thus to @xmath46 and @xmath47 .
thus , linear superposition of the solutions of the sources is valid .
the time - dependent solutions can simply be given by a weighted summation over the results of a single flash .
the weight of the summation is proportional to the time - dependent flux of the light source .
we will employ the same mc algorithm as used in @xcite , of which some of the details are given in appendix b and c. the major modification from the earlier methods , such as @xcite , is to record the time at each collision of the photon , and to take snapshots of photon distribution in spatial and frequency space based on these time stamps .
thus , the time - dependent solutions @xmath46 and @xmath47 of an arbitrary source can be given by a synthesis of the fluxes at different epochs from a single flash source .
as a test of the mc method , we calculate the flux @xmath54 at the outer boundary @xmath55 of a dla halo . this result
is shown in fig . 1 , in which the curves are the mc results with parameter @xmath56 , 2000 and 3000 .
they show typical double - peaked profile .
the curves of @xmath57 and 3000 actually are the same .
the overlapping indicates that @xmath54 is already at the limiting asymptotic state , or saturated for time @xmath58 . in fig .
1 , the data points are given by the weno numerical solutions of @xcite .
therefore , the mc method can match the time - dependent weno solutions .
photon flux @xmath59 from a dla halo with radius @xmath60 and parameter @xmath61 .
the curves are from mc simulations at time @xmath56 , 2000 and 3000 , respectively from bottom up .
the last two curves ( 2000 and 3000 ) are already overlapped with each other .
the symbol points marked by
+ " and x " are from the weno numerical solution of the radiative transfer equation @xcite at epochs @xmath56 , 2000 , respectively.,width=302 ] in the weno method , the solutions of the angularly averaged specific intensity @xmath62 and flux @xmath54 at boundary @xmath63 should satisfy the condition @xmath64 ( unno 1955 ) .
this is the result of eddington approximation and the assumption of no incoming photons at the boundary .
our mc simulation results of @xmath65 at time @xmath66 are shown at various radius in fig .
2 . at the surface
where @xmath67 , we see @xmath68 at the center frequencies when compared with fig.1 .
it shows that the mc simulation can pass the test of unno s boundary condition .
we also find that @xmath69 relation is only valid at the boundary at center frequencies .
slightly beneath the surface , at radius @xmath70 , we find that @xmath71 , and @xmath72 , respectively .
the enhanced photon intensity is a result of backward scattered photons near the boundary . at time @xmath56 for dla halo model of parameter @xmath73 with size @xmath55 .
the curves correspond to @xmath74 100 , 99 , 98 , 95 , 90 , 80 , respectively from bottom up .
the flux @xmath47 at time @xmath66 at the boundary @xmath67 from fig.1 is marked with
+ " symbols for comparison . ,
width=302 ] besides the curve of @xmath75 , fig .
2 also plots @xmath76 at time @xmath56 , but for @xmath77 , 90 , 95 , 98 , 99 .
we see that all the curves of @xmath78 are almost flat in the range of @xmath79 .
it means that the frequency distribution of photons is thermalized near the resonant frequency @xmath10 .
that is , the frequency distribution is of boltzmann @xmath80 , where @xmath14 is the kinetic temperature of neutral hydrogen gas in the halo .
this is the w - f local thermalization @xcite .
2 tells us that the w - f local thermalization is achieved by resonant scattering even at @xmath81 . yet
photons at the outermost layer @xmath82 have not yet been thermalized , as the optical thin layer does not carry enough number of scattering .
this result is similar to the weno solution @xcite .
the second test is on the @xmath83-@xmath84 relation , where @xmath85 are frequencies of the two peaks of the double - peaked profile as shown in fig . 1 ,
and @xmath84 is the optical depth parameter defined in eq.(1 ) .
this relation has been studied by many time - independent solutions based on the fokker - planck equation .
the major conclusion is the so - called @xmath86-law , i.e. @xmath87 , where @xmath88 is a constant of order 1 @xcite .
it is well known that the @xmath89-law is available only when the optical depth @xmath84 is very large .
however , the @xmath85-@xmath84 relation available for various @xmath84 has not been calculated until very recently .
it may be due to the absence of proper numerical solver of the integro - differential equation of resonant photon transfer in optical thick medium .
the weno solver provided the first @xmath85-@xmath84 relation of dla halo with moderate and high optical depth . vs. @xmath90 given by mc simulation ( data points ) of a dla halo with optical depth @xmath84 .
the parameter @xmath41 is taken to be @xmath91(bold line ) and @xmath92 ( light line ) . the dashed straight line of @xmath93-@xmath94 with slope 1/3 is to show the @xmath86-law.,width=302 ] the @xmath85-@xmath84 relation given by the mc simulation is presented in fig .
3 , which shows that the @xmath89-law is significantly deviating from the mc results when @xmath95 , where the optical depth is smaller than @xmath96 at @xmath97 .
this result is the same as the weno solution . in fig .
3 the parameter range of [ @xmath98 $ ] is larger than that of weno solution [ fig . 4 of @xcite ] . in the range @xmath99 ,
the @xmath83-@xmath84 relation is almost flat with @xmath100 .
it is because the double - peaked profile is from photons stored in the frequency range of @xmath79 and in the local thermal equilibrium state .
the positions of the two peaks , @xmath101 , actually is about the same as the frequency range of the local thermalization .
the frequency width @xmath102 of the local thermal equilibrium state is determined by the doppler broadening , and very weakly dependent on @xmath84 .
thus , once the photons in local thermal equilibrium state are dominant , we always have @xmath103 .
this point can also be seen with figs . 1 and 2 , in which the positions of the two peaks are kept to be @xmath104 despite that the intensity of the flux increases with time significantly .
when @xmath105 , the curves of @xmath83 are no longer determined by one variable @xmath106 , instead by variables @xmath41 and @xmath84 , separately .
the value of @xmath83 shows a quick drop to zero at @xmath107 for @xmath108 , and @xmath109 for @xmath110 .
the w - f thermal equilibrium can not be established in halos with small optical depth @xmath111 , and therefore , photons from these halos do not have double - peaked profile .
since thermalization will erase all frequency features within the range @xmath102 , the double - peaked structure does nt retain information of the photon frequency distribution within @xmath79 at the source .
it is impossible to probe the frequency profile for @xmath79 ly@xmath0 photons of the source from the escaped ly@xmath0 photons .
this property can also be used as a test of simulation code .
that is , the simulation results should be independent of the profile of ly@xmath0 emission from the sources , only if the profile is non - zero within the range @xmath79 , i.e. it should not matter whether the source is monochromatic , or has a finite width around @xmath10 .
it is well known that the spatial transfer of ly@xmath0 photon in optically thick halo is not simply a brownian random walk .
the time scale of ly@xmath0 photon escape from optically thick halo is much shorter than that of brownian diffusion . it is because the spatial transfer depends on the diffusion in frequency space .
this is the so - called `` single longest excursion '' process @xcite .
however , earlier estimates of the escaping time scale based on `` single longest excursion '' can not describe the details of the time - dependent behavior of photons escaping from optically thick medium . if the central source of a dla halo is assumed to be a photon flash , the time dependence of the luminosity of the photon source is proportional to @xmath112 . without scattering ,
the luminosity at the boundary of the halo with size @xmath22 should still be a delta function as @xmath113 , i.e. it is also a flash , but with a retarded time @xmath22 , which is the time needed for a freely streaming photon from the center to the edge of the halo with speed of light . considering the effect of resonant scattering , the luminosity of escaped photons will no longer be a flash . the light curve of the luminosity from such a source for a halo with size @xmath114 and @xmath115 is shown in fig .
4 , in which @xmath116 is the flux integrated over all frequencies of escaped photons .
photons in a dla halo model of @xmath117 with optical depth @xmath118 .
one photon is supposed to be emitted from a central source at time @xmath119 , i.e. the integral of the light curve is equal to 1 .
, width=302 ] fig .
4 shows that the light curve lasts from time @xmath120 to @xmath121 .
the peak of light curve @xmath122 is at @xmath123 , and the time duration @xmath124 of @xmath125 , is also about @xmath126 . since the source of photons does not contain any time scale , both the numbers @xmath127 and @xmath124 are from the size or optical depth of the halo .
the time scale @xmath127 means that the retarded time of the escaping of photons from the halo is as large as @xmath128 or 10@xmath84 .
the amount @xmath124 means that the time - distribution of the escaped photons are significantly spread out from a delta function @xmath112 to @xmath129 . without resonant scattering
, photons emitted from the source will escape from the halo at time @xmath130 .
with resonant scattering , the majority of photons emitted from the source will not escape from the halo until time @xmath131 .
therefore , a huge number of resonant photons is stored in the halo .
the resonant nature of ly@xmath0 photon scattering let the photons to stay in the halo with a time scale equal to 10 times of the optical depth @xmath84 of the halo . in our model
the destruction processes of ly@xmath0 photons , such as the two - photon process , are not considered , and dust absorption is ignored too . the number of photons is conserved .
thus , we have @xmath132 .
therefore , the light curve , @xmath133 of fig .
4 can be understood as the probability distribution of the time of ly@xmath0 photon escaped from a @xmath114 halo . with dimensionless variables , the curve of fig .
4 is also the probability distribution of the total length of the path of a photon transferring from @xmath23 to @xmath114 . in this context , @xmath127 and @xmath124
can be used as the most probable length of the path , and the variance of the distribution , respectively . considering that the most probable path length @xmath127 and the variance @xmath124 have the same order ,
the spatial transfer of ly@xmath0 photon in optically thick halo essentially is still a random process of diffusion . the light curve of a flash source ( fig .
4 ) can easily be generalized to arbitrary sources . considering
the total flux is linearly dependent on sources , the total flux @xmath134 is given by @xmath135 where @xmath133 is the curve of fig . 4 , and @xmath136 is the time - dependent ly@xmath0 photon flux of the sources . for a source with stable luminosity of ly@xmath0 photons , escape coefficient of ly@xmath0 photons emergent from a halo with size @xmath22 can be calculated by @xmath137 if the luminosity of sources is normalized .
since the number of ly@xmath0 photons is conserved , the total number of escaped photons should be equal to the total number of emitted photons by the source when a stable state is reached .
thus , the escape coefficient of a stable source should reach to 1 when @xmath21 is large enough .
however , before the system approaches to stable state , the escape coefficient @xmath138 can be much less than one .
we have calculated the time dependent solution of the escape coefficient @xmath139 for a stable photon source with unlimited life time located at the center of the halo with optical depth @xmath140 . at time @xmath141 , the escape coefficient @xmath142 . if we define the time scale @xmath143 of reaching saturated or stable state as @xmath144 , it yields @xmath145 , or @xmath146
. therefore , @xmath147 is always significantly less than 1 , when the age is less than about @xmath148 or 30 @xmath22 . for photon sources located in the center of a dla halo of @xmath149
k with optical depth @xmath150 .
the life duration of the source @xmath151 is taken to be @xmath152 10@xmath3 , 10@xmath153 , 10@xmath154 , 10@xmath155 and 10@xmath156 , respectively , from bottom up using color code orange , violet , cyan , green , and black .
, width=302 ] fig . 5 presents the time dependent solutions of escape coefficient @xmath139 for photon sources in a @xmath157 k neutral medium with limited life time to be @xmath158 and @xmath159 , respectively , from bottom up .
it shows that the escape coefficient is always less than 1 if @xmath160 . in this case , the time scale of the escaping of ly@xmath0 photons in @xmath161 halo is much larger than the life time of the source , and therefore , photons emitted within the time duration @xmath119 to @xmath162 are fully spread over a time scale @xmath163 .
thus , the escape coefficient of a source with short life time is always much less than 1 when the dla halo is optically thick .
this mechanism would be important for understanding the observable features of high redshift objects such as grbs or first stars .
even when dust absorption is negligible , the escape coefficient can be small , even very small , when the age of the object is small .
it is interesting to see that although different curves of fig . 5 correspond to very different @xmath164 , all the curves with @xmath160 approach their maximum at about the same time @xmath165 .
this time scale actually is the @xmath127 shown in fig .
4 . the stored photons yield a delayed emission with time scale of about @xmath166 .
the delay time is independent of the life - time @xmath164 of the source , but dependent only on the optical depth of the halo . for a stable source ,
the time - independent surface brightness ( sb ) of a stable source located in dla halo is simply proportional to @xmath167 , where @xmath168 is the radius of the halo . when the life time of the source is short , the time - dependent behavior of the surface brightness , @xmath169 , is not simply proportional to @xmath167 . to demonstrate this point
, we calculate the surface brightness of a source with life time 10@xmath3 years and ly@xmath0 luminosity @xmath170 erg s@xmath6 in a medium of @xmath171 .
the results are presented in fig .
6 , in which the size of the halos is taken to be @xmath172kpc , 10 kpc , 100 kpc , and 1 mpc . in fig .
6 , the optical depth of the dla halo surrounding the source is assumed to be @xmath173 , which correspondes to a typical dla halo with column density @xmath174 @xmath7 ( eq.(1 ) ) .
k with optical depth @xmath175 .
the source emits ly@xmath0 photons with stable flux @xmath170 erg s@xmath6 in the time range from @xmath176 to @xmath177 years .
the size of dla halo is taken to be @xmath172kpc , 10 kpc , 100 kpc and 1 mpc , respectively from top down using color codes black , red , orange and green .
, width=302 ] from fig . 6 , we first see that the curves do not show the behavior of being proportional to @xmath167 .
the maximum surface brightness of halo with @xmath178 1 kpc is about @xmath179 erg s@xmath6 @xmath7 arcsec@xmath8 , while the maximum surface brightness of @xmath178 1 mpc is only @xmath180 erg s@xmath6 @xmath7 arcsec@xmath8 . that is , when @xmath167 decreases by a factor @xmath181 , the maximum surface brightness descreases by a factor 10@xmath154 .
secondly , the shapes of the curves of fig . 6 for different @xmath168 are very different .
the curve of @xmath1781 kpc shows saturation at @xmath182 years , while all others do nt have a saturated phase .
these time - dependent behaviors are also due to the time scale of ly@xmath0 photon transfer . for a halo with given column density @xmath183 , or @xmath84 ,
the time scale of the ly@xmath0 photon transfer is about @xmath184 , which is independent of the size @xmath168 of the halo .
however , @xmath184 corresponds to a physics time @xmath185 , which is @xmath168-dependent . for dla halo with @xmath1781 kpc , the time scale
@xmath186 years .
it is much less than 10@xmath3 years , and therefore , the surface brightness is saturated when @xmath187 year . for dla halo with @xmath178100 kpc , the time scale @xmath188 years .
it is larger than the life - time of the source , and therefore , the surface brightness can not approach a saturated state .
for dla halo with @xmath189 100 kpc , the time scale @xmath185 would be much larger than 10@xmath3 years .
the source is more like a single flash and its radiative transfer is highly time dependent ( fig .
thus , we may conclude that for dla halos with a given column density , the surface brightness will be proportional to @xmath190 and @xmath191
. this result would be useful to estimate the size of the dla halo with observed surface brightness and the model of the sources .
the mechanism of the escaping of ly@xmath0 photons from expanding opaque igm at high redshift is different from that of dla halos . for the former , besides the diffusion in frequency space caused by the the resonant scattering , we should also consider the frequency redshift caused by cosmic expansion .
photons will escape from gunn - peterson trough , once their frequency is redshifted enough .
for a source at redshift @xmath2 surrounded by the igm of @xmath48k of the expanding universe .
the time duration of the emission of the source is taken to be @xmath192 , 10@xmath193 , 10@xmath194 , 10@xmath3 , 10@xmath153 , and 10@xmath154 years , respectively from bottom up using color codes blue , cyan , green , yellow , orange , and black .
@xmath18 is the delayed time of an escaped ly@xmath0 photon with respect to the first escaped free streaming photon from the source , measured at the redshift of the source.,width=302 ] similar to the previous section , we calculate the escape coefficient using the number of escaped photons .
7 presents the time - dependent solutions of the escape coefficient @xmath195 for sources at redshift @xmath196 surrounded by the igm of hubble expanding universe .
the number density of neutral hydrogen is given by the cosmology parameters of the @xmath1cdm model .
the igm is assumed to be not reionized yet .
the time duration @xmath197 of photon source has been explored with values @xmath198 , and @xmath199 years . for sources at @xmath196 , @xmath197 can not be larger than the age of the universe @xmath200 years .
the source is starting to emit photons at @xmath201 .
the curves of fig .
7 look very similar to those of fig .
5 . therefore
, one can explain fig.7 in the same way as fig .
5 , replacing the optical depth of the dla halo with the gunn - peterson optical depth ( eq.(a5 ) ) of the expanding igm at @xmath15 . in fig . 7 , all the short life time curves with @xmath202 years reach their maximum at about the same time @xmath203 years , which is the time scale of ly@xmath0 photon escape from the gunn - peterson trough in an expanding igm at @xmath2 .
one can then conclude that the escape coefficient of sources with life time less than @xmath204 years should always be less than 1 , even without dust absorption .
the curve of the shortest life time ( @xmath192 ) of fig .
7 can be thought to represent the light curve of a flash source in igm model , like that of fig . 4 for dla model . the curve of @xmath192 of fig .
7 has a long tail .
the long tail basically is a power law of @xmath18 , and is a joint result of radiative transfer and hubble expansion velocity field . as mentioned in 4.1 , for the light curve of fig .
4 , the maximum @xmath127 and variance ( or the width of the light curve ) @xmath205 are about the same . yet
, the power law long tail leads to the maximum @xmath206 of the curve of @xmath192 of fig .
7 to be much less than the width of the curve . since the suppression of escape coefficient is mainly given by the width of the light curve .
therefore , the power law long tail of fig .
7 implies that the stable state of @xmath207 takes a much longer time to approach , or can never be approached within the age of the universe .
although fig .
7 shows that the escape coefficient of sources with @xmath208 years is larger than those of sources with @xmath209 , it does not mean that the former is easier to be observed than the later .
this point can more clearly be revealed with the time - dependent solution of surface brightness .
we calculate the mean surface brightness @xmath210 defined as the flux divided by the averaged area of the source , where @xmath211 is the projected distance to the source on the sky at which a photon escapes .
thus , @xmath212(t ) is the mean of the area at time @xmath18 on the plane perpendicular to the line of sight .
the result is presented in fig .
8 , which plots the @xmath169 for sources with ly@xmath0 luminosity ( erg / sec ) and life time ( year ) paired as : ( @xmath213 , 10@xmath214 ) , ( @xmath215 , 10@xmath193 ) , ( @xmath216 , 10@xmath194 ) , ( @xmath217 , 10@xmath3 ) , ( @xmath218 , 10@xmath153 ) , and ( @xmath219 , 10@xmath154 ) .
that is , all sources emit the same amount of @xmath220 ly@xmath0 photons in their whole life span .
the sources start to emit at time @xmath201 .
k in expanding universe .
the source is assumed to be located at redshift @xmath196 , and emitting a total of @xmath221 ly@xmath0 photons from time @xmath222 .
the time duration of the emission of the source is taken to be @xmath223 , and @xmath199 years , respectively from top down using color codes cyan , green , yellow , orange , and red .
the flash source ( @xmath224 years ) is drawn with black dashed line.,width=302 ] fig .
8 has two remarkable features .
first , the four curves of ( @xmath213 , 10@xmath214 ) , ( @xmath215 , 10@xmath193 ) , ( @xmath216 , 10@xmath194 ) , and ( @xmath217 , 10@xmath3 ) are almost the same .
the common curve has a maximum at @xmath225 year and then starts decaying with a power law @xmath226 .
the time scales of the maximum and the power law tail are about the same as that of the light curve of fig . 7 with @xmath192
therefore , for sources with life time less than @xmath225 years , the surface brightness is only dependent on the total number @xmath227 of ly@xmath0 photons emitted from the source , regardless of their life time .
this is because the photons should wait for about @xmath228 years before their escaping from the igm .
the stored photons are locally thermalized , and therefore , the information of the `` age '' of the photons emitted at time @xmath229 will be forgotten during the thermolization .
this property actually is also valid for the sources of ( @xmath218 , 10@xmath153 ) , and ( @xmath219 , 10@xmath154 ) .
for these two cases , the total numbers of ly@xmath0 photon emitted in @xmath230 years are smaller , respectively are @xmath231 , and @xmath232 and therefore , their @xmath169 at @xmath233 is less than that of @xmath234 by factors 10 and 100 , respectively in the figure .
the curves of @xmath235 years become scalable from @xmath236 myr .
the @xmath237 yr curve is just starting to move away from the flash source solutions .
the relaxation time looks to be around @xmath238myr when the flash source reaches its maximum .
this is in agreement with the estimate from @xcite where the relaxation time is @xmath239 myr for our chosen redshift ( @xmath2 ) and temperature ( @xmath240 ) .
8 shows various scaling relations .
for example , for the flash source of @xmath241yr , the asymptotic slope sets in from @xmath236 myr as a result of radiative transfer in hubble expansion velocity field . for the @xmath242 yr curve in the figure , another straight part sets in from 1 myr to 50 myr even before the asymptotic slope is reached , as a joint result of photon emission , photon transfer and hubble expansion .
these time scales are comparable to the possible life time of the sources , which was previously pointed out by @xcite .
the second feature of fig . 8 is the monotonous decrease of @xmath169 when @xmath18 is large , regardless of the life time of the source .
it is very different from all the curves of fig . 5 , which will approach a stable or saturated state if the life time of the source is long enough .
for any sources in expanding neutral igm , the mean surface brightness @xmath169 does not approach a saturated or stable state .
in other words , a time - independent solution of the surface brightness does nt exist in the igm model .
this is simply because the increase of volume of the expanding igm , which stores ly@xmath0 photons , is faster than the number of ly@xmath0 photons redshifted to frequency @xmath243 .
this feature is similar to the evolution of ionized halos around a uv photon source in expanding universe .
the ionized radius can never approach a stable state required by a strmgren sphere , because the increase of the ionized radius is always lower than the comoving velocity@xcite .
therefore , in fig .
8 there is no flat section for every curve .
the maximum of the surface brightness of a ly@xmath0 photon source with luminosity @xmath4 at @xmath2 scales approximately as @xmath244 erg s@xmath6 @xmath7 arcsec .
we should emphasize that the maximum can be reached only for sources with age equal to about 10@xmath3 years .
the surface brightness would be less than the maximum when the source age is younger or older than 10@xmath3 years . that is , in term of surface brightness , a stable source will yield a time - dependent curve .
a typical problem of ly@xmath0 absorption at high redshift is the red damping wing , or hi damping wing of high redshift sources .
the optical afterglow of grbs at high redshifts has shown this feature(e.g .
we calculate the time - dependence of the red damping wing of a dla halo with radius and optical depth @xmath84 to be @xmath245 .
the frequency spectrum of the central photon source is assumed to be flat with flux equal to one .
if the source starts to emit photons at time @xmath119 , photons without undergoing scattering or collisions will escape from the halo at time @xmath246 .
the profiles of red damping wing at times later than @xmath247 are plotted in fig .
9 , in which we take @xmath248 , where @xmath249 , 0.5 , 2 , 5 , 10 and 50 . with size @xmath22 and optical depth @xmath84 set to be @xmath250 .
the central source starts to emit photons with a flat spectrum at @xmath119 .
the profiles of the red damping wing are shown at times @xmath248 , where @xmath251 is the time of free flight from @xmath23 to @xmath22 , and @xmath252 is taken to be @xmath253 , 0.5 , 2 , 5 , 10 , and 50 , respectively from bottom up using color code violet , blue , orange , yellow , green , and cyan .
the black curve , which is almost the same as the violet curve , is given by the voigt absorption , i.e. @xmath254 and @xmath255 is from eq.(1).,width=321 ] fig .
9 shows that the red damping wing at time @xmath256 can be well described by a voigt profile @xmath254 , where @xmath255 is given by eq.(1 ) .
therefore , it would be fully reasonable to fit the red damping wing of grb s optical afterglow with a voigt absorption profile , because the red damping wing is measured only a few hours or a few days after the grb explosion , the time @xmath21 is very close to @xmath247 .
even when @xmath257 ( the blue line in fig.9 ) , the red damping wing can still be approximately fitted by a voigt profile .
however , the fitting at time @xmath257 will yield a smaller @xmath84 than that of the fitting at @xmath258 .
therefore , the column density of hi atoms given by the fitting at @xmath257 is underestimated .
the red damping wing at @xmath259 shows a shoulder with frequency similar to that given by fig .
3 with parameters @xmath110 and @xmath150 . therefore
, the shoulder is from the stored resonant photons . fig .
9 shows that the curve of red damping wing has become saturated at time @xmath260 .
this is larger than the time scale @xmath261 shown in fig .
4 . it is because the photons with frequency @xmath262 of the continuous spectrum can also be stored .
according to the redistribution function ( a6 ) , the probability of transferring a @xmath263 photon to @xmath79 is larger than that from @xmath79 to @xmath263 , and therefore , in frequency space the net effect of the resonant scattering is to bring photons of the continuous spectrum background to become ly@xmath0 .
more photons be stored leads to the larger time scale of the saturation .
the problem of the red damping wing for the igm model is very different from that of the dla model .
if the frequency spectrum of source is flat , cosmic redshift will continuously provide ly@xmath0 photons from blue side of the spectrum .
since the time scale of the w - f thermalization is much shorter than the time scale of the cosmic expansion , the photons moved into the frequency range @xmath264 from @xmath265 is quickly thermalized .
thus , a huge number of thermalized photons is stored in the frequency range @xmath79 @xcite .
this process marks the difference between the igm and the dla models , and also the difference between our igm model and igm models of some others @xcite , in which the source emits only @xmath10 photons . starts to emit photons with flat spectrum at @xmath201 into igm of t=100k .
the curves correspond to @xmath266 , 10@xmath3 , @xmath267 , and 10@xmath153 years , respectively from bottom up using color codes blue , orange , cyan , and black .
the red line is given by pure absorption @xmath254 where @xmath255 is from eq.(c1 ) .
, width=294 ] fig .
10 shows the red damping wings of a source in t=100 k igm at @xmath2 with age @xmath268 , 10@xmath3 , @xmath267 , and 10@xmath153 years . when the time is less than @xmath269 year , the ly@xmath0 photons have not been significantly stored yet , and the profile of red damping wing is about the same as a pure absorption , which has been addressed by @xcite .
once the time is larger than @xmath270 year , the stored photons yield a shoulder at peak frequency @xmath271 .
10 also shows that the profile of the red damping wing seems not yet to approach a saturated or time - independent state at @xmath272 years .
the time - dependent behaviors in physics and frequency spaces of the ly@xmath0 photons emergent from optically thick medium have been extensively studied with monte carlo method .
the first conclusion is that time - dependent solutions are essential not only for strong time dependent sources like grbs , but also for stable sources , especially when the cosmic expansion needs to be considered .
cosmic expansion makes the radiation transfer equation not invariant with respect to the transformation of time shift .
consequently , the radiative transfer equation does nt have time - independent solutions in principle .
time - dependent behavior are essential . with the model of igm in the hubble expanding universe
, we show that the time - dependence of the brightness of a stable source at high redshift is like that of the light curve of an `` explosion '' , i.e. in the first phase , the surface brightness is increasing and approaching a maximum , and then , in the second phase , it decays monotonously . although our calculation is only based on a model source at redshift @xmath15 , we believe that the feature of the nonexistence of a time - independent solution , or the nonexistence of a saturated state of the surface brightness by igm scattering , would be hold for various high redshift sources
. the igm model may be too simple and ideal .
many effects have not been considered , such as the effects given by the inhomogeneity of the hi density distribution , the nonuniform ionization , the bulk velocity , the wind and the turbulence of the igm field , and the observational aperture , etc .
however , one point seems to be clear that under these circumstances , stable solutions will not exist .
all these effects should be studied with time - dependent solutions of the ly@xmath0 photon transfer .
since some high redshift ly@xmath0 emitters have already been observed , static solutions may not be enough to understand these observations . for the dla model , the saturated state or the time - dependent state , does exist .
therefore , it is reasonable to find this state directly by time - independent solution .
however , time - independent solutions can not tell us the suppression of the escape coefficient in the time range of @xmath273 , @xmath84 being the optical depth of the halo . when @xmath84 is large , the escape coefficient will be substantially suppressed for time range comparable to the age of the universe .
this is equal to say that the escape coefficient is much less than one in all time .
low escape coefficient is often explained by dust absorption .
dust extinction is effective only when the medium is optically thick .
therefore , once we use the model of dust absorption at high redshift , the time - dependent behavior must be considered .
a weno time - dependent solution of the ly@xmath0 photon transfer in dusty medium will be reported soon .
we did nt consider the ly@xmath0 photons from the recombination in the ionized halo around the center source
. simply changing the luminosity of the central light source by adding the photons from stable recombination in the halo is a poor approximation , as the time scale of the recombination is not small
. therefore , with the dla model , time - dependent solutions are also essential for understanding ly@xmath0-related phenomenon .
this work is supported by the ministry of science and technology of china , under grant no .
with dimensionless variables , the specific intensity @xmath53 of photon number is a function of @xmath21 , @xmath22 , @xmath28 and @xmath274 .
when the optical depth is large , we can take the eddington approximation .
radiative transfer equations are @xcite @xmath275 @xmath276 @xmath277 where @xmath278 are the rescaled quantities of physical intensity , @xmath279 is the angularly averaged specific intensity and @xmath280 is flux .
the mean intensity @xmath51 describes the @xmath281photons trapped in the halo by the resonant scattering , while the flux @xmath52 describes the photons in transit .
the parameter @xmath49 can be calculated as @xmath282 the term @xmath283 describes a constant physical photon source . since eqs.(a1 ) and ( a2 ) are linear with respect to @xmath46 and @xmath47 , the solutions of @xmath46 and @xmath47 for sources @xmath284 can be given by a superposition of solutions @xmath285 and @xmath286 , which are the solutions of the source @xmath287 .
the resonant scattering is described by the redistribution function @xmath288 which is the probability of a photon absorbed at the frequency @xmath289 , and re - emitted at the frequency @xmath28 .
it depends on the details of the scattering @xcite .
if we consider coherent scattering without recoil , the redistribution function with the voigt profile is @xmath290du\end{aligned}\ ] ] where @xmath291 and @xmath292 .
this re - distribution function is normalized as @xmath293 .
thus , the frequency of photons will be changed during the evolution governed by eqs.(a1 ) and ( a2 ) , while the total number of photons is conserved .
that is , the destruction processes of ly@xmath0 photons , such as the two - photon process @xcite and dust absorption , are ignored in equations ( a1),(a2 ) .
monte carlo simulation starts with releasing a photon at the source placed along a random but isotropic direction .
the frequency distribution of the new photon follows that of the source , either a gaussian distribution with a doppler core , or a continuum . once the photon enters the gas medium , the length of free path
is determined by calculating the optical depth variable @xmath294 traveled during the free flight , following the distribution function @xmath295 .
it is then straight forward to convert from @xmath294 to distance .
the location of the next scattering is thus determined .
if the place is outside the hi cloud in a halo model , or if the traveled optical depth is larger than gunn - peterson optical depth in an expanding igm model , the photon is labeled escaped . at the site of scattering , the velocity of the hi atom is generated by two steps .
first , the velocity components @xmath296 and @xmath297 ( normalized to doppler velocity @xmath298 , and @xmath299 is the propagation direction of the photon ) are generated following a maxwellian distribution @xmath300 .
second , the velocity @xmath301 is generated following the distribution : @xmath302 which is the joint requirement of gaussian distribution and lorentz profile for the rest frame cross section of resonant scattering .
the direction of the resonantly scattered photons is assumed to be isotropic , but can be easily adapted to other types of angular dependence . once the direction is generated , frequency of the outgoing photon can be calculated . using the notation of @xmath28 , @xmath289 to represent the laboratory frequency of the incoming and outgoing photon
, we have @xmath303 where @xmath304 is the recoil parameter , @xmath21 is the angle between the incoming and outgoing photons , @xmath305 and @xmath306 are the two spherical coordinates of the outgoing photon where the coordinates are chosen such that the incoming photon is in @xmath299 direction .
( we follow field 1959 s scattering geometry and notations . ) with this new set of frequency and direction of the photon , we repeat the above procedures of calculating the next scattering and the determination on escape . each photon is followed all the way along its path until it escapes .
since the effectiveness of generating @xmath301 determines crucially the speed of calculation , special algorithms have been proposed ( @xcite , zm02 hereafter ) .
we basically follow zm02 s algorithm for medium to large @xmath28 ( @xmath307 ) . for smaller @xmath28 ( @xmath308 ) , methods of plain rejection
( not employing zm02 s algorithm ) is used .
for very large @xmath28 ( @xmath309 ) , our treatment for @xmath310 is similar to zm02 , but for @xmath311 , we switch the roles of the two functions , using the distribution function @xmath312 as the transformation method to generate @xmath301 , and then use @xmath313 as the comparison function to reject .
twenty million of photons are experimented by monte carlo simulation for each model .
simulations are performed only for sources with single flash of photons . in these simulations ,
a time stamp can be recorded for each photon at each step of collision , and at its escape . for sources of arbitrary time dependence
, a new random variable is used representing the birth time of the photon .
the time stamps can be generated by adding a photon s birth time to the recorded time stamps from a single flash simulation .
furthermore , with , say , @xmath314 trials of randomly generated birth time for each original photon in a single - flash simulation , we form a subgroup of @xmath314 photons , which have the same history of collisions but happen at different epochs . by this way we greatly improve the very low usage rate when
the data of each recorded photon is coupled with only one birth time .
statistics over this enlarged group of photons give better continuity and smaller poisson errors on the surveyed quantities , but do not add new physics .
the free path of a ly@xmath0 photon in a hubble streaming igm can be derived by using gp optical depth for frequency @xmath28 measured at source redshift @xmath299 , which is @xmath315 where the gunn - peterson optical depth parameter is given by equation ( a3 ) . in mc simulations , the hubble streaming always makes frequency @xmath28 lower and gp optical depth smaller .
suppose the photon frequency is @xmath316 immediately after the last scattering and @xmath317 before the next scattering , the change of gp optical depth @xmath318 during the free flight is always negative and equals to -1 on average .
the particular value of each @xmath318 is determined by generating a random number @xmath319 following the distribution function @xmath320 . with this @xmath318
we can derive the new frequency @xmath317 by solving the equation @xmath321 by compiling a large data table of @xmath322 and using linear interpolation , @xmath317 can be solved fast . the free length @xmath323 before the next scattering can be found by considering that the frequency change is caused by doppler effect of hubble motion , @xmath324 where hubble constant at redshift @xmath299 is @xmath325 in which all @xmath326 refer to values at present . | with monte carlo simulation method , we investigate the time dependent behavior of ly@xmath0 photon transfer in optically thick medium of the concordance @xmath1cdm universe . at high redshift ,
the ly@xmath0 photon escaping from optically thick medium has a time scale as long as the age of the luminous object , or even comparable to the age of the universe . in this case
, time - independent , or stationary solutions of the ly@xmath0 photon transfer with resonant scattering will overlook important features of the escaped ly@xmath0 photons in physical and frequency spaces .
more seriously , the expansion of the universe leads to that the time - independent solutions of the ly@xmath0 photon transfer may not exist .
we show that time - dependent solutions sometimes are essential for understanding the ly@xmath0 emission and absorption at high redshifts
. for ly@xmath0 photons from sources at redshift @xmath2 and being surrounded by neutral hydrogen igm of the @xmath1cdm universe , the escape coefficient is found to be always less , or much less than one , regardless of the age or life time of the sources . under such environment
, we also find that even when the ly@xmath0 photon luminosity of the sources is stable , the mean surface brightness is gradually increasing in the first 10@xmath3 years , and then decreasing with a power law of time , but never approaches a stable , time - independent state .
that is , all @xmath2 sources in a neutral hubble expanding igm with ly@xmath0 luminosity @xmath4 have their maximum of mean surface brightness @xmath5 erg s@xmath6 @xmath7 arcsec@xmath8 at the age of about 10@xmath3 years .
the time - dependent effects on the red damping wing profile are also addressed .
cosmology : theory - intergalactic medium - radiation transfer - scattering |
Subsets and Splits